The Student Conference on Medical Engineering Science is an annual event at the BioMedTec Science Campus Luebeck. The Student Congress is organized by the University of Lübeck and supported by NORGENTA, the life science cluster agency in north Germany. Master students of programs related to medical engineering science present results of their recent research projects.
Die Studierendentagung Medizintechnik findet jährlich auf dem BioMedTec Wissenschaftscampus Lübeck statt. Der Kongress wird von der Universität zu Lübeck organisiert und von der norddeutschen Life-Science-Clusteragentur NORGENTA unterstützt. Studierende in Masterprogrammen der Medizintechnik und der Lebenswissenschaften präsentieren die Ergebnisse ihrer jüngsten Forschungsprojekte.
Contents
1 Biomedical Engineering I
Development of a program to analyse and visualize ciliary beat frequency ex vivo
Christian Myrtus
Plug-in LED lighting for ureteroscopes
Milan Öri
CE certification of TTI Imaging
Jan Krieger, Burkhard Zander, Dagmar Lühmann
EMG-based estimation of wrist kinematics using Fisher‘s linear discriminant analysis
Nina Rudigkeit, Liliana P Paredes Calderon
Material compatibility with different sterilization procedures
Yannik Schröder, Johannes K.-M. Knobloch
2 Biomedical Engineering II
Electrical Impedance Tomography Image Reconstruction with EIDORS
Julia Henschel, Steffen Kaufmann, Aram Latif, Windy C. Saputra, Tanner Moray and Martin Ryschka
Development and implementation of a method for producing directional solidified, electrospun hybrid structures as nerve guidance channels
Christopher Janssen, Stephan Klein, Birgit Glasmacher, Soenke Wienecke, Tanmay Chakradeo
Spectral light modulation using a digital micromirror device (DMD) for the calibration of pulse oximetry sensors
Stefan Marx, B. Weber, B. Nestler, H. Gehring
Multi-frequency Electrical Impedance Tomography for irreversible Electroporation
Windy Saputra, Steffen Kaufmann, Tanner Moray, Aram Latif, Julia Henschel and Martin Ryschka
Filtering cardiac artefacts from transdiaphragmal pressure for the validation of a non-invasive method to assess work of breathing
Merle Strutz, K. Lopez-Navas and U. Wenkebach
3 X-Ray and Computed Tomography
Emissivity factor comparison of different coatings for medical x-ray tube housings
Imke Zeuner
Models for osteoarthritis assessment from digital x-ray images of the lower extremity
Alexander Mikula, André Gooßen and Rolf-Rainer Grigat
Phantom-based Determination of Noise Distribution in Computed Tomography
Christian Kaethner, J. Müller and T. M. Buzug Construction and Calibration of a Micro-CT Phantom for the Determination of Iron Oxide Concentrations in Ferrofluids
Christina Maria Debbeler, Jan Müller and Kerstin Lüdtke-Buzug
4 Magnetic Particle Imaging
Localization of small ferromagnetic samples in a magnetic particle imaging scanner
Nils Nothnagel, J. Rahmer
3-Dimensional FFP-MPI-Scanner Simulation using X-Space Theory
Matthias Weber, P. Goodwill and S. Conolly
Realistic Simulation of a Movable and Rotatable Field-Free Line in Magnetic Particle Imaging
Klaas Bente, T. F Sattel and T. M. Buzug
5 Magnetic Resonance Imaging
Flexible Probe Positioning for Workbench Measurements on MRI Coils
Lars Kreutzburg, Torsten Hertz
Curved saturation for spine imaging in magnetic resonance imaging
Britta Lehmann, Dieter Ritter, Josef Pfeuffer
Visualization of tumor tissue in the peripheral zone of the prostate using multi-parametric MR images
Erik Slowikowski
Rotation estimation in k-space for different trajectories
Anselm von Gladiß, Christoph Kolbitsch, Tobias Schaeffter and Ghislain Vaillant
6 Biomedical Optics
Development of a Novel Fractional Laser Device Utilizing a Tunable Cr2+:ZnSe Infrared Laser
Man Linh Ha, Dieter Manstein
Full range Fourier domain optical coherence tomography via piezo-driven reference mirror
Susanne Luft, Marc Krug, Eva Lankenau
Resection of human calcified aortic heart valves in vitro by using a Thulium laser
Jennifer-Magdalena Masch, Ingo Rohde, Dirk Theisen-Kunde, Ralf Brinkmann Determining the accuracy and repeatability of a multidimensional eye tracker designed for laser refractive surgery
Laila Paulsen, M. Abraham
7 Medical Image Computing
Visualization of self-expanding stent systems and reject minimization
Maresa Glanert, H. Paulsen, C. Wöhry
Preprocessing of Spectral Retinal Images for Registration
Florian Griese, L. Lensu, R. Bruder
Application of Machine Learning Regression Techniques on Predicting Clinical Outcome in Primary Progressive Multiple Sclerosis
Viktor Wottschel, Benedetta Bodini, Olga Ciccarelli, Alan J. Thompson, Daniel C. Alexander
Mathematical modelling of breast tumour growth and treatment
Anna Heye, Mark A. J. Chaplain
Camera and tracking system calibration for image guided bronchoscopy
Pedro Manuel Baptista Néova, A. Schlaefer
Conference Chair
Thorsten M. Buzug (Chair), Institute of Medical Engineering, University of Lübeck
Stephan Klein (Co-Chair), Center for Biomedical Technology, University of Applied Sciences Lübeck
Local Coordination
Kanina Botterweck, Medisert, BioMedTec Science Campus
Eugenie Ewert, Medisert, BioMedTec Science Campus
Christian Kaethner, Institute of Medical Engineering, University of Lübeck
Gisela Thaler, Institute of Medical Engineering, University of Lübeck
Scientific Program Committee
Erhardt Barth, Institute of Neuro- and Bioinformatics, University of Lübeck
Reginald Birngruber, Institute of Biomedical Optics, University of Lübeck
Henrik Botterweck, Center for Biomedical Technology, University of Applied Sciences Lübeck
Ralf Brinkmann, Institute of Biomedical Optics, University of Lübeck
Thorsten M. Buzug, Institute of Medical Engineering, University of Lübeck
Jens Christian Claussen, Institute of Neuro- and Bioinformatics, University of Lübeck
Jan Ehrhardt, Institute of Medical Informatics, University of Lübeck
Hartmut Gehring, Clinic of Anesthesiology, University Medical Center Schleswig-Holstein, Campus Lübeck
Heinz Handels, Institute of Medical Informatics, University of Lübeck
Christian Hübner, Institute of Physics, University of Lübeck
Gereon Hüttmann, Institute of Biomedical Optics, University of Lübeck
Josef Ingenerf, Institute of Medical Informatics, University of Lübeck
Stephan Klein, Center for Biomedical Technology, University of Applied Sciences Lübeck
Martin Koch, Institute of Medical Engineering, University of Lübeck
Kerstin Lüdtke-Buzug, Institute of Medical Engineering, University of Lübeck
Alfred Mertins, Institute for Signal Processing, University of Lübeck
Jan Modersitzki, Institute of Mathematics and Image Computing, University of Lübeck
Bodo Nestler, Center for Biomedical Technology, University of Applied Sciences Lübeck
Alexander Schläfer, Institute for Robotics and Cognitive Systems, University of Lübeck
Alfred Vogel, Institute of Biomedical Optics, University of Lübeck
Preface and Acknowledgements
The First Student Conference on Medical Engineering Science has been organized on March 29/30, 2012 by the BioMedTec Science Campus Lübeck in cooperation with Norgenta, the North German Life Science Agency and the technology transfer platform Medisert. Master and diploma students presented their recent research results to broad public from academics and industry
Students from the Life Sciences programs at the BioMedTec Science Campus presented their results from projects carried out at the Laboratories and Institutes of Lübeck’s Universities, in international research facilities or research-oriented industrial companies. The conference focus has been placed on topics from medical engineering. Biomedical engineering has been established at the University of Applied Sciences Lübeck for decades and Medical Engineering Science (MIW) is an important bachelor and master program at the University of Lübeck as well. Both Universities jointly offer the international master degree course Biomedical Engineering (BME). This is complemented with further life-science oriented programs of the University (Computer Sciences, Medical Computer Sciences, Mathematics in Medicine and Life Sciences, Molecular Life Science, Medicine) which contribute to the success of the interdisciplinary Medical Engineering Science and of Biomedical Engineering
These competencies, which the conference program impressively reflects, also meet the requirements of biomedical industry. It is known that Germany lacks graduates of the MINT programs in order to be able to compete on the global market. And MINT means Mathematics, Informatics, Natural Science and Technology. In Lübeck, one may optionally replace the „M“ with medicine. As can also be seen in the conference program, the fields of imaging and image processing are further foci in Lübeck. Excellent research achievements of Institutes,
Laboratories and Clinics at the BioMedTec Science Campus are closely linked with lecture plans, so that many students can demonstrate their competencies in project work in renowned national and international research facilities during their master degree program
Finally, I want to thank all the people who worked with enthusiasm and dedication to make the conference a successful event. Without the financial support of Norgenta, the North German Life Science Agency, and the commitment of Angela Wäsche, this conference would not have been possible. Moreover, my thanks go to the technology transfer platform Medisert of the BioMedTec Science Campus. The professional management of Kanina Botterweck and her Medisert team has contributed substantially to the success of this conference. In the context of the project “Encounter with Research” of the Lübeck Engineering Laboratory (LILa), interested pupils of the upper secondary schools in Lübeck and surroundings are invited to participate as guests at the Student Conference. I would also like to express my thanks to the coordinators of LILa, Julia Hamer and Tina Anne Schütz, for the organization of this part of the program. Thanks too to the participants of the companies who, in workshops on the first day of the Students’ Conference, give insights into what companies expect from graduates: Philips Medical GmbH; Birte Loffhagen, Dräger Medical; Pia Jedamzik, Stryker Osteosynthesis; Dr. Ulrich Hoffmeister, Lübeck Chamber of Industry and Commerce; Dr. Frank Schnieders, Provecs Medical GmbH. Especially - and therefore as the final point - I would like to thank Bäbel Kratz personally and on behalf of all colleagues of the BioMedTec Science Campus. Bärbel Kratz from the Institute of Medical Engineering has been the first contact point for all questions of students and the program committee. Her excellent overview of all details of this event was the key to the success of the first Student Conference at the BioMedTec Science Campus
Lübeck, March 29/30,
Prof. Dr. Thorsten M. Buzug
Vicepresident of the University of Lübeck
Chair of the Student Conference on Medical Engineering
Science
Biomedical Engineering I
Development of a program to analyse and visualize ciliary beat frequency ex vivo
Christian Myrtus
Abstract—A main topic in lung research is to understand the mechanisms that continuously clean the airways from inhaled particles by transport of mucus. This transport is maintained by the continuous beating of cilia that are present on airway epithelial cells. To date, the beat frequency of individual ciliated cells is measured as an indicator of mechanical clearance activity. However, the present frequency analysis was, due to many constraints, very time consuming, inflexible, limited in features and very error prone. The main task of this work was the improvement of the analysis tools to analyze ciliary beat frequency. This was performed by the development of an encapsulated program environment, which was implemented with Matlab (The MathWorks, Inc.). The new program improved the analysis of ciliary beat frequency of individual cells but also generated a tool that allows to visualize the changes in ciliary beat frequency of all cells present in the microscope field of view opening up new possibilities for analysis.
I. Introduction
In the research to combat respiratory and lung diseases the function of the airway epithelium plays a central role. The beating of ciliated cells in the airway epithelium is the main mechanism how mucus and particles are cleared from the airways. The velocity of mechanical clearance of mucus and particles is proportional to the ciliary beat frequency. For the evaluation of mechanical mucus clearance ciliary beat frequency is an important indicator, whose monitoring and analysis were the focus of this work. Figure 1 visualizes schematically the normal motion-sequence of the ciliary beat that induces the clearance of the airway covering mucus (see [I], to get more details).
The previously used analysis method of ciliary beat frequency was carried out using different programs that made, on the basis of their individual functional properties, the analysis very slow. Thus, for example, regions of interest (ROI) had to be defined sequentially, for which the frequency analysis of grey value changes over time, using discrete Fourier transform (DFT), had to be carried out in a separate program environment. The result revealed a frequency spectrum of an considered ROI that gave the scientist the opportunity to associate an ’’appropriate” frequency to the ciliary movement. The following steps of this analysis process represented a potential source of error:
- Definition of ROI (size and position)
- Strategy (formation of the average grey values per frame) to calculate the data for the execution of the DFT
Christian Myrtus - Medizinische Ingenieurwissenschaft, Universität zu Lübeck; the work has been carried out at the Institute of Anatomy, University of Lübeck
illustration not visible in this excerpt
Fig. 1: Schematically illustration of the mechanical clearance of mucus by ciliary beating. By the movement of the cilia a mechanical force is applied to the mucus and the periciliary layer, leading to a directed movement, (modified after [1])
- Data fonnat incompatibilities
- Exclusion of digital filters to enhance interpretation of the frequency spectrum
- Individual assignment of an ’’appropriate” beat frequency by scientists
Another major disadvantage was the inability of the beat frequency analysis of the field of view of the microscope. Thus, the main task of my work consisted in the development of a program to determine the ciliary beat frequency, without the above mentioned sources of error.
Validation experiments were performed using transmission light microscopy on tracheae of mice ex vivo.
II. MATERIAL AND METHODS
The program for ciliary frequency analysis was developed with Matlab (The Math Works, Inc.) and based on the DFT of video data. For this purpose the video data are converted into a three-dimensional matrix Mo. The nm-dimension corresponds to the respective frame and the third dimension 1 corresponds to time (see figure 2).
The applied discrete Fourier transform is shown in the following equation:
illustration not visible in this excerpt
where[illustration not visible in this excerpt]number of video frames and
illustration not visible in this excerpt
f = sampling frequency, ω = angular frequency (normalized by the sampling frequency) , T = sampling time [2]
illustration not visible in this excerpt
Fig. 2: Exemplary visualization of a three dimensional Matrix M0, which represents the video data. The nm-plane corresponded to the video frame. Each DFT will be executed in l-direction, which corresponds to the elapsing time. The red cylinder illustrates for example an vector of grey values, which can be transformed into frequency space using the DFT.
A. Filter
The developed software provides two filters. ’’Unsharp masking” produced edge sharpening, whereby the cilia should be improved distinguished from the background and thus the movement could be better identified (see [3]). The ’Bandpass filter” should allow the user to specify a frequency range that is used for frequency analysis. Thus it is possible to ignore dominant (interfering) frequencies. Both makes the assignment of frequencies of the ciliated epithelial cells more reliable.
B. Visualization of the Fourier spectrum of a region of interest
The basic principle of visualization of the spectrum was the definition of an region of interest (ROI) (see figure 3). Crucial for the frequency analysis are the size and the position of the ROI. For the technically frequency analysis only the submatrix Mx (x 2 N \{0} - x is an index to distinguish between the several defined ROI) of the Matrix M0, was used.
1) Average of grey values: This method provided the calculation of the average of the grey values for each (sub-) frame (layer) of the sub-matrix Mx (see equation 4). The results were represented by an vector of grey values, to which the DFT has been applied.
illustration not visible in this excerpt
with Rx = number of rows, Cx = number of columns, l = frame index
illustration not visible in this excerpt
Fig. 3: Association between the Matrix M0 and the sub-matrix M1, which corresponded to the user defined ROI. The shape of an ROI only can be an rectangle. The index x(x 2 N \{0}) of the sub-matrix denotation corresponds to an specific identifier.
This calculation method was made available to scientists, so they can compare their former results with the new ones.
C. Additional calculation strategies to determine the dominant frequency of an region of interest
The developed program provides two different calculation methods. Both are using the defined ROIs but differ from the method described above. In contrast to the previous method the following methods are performed in frequency space. Thus, for each Pixel inside of the ROI, the dominant frequency was calculated. This enhancement should make the results more resistant to interference.
1) Average of dominant frequencies: Here the average of the matrix of dominant frequencies fd was calculated. The calculation is analog to the equation 4.
2) Median of dominant frequencies: Analog to the calculation of the average of the dominant frequencies, this method calculates the median. This means that the dominant frequencies are sorted by size, followed by selecting the value in the middle.
The calculation of the mean as well as the median of the frequencies produces a representative ciliary beat frequency for the considered ROI.
D. Color coded visualization of the frequency distribution
This type of visualization calculates the dominant frequency of each pixel of one frame. This creates a two dimensional matrix Fdom of dominant frequencies. With a color table each dominant frequency can be assigned to a color. The result of color coded frequencies visualizes a distribution of frequencies in where ciliated epithelial cells can be identified by the user. The video matrix M0 can also be decomposed into sub-matrices of a defined length. The length should be chosen depending on the maximum of expected frequency. Is a matrix decomposition made, the frequency distribution can be calculated and visualized for each sub-matrix. These resulting different images can be saved on the hard disk in tagged image file format (TIFF). With an additionally developed program these images can be converted to visualize changes in ciliary beat frequencies over time.
E. Graphical User Interface (GUI)
A GUI was developed in cooperation with the users and ensures a quick overview over all functions and allows a quick change of analysis parameters. Additionally it provides an error free program execution, visualization and export of results to various data formats.
A detailed description of the construction and operation of the GUI is omitted because of space limitations.
F. Experiments to evaluate the developed program
The frequency analysis of ciliated cells was performed using the explanted trachea of the mouse. The trachea was extracted from the mouse immediately after determining of death. It was cut open lengthwise and fixed with insect needles in a Petri dish was equipped with a base of Sylgard (silicone elastomer), which was filled with 2 ml of Hepes-Ringer solution. Hepes-Ringer serves to counteract possible ion gradients and keeps the pH-value in the physiological range. In addition, the temperature in the Petri dish was controlled by an ¿{Environmental Culture Dish Controller (Bioptechs Inc., Delta T5). For imaging, a Zeiss Axio Examiner AX 10 fixed-stage light microscope with an iimnersion objective (Zeiss wPlan - Apochromat, 20x/1.0, DIC (UV)VIS-IR) and a digital camera Sumix ME-150 (Monochrome 2/3” 1.3 mega pixel CMOS camera) was used.
With the external software StreamPix (NorPic, Inc.), the video data acquired and stored in Audio Video Interleave (AVI) format. The resulting video was used for the frequency analysis.
The ciliary beat frequency was modulated in two independent experiments using either modulation of environmental temperature or application of ATP to the Hepes-Ringer solution. The Results were recorded with the microscope camera which generates an video with a resolution of 640 x 480 pixel. The maximum ciliary beat frequency was estimated to be about 50 Hz, thus, based on the Nyquist criterion, a sampling frequency of 100 Hz was chosen.
1) Analysis settings: For the visualization of the color coded frequency distribution the matrix M0 was analyzed in equidistant discrete time steps over a length of 100 frames. The ROI based process used the same analysis intervals, where a selected quantity of representative ROIs were defined. These ROIs were positioned in the center of ciliated epithelial cells and were examined using the methods introduced in section II-Cl and section II-C2.
The protocols of experiments were as follows:
1) Addition of ATP to the Hepes-Ringer solution
Increase of the ciliary beat frequency was achieved by the addition of ATP (20 μΐ, 10 -mol/l) to the Hepes- Ringer solution (2 ml). Ciliated epithelial cells have a ATP receptor that upon activation increases ciliary beat frequency.
The environmental temperature was kept at 30 °C (±0.3 °C) and no refocusing was necessary. The recording time was 60 minutes to recognize a possible normalization of ciliary beat frequency. 20 ROIs were analyzed.
2) Increase of the environmental temperature
In this experiment the environmental temperature was increased from 31 °C to 46 °C. During the heating refocusing of the tissue was necessary. 12 ROIs were analyzed.
III. Results and Discussion
A. Results of the first experiment
Figure 4 illustrates the ROI based changes in ciliary beat frequencies after adding ATP to the Hepes-Ringer solution. The ciliary beat frequencies increased over time. This increased frequency was stable over a long period, but decreased slightly to the end of execution. Selected results of the color coded frequency distribution are shown in figure 5. They should illustrate the advantage of this new method to analyze the ciliary beat frequency.
illustration not visible in this excerpt
illustration not visible in this excerpt
Fig. 4: Changes of the ciliary beat frequency after adding of ATP. Average frequency (blue - see section II-Cl) and median frequency (red - see section II-C2) are shown.
The experiment showed:
- In contrast to the ROI based analysis (see Figure 4) the color coded analysis (see Figure 5) provides the ciliary beat frequencies of all ciliated cells in the field of view.
- The dynamics of the system, such as the lack of tight frequency synchronization of different ciliated cells were seen without further analysis.
- It could be determined that in all examined Experiments the measured ciliary beat frequency of the movement was in the range of 8 Hz to 50 Hz.
- In the color coded frequency distributions of ciliated cells, areas with doubled frequencies within the ciliated cells were displayed. This frequency doubling gave information about the direction of the ciliary beat.
B. Results of the second experiment
Figure 6 and Figure 7 visualized the marked increase of the ciliary beat frequency during the heating process. Both methods illustrated the increase of ciliary beat frequency with temperature. However the ROI based analysis (see Figure 6) failed to detect the death of the ciliated cells at the temperature of 46 °C. In contrast Figure 7d demonstrated that cell death had occurred. Only noise of the background was detected. This
illustration not visible in this excerpt
Fig. 5: These four figures visualize the observed beat frequencies of the ciliated epithelial cells at four different time points. At the beginning 20 μΐ (10 bnol/l) of ATP were added to the Hepes-Ringer solution. Based on the changes of color, an increases of ciliary beat frequency can be seen that decreases slightly with time.
is a major advantage of the visualization of the color coded frequency distribution.
illustration not visible in this excerpt
At the Klinik für Kinder- und Jugendmedizin Lubeck, the method presented was also successfully used to analyze material from patients. Ciliary beat frequency analysis is an important factor in the diagnosis of Primary Ciliary Dyskinesia (PCD) [4].
IV. CONCLUSIONS
The developed analysis program gives the opportunity to analyze the ciliary beat frequency of individual cells faster and more reliable than with the previously used methods.
illustration not visible in this excerpt
Fig. 7: These four figures visualize the determined beat frequencies of the ciliated epithelial cells in respectively four different environmental temperatures. Accompanied with the temperature rise the frequency increases, but this modulation was not observed for all cells equally.
Furthermore, the color coded visualization of ciliary beat frequency over time gives the possibility to quickly assess the behavior of all ciliated cells present in a video. A feature that was not possible with previous analysis methods.
Currently the method is further evaluated for its value in diagnostics to detect abnormalities of ciliated cells in human samples.
To diagnose PCD directly in patients we currently adapt optical coherence tomography (OCT) to carry out ciliary beat frequency analysis in vivo.
Acknowledgment
This work was part of an internship, which I completed in the Institut für Anatomie, Universität zu Lübeck under the supervision of PD Dr. Peter König, for which I am very grateful.
I also want to thank Dr. Karina Weinhold and Dr. Mario Pieper for the excellent mentoring and their help in the execution of the experiments.
References
[1] Michael R. Knowles and Richard C. Boucher, Mucus clearance as a primary innate defense mechanism for mammalian airways, The Journal of Clinical Investigation, Volume 109, Number 5, March 2002.
[2] Alfred Mertins, Signaltheorie - Signaltheorie Grundlagen der Signalbeschreibung, Filterbnke, Wavelets, Zeit-Frequenz-Analyse, Parameter- und Signalschtzung, 2. Auflage, Vieweg+Teubner, 2010.
[3] Wilhelm Burger and Mark J. Burge, Principles of Digital Image Processing - Fundamental Techniques, Springer, 2009.
[4] Paul C. Stillwell, M.D., Eric P. Wartchow, B.Sc., and Scott D. Sagel, M.D., Ph.D., Primary Ciliary Dyskinesia in Children: A Review for Pediatricians, Allergists, and Pediatric Pulmonologists, PEDIATRIC ALLERGY, IMMUNOLOGY, AND PULMONOLOGY, Volume 24, Number 4, 2011[1]
Plug-in LED lighting for ureteroscopes
Milan Oeri
Abstract-In endoscopic imaging an adequate illumination of body cavities is essential. The use of xenon lamps has several drawbacks like size, short lifetime and high cost. An alternative light source is desirable. The increase of LED (light emitting diode) light efficacy opens ways for the replacement of xenon lamps. This is a feasibility study of a plug-in LED which can be attached to ureteroscopes and illuminate small body cavities. Experiments showed promising results regarding sufficient illuminance by LED lighting also for greater body cavities besides urology. In view of this ability, a first design of a mains- opereated LED lighting is depicted.
I. Introduction
High quality imaging in medical endoscopy is limited by the transfer of light into body cavities. Light sources with a sufficient light output are essential. currently, lighting is realized by external metal halide, halogen or xenon lamps mounted onto unfavorable supply towers and light has to be transferred by light conduction cables. On its way to the proximal end of the endoscope a portion of light is absorbed. cold-light sources as xenon lamps produce light containing a highly reduced infrared fraction. Resulting lamp heat has to be cooled by thermal convection though. The ineffective conversion of electric power into light produces large amounts of heat. Furthermore, the handicap of the user due to wiring and the potential of examinations devoid of expanded endoscope supply towers give reason to attachable sources of light in endoscopy (compare Fig. 1). In this work the illuminances produces by high-power LEDs were examined as a possible miniaturized device replacing xenon lamps.
Rapid improvements in the performance of light emitting diodes (LEDs) facilitate the substitution of former lamps in medical instruments, as the efficacy factor is twice as high, lifetime a hundredfold and therefore costs lower [1]. Recently, integrated or attachable LED lightings were implemented for examination of body cavities of small-sized volumes as in nasopharyngoscopy or as diode-array for abdominal examinations [2], [3]. A battery-driven LED lighting (EndoLED, Olympus, Hamburg) is limited to laryngoscopy, as it here not intense illumination is needed.
The elevation of light output of high power LEDs to 160 lumen/watt is assumed to enable their utilization in urology, despite the demand of light coupling into the optic fiber adapter of the endoscope and the heat management of an attachable device, respectively.
With respect to user’s affinity to the light of operational lamps the color temperature should be between 5000 and 6000 K. Metal halide lamps emit light with a color temperature of 2000 K, halogen lamps 3000 K and xenon lamps 5000 K.
Recent LEDs are available in “natural white” (4000-5000 K) and “cold white” (> 5000 K). Most white LEDs are based on a blue LED combined with a yellow, luminescent phosphor [4]. Since blue LEDs have the highest effectiveness, a high fraction of blue is used in high power LEDs. The higher the blue fraction within the emission spectrum, the higher the correlated color temperature is. As a maximum luminous flux is required to make replacement feasible, “cold white” color temperature of 5000-6000 K is favored.
illustration not visible in this excerpt
Fig. 1. Sketch of the LED lighting (A) attached to an uretero- resectoscopes. As wiring (CO2-supply (B), exhaust supply (C) and power supply for HF-surgery (D)) handicaps the user, a mountable lighting is required. The device should be supplied with power of the camera head (E).
Because color temperature does not define how natural colors of objects appear, the color rendering index (CRI) is more significant. Even if two light sources are of the same color temperature, the CRI can be different. Maximum CRI (100) is reached, if the light source shows no difference in appearance compared to a reference at same color temperatures. The CRI of white LEDs is in the range of 70-85. The reference source from 2000 K to 5000 K is the ideal black body. Above 5000 K daylight locus is the reference. The higher the index the more authentic the look of an object is. Recently, new color rendering evaluation systems are discussed due to striking consistency of the CRI for light sources with a spiky emission as white LEDs since at 5000 K the CRI is discontinuous and the light reference switches [5].
Still heat management is important for high power devices, which are driven by currents exceeding 350 mA. As recombining charge carriers emit light, light output is nearly proportional to the current. However, the increase of current leads to a characteristic increase of forward voltage and therefore to high junction temperature of the semiconductor. The higher the current and forward voltage, respectively, the higher the light output. Efficacy is decreasing and dissipation losses are rising. Therefore one must balance the increase of forward current and junction temperature regarding the required luminous flux and long lifetime.
Accordingly, the optimization of heat dissipation by optimal positioning of LED with respect to the heat sink is necessary. The heat transfer between chip, epoxy-coated conduction board, mounted plate and heat sink can be specified as an array of resistances. Due to the increase of dissipation loss with higher current, maximum efficacy (lumen/watt) is reached for minimal current. Also, manufacturer predict long lifetime at 700 mA current.
As light of the portable device has to be coupled into the fiber cone of the instrument, focusing of the LED output is necessary. Usually high power LEDs provide a wide angle of radiation of 90° to 115°. The surface mounted device LED consists of a metal mirror basement and a surface mounted lens. A secondary optic, a focusing lens, has to be brought into the beam enabling light coupling into the fiber cone of the endoscope. In comparison both couple areas of laryngoscopes and ureteroscopes were examined. It results a ratio of 1.47, i.e. plugging the EndoLED lighting onto ureteroscopes leads to a 1.5-times darker spot. Theoretically, the maximum possible coupling efficiency is characterized by the quotient of offered ( Pled ) and coupled power ( PC ):
illustration not visible in this excerpt
Emitted power can be approximated by modeling the plane LED chip as a Lambertian radiator:
illustration not visible in this excerpt
II. Material and Methods
Methods consisted of four studies. First study concentrated on the feasibility of an energy clip for ureteroscopes. In the following study, guideline values for illuminance were measured and LED characteristics compared with respect to maximum light efficacy. Thirdly, illumination of most efficient LED at working distance was examined and compared to xenon lamp and EndoLED. To prevent LEDs from strong heating, heat sinks were designed for each setup. In a last step, a design of the modified LED light source was worked out.
The construction of the energy clip was done using CAD- software (Unigraphics, Siemens, Germany). Usually construction in endoscopy is rotational symmetric and therefore in a first step a two-dimensional sketch is designed. By rotating the sketched shape about the main axes, objects, e.g. lens holders, will be created. What is more challenging is the design of a clip, exemplarily shown as paper draft in Fig. 2. Normally, this is performed by using freeform tools. The model was instead designed by classical means due to unavailability of the freeform tool kit.
illustration not visible in this excerpt
Here, rLED is the radius of the emitting plane, LLED radiance and Ω solid angle. Further coupled power is given by:
illustration not visible in this excerpt
Fig. 2. The device clips on the main body of the instrument and separates into two conductive pieces supplying the lighting with energy.
illustration not visible in this excerpt
As a convention, fiber profile parameter a is striving against œ for step index fiber with radius x . For rLED « x, rmax is equal to rLED . With last equations it results:
illustration not visible in this excerpt
The numerical aperture (N.A.) of 0.66 of the present optic fiber leads to a coupling efficacy of 3 = 0.43[6]. As maximum coupling efficacy is required, the following factors have to be considered: radiation pattern, N.A. of the optical fiber, distance between emitter and fiber, adjustments of both optical axes and optimal secondary lenses. Further coupling methods were explained in chapter II.
Since there are only battery-driven LED light sources for endoscopes yet, an energy interface for mains powered LED lighting has to be considered. Here, it is postulated, that the camera head supplies the light source with energy. Electric circuit is arranged by conductive energy interface between adapter and eye piece, respectively, and light source, which is to be plugged on the corresponding cold-light adapter of the endoscope main body. The adapter allows the user to rotate the instrument freely around its main axis, which requires a rotatable energy connection. A first concept is to design the energy interface as a molded interconnecting device, a threedimensional PCB, which is isolating against the main body. An idea is to design a “clip-on energy interface” clipping onto the main body and eye piece. Thus, the setup including the light source is demountable.
The construction steps were as follows: sketching of the silhouette with cylindrical shape, revolving outline, inserting a slot that the clip can be slipped over the instrument, trim body and define a spline curve along the ureteroscope surface to get two side pieces, which loop along the main body and end in contact tips. A tight spot was added at the end of the conductive side pieces. The spot sticks to the optical fiber adapter of the destined instrument which gives a greater halt.
The clip was manufactured by a prototype device allowing three-dimensional print (Fig. 3.). As this can be seen as a feasibility study for a clipped device, no further effort is taken towards conductive paths, except for chamfers spending space to isolated electric lines. Investigation regarding the idea of an energy clip bridging the energy gap between the camera adapter and the interface including the rotatable connection to the camera adapter is further done within another study.
illustration not visible in this excerpt
Fig. 3. Left: side view of the energy clip prototype. The tight spot in the front clips on the optic fiber adapter. Middle: proximal part is plugged into the camera adapter, which provides the clip with energy. Right: top view of clip with wires demonstrating conductive paths.
Illuminance of the EndoLED (XLamp XR-E, Cree, USA) attached to a 4-mm-ureteroscope (A22001A, Olympus, Hamburg) with viewing angle of 12° was measured and compared to the xenon lamp. The requested illuminance is stated to be optimal at 2.1 klx (internal procedure). Within the measurements light is coupled into the ureteroscope and illuminance of the optical system detected by an integrating sphere (LMT B360, Lichtmesstechnik, Germany). Table I shows the results.
Table I Comparison of current light sources with respect to the required Illuminance
illustration not visible in this excerpt
[1] 4 mm diameter, viewing angle of 12 °, radiation angle of 60°
In contrast to EndoLED, maximum illuminance of 7.1 klx using xenon lamps is possible, which is a fourfold of the requested illuminance in urology. Plugging the EndoLED onto ureteroscopes leads to insufficient illuminance by factor 0.57. As it is designed for coupling into larger fiber diameters of laryngoscopes an optimization of the light output is necessary to overcome the lack of illuminance. An alternative is to use more efficient LEDs.
LED characteristics by manufacturers compared to xenon lamp are shown in Table II. Three LEDs (XLamp XP-E and XP-G, Cree, USA; Luxeon Rebel, Philips Lumileds, USA) were compared to conventional xenon lamp (Visera CLV-540, olympus, Hamburg). All LEDs differ in their characteristics. Newer chips are significantly smaller and therefore promise a miniaturization of the lighting design. Here, the influence of the angle of radiation, color temperature, CRI and thermal resistance with respect to optimal illuminance of the field of view (FoV) and heat management is observed.
Table II Light Source Characteristics of LEDs and xenon lamp
illustration not visible in this excerpt
at current of 0.35 A; [2] at maximum current (differs between LED types)
Color temperatures are in the range of natural light, but by 1500 to 2000 K colder than xenon lamps. CRI is compared to xenon lamps by approximately 20 % lower. The most efficient XP-G runs a maximum current of 1.5 A, unlike the other LEDs operating at a current up to 1 A. For these LEDs, a running at 0.7 A is recommended by manufacturer. Against that, for XP-G a maximum current drive at 1 A is recommended. EndoLED is driven by a current of 0.35 A.
Here, high efficacies comes with wide emission angles of radiation, which are unfavorable for light coupling. As secondary lenses (10034, Carclo, UK) were used to couple light into the fiber cone, wide angles of radiation were countered with specific lenses that are optimized for coupling into 6 mm optic fiber diameter by a distance of 16.8 mm. Therefore illuminances of LEDs before and after coupling into the endoscope were examined equal to measurements before. The manufacturer states efficacy for coupling LED emission into optical fibers with a diameter of 3 mm to 43 %. As the observed fiber diameter is only 2.8 mm, the coupling efficacy & is to be expected lower than theoretical maximum of 43 % (compare 4). No further loss is expected during light transport, as total reflection causes no loss and due to short transport length, no absorption is expected.
For xenon lamps are of low efficacy compared to LEDs, special cooling is essential. Regarding SMD LEDs, dissipation loss is lower. Thermal resistance of the LED pad should be low though, as to have a maximum heat flow. It is expected, that XP-G having the lowest thermal resistance is a feasible replacement for the conventional xenon lamp.
The setup consisted of an adjusting desk allowing translation of the LED in three dimensions. LED on PCB and thermal pad was mounted onto a designed aluminum heat sink. In order to enable optimal heat flow, inter layer was coated with heat transfer paste with high thermal conductivity 0.65 W/mK (RS Components, UK) compared to 0.02 W/mK for air. Conductive lines were soldered onto the LED PCB and supplied by laboratory power (HM8143, Hameg, Mainhausen) enabling constant current. Light beams were then adjusted overhead into the fiber cone as part of the light adapter of the ureteroscope, inserted into an integrating sphere 30 mm (test procedure standard) and quantified by digital analyzer (B360, Lichtmesstechnik, Germany). Heat flow was controlled by IR- thermometer (Lasersight, optris, Berlin). All three types, three LEDs per type, were examined equally. Finally homogeneity of intensity in the field of view was observed by goniophotometer (optronik, Berlin) with respect to the working distance of 15 mm over an angle of 34° in both directions and a step size of 2°.
III. Results and Discussion
Examinations of the different LEDs showed a significant increase in illuminance efficacy and promise therefore a substitution of xenon lamps in the exploration of small cavity bodies. Table III displays resulting percentage of mean illuminance under different currents. Cree XP-G turned out to be sufficient as light source for ureteroscopes.
Table III Percentage of illuminance dependant of the current of LEDs and EndoLED
illustration not visible in this excerpt
[1] compared to requirement (2.10 klux); [2] chip bottom to endoscope light adapter
As XP-G can be driven by a current of 1 A under consideration of a long lifetime, 2.2 klux overcome required 2.1 klux of xenon lamps by 4 % and performs a twofold of the illuminance of EndoLED for laryngoscopes (1.2 klux).
Minimal coupling diameter is observed to be at 14 mm. As the EndoLED has a coupling distance of 27 mm, a miniaturization of the light source is feasible.
Here, no critical thermal flow on the heat sink and the PCB, respectively, was observed. As EndoLED is driven by 0.35 A only and experiments didn’t provide heat flow measurements concerning the design of a lighting housing, more heat dissipation especially at 1 A is expected. Further investigations towards the heat flow under user conditions have to be carried out.
An average coupling efficacy of only 24 % was observed within the measurements for all LEDs with respect to the theoretical maximum of 43 %. Stated coupling efficacy by Carclo of 43 % for 3 mm fiber was therefore not reached. Residual loss might be due to smaller fiber diameter (2.8 mm) and insufficient adjustment.
illustration not visible in this excerpt
Fig. 4. Relative intensity in the field of view of xenon lamp (left), XR- E (middle) and XP-G (right). Measurements were carried out by goniophotometer. XP-G provides at best 56 klux and a homogenous FOV.
As the small XP-G turned out to be optimal for xenon lamp substitution and as replacement of the EndoLED, illumination homogeneity of xenon and XR-E were examined and compared to most efficient XP-G. The lighter the color, the higher the relative intensity is. The cross marks the point of highest illuminance in FOV (Fig. 4).
For xenon and XR-E the illumination is nearly homogenous. The brightest spot is reached by the xenon lamp (84 klux), followed by XP-G (56 klux) and XR-E (47 klux). The xenon lamp spot fills the FOV best. In contrast to XR-E newer XP-G provides a larger FOV with light.
endoscope. Thermal dissipation is provided by cooling fins of the synthetic housing. With regard to an autoclavable device, further examinations on the feasibility of cooling fins are essential as a high thermal flow is requested.
IV. Conclusions
It was demonstrated that the replacement of xenon lamps by LED lightings is feasible. XP-G LED supplies high illuminance not only in urology being the starting point of this examination but also for greater body cavities. It fulfills the requirement of an illumance of 2.1 klux when plugged onto an ureteroscope. The coupling distance could be reduced to 14 mm, for which reason the device can be minimized. Smaller LED lightings were feasible by using an energy interface bridging the energy gap between light source and camera head instead of batteries. Further investigations concerning heat development of the device under user conditions and users perception for LED color temperature and color rendering, respectively, are recommended.
References
[1] M. R. Krames et al., “Status and future of high-power light-emitting diodes for solid-state lighting,” IEEE Journal of Display Technology, vol. 3, no. 2, pp. 160-175, June 2007.
[2] T. Hu, P. Allen and D. Fowler, “In-vivo pan/tilt endoscope with integrated light source,” IEEE Conf. on Intelligent Robots and Systems, pp. 1284-1289, Oct. 29 2007 - Nov. 2 2007, San Diego, CA.
[3] H. Yanai et al., “Preliminary experience with a gastrointestinal endoscope using a white light-emitting diode,” Thieme eJournals: Endoscopy, 38(3): 290-291, 2006, New York.
[4] Y. Ohno, “Color rendering and luminous efficacy of white led Spectra,”Proc. Of SPIE, Vol. 5530, 2004.
[5] P. Bodrogi, “Color rendering: past, present (2004) and future,” CIE Expert Symposium on LED Light Sources, June 7-8 2004, Tokyo.
[6] W.-J. Becker, K.-W. Bonfig and K. Höing, “Handbuch elektrische Messtechnik, “ Hüthig Verlag, Heidelberg, 1st edition, 1998.
illustration not visible in this excerpt
Fig. 5. Design of the LED lighting for ureteroscopes. The Lighting can be plugged mechanically onto the endoscope light adapter and is a modified (minimized) version of the EndoLED.
The resulting design of the LED lighting using Unigraphics is shown in Fig. 5. It consists of an aluminum insert, LED body with secondary optics and a mounted aluminum heat sink. Light is focused onto the plugged light adapter of the
CE certification of TT! Imaging
Jan Krieger, Burkhard Zander, Dagmar Lühmann
Abstract—Products subject to certain EC directives providing for CE marking have to be affixed with the CE marking before they can be placed on the European market [10]. Manufacturers have to check, on their sole responsibility, which EU directives they need to apply for their products [5]. For medical devices is this the Medical Device Directive 93/42/EEC [6]. The aim of this internship was to carry out the CE certification for the TTI (Transamerican Technologies International [9]) Imaging software from EYETEC. Its a picture capture and processing software which is used in the ophthalmology. The most important part from the directive 93/42/EEC is the risk analysis which accounted for the bulk of the work. For this purpose 40 risks from 30 different causes and 26 measures to minimize the risks were worked out in detail. After completing the conformity assessment procedure, we issued the conformity certification and reported the medical product to the authorities. EYETEC is now allowed to sell TTI Imaging on the European market.
I. Introduction
A. The European Union and the CE certification
In 1993, the members of the European union created the single European market. This economic zone guaranteed for all members: free movement of goods, free movement of capital, free movement of services and free movement of people [1].
At that time, the statutory provisions of the member states regarding effectiveness, safety and performance of products differed in content and scope.
Furthermore there were disparities in the certification and inspection procedures from one Member State to another. The national provisions for the safety and health protection of patient, users and, where appropriate, other persons, with regard to the use of the product or the installation should be harmonized in order to guarantee the free movement of such devices within the internal market without any national borders.
An important step was the introduction of Europe-wide harmonized technical standards. These technical standards are required for all Member States and are worked out by three different European standardization organizations:
CEN (Comit Europen de Normalisation),
CENELEC (Comit Europen de Normalisation lectrotechnique) and
ETSI (European Telecommunications Standards Institute)
J. Krieger, Medizinische Ingenieurwissenschaft, Universität zu Lubeck, the work has beem carrid out ät EYETEC GmbH, Maria-Goeppert-Str. 1 23562 Liibeck (telephone: 017663282669, e-mail: jan-krieger@web.de).
B. Zander, managing partner EYETEC GmbH (telephone: 045150570360, e-mail:bz@eyetec.com).
Dr. med. D. Lümann, Universiätsklinikum Schleswig-Holstein, Institut fíir Soziälmedizin (telephone: 04515005881, email: Dagmar.Luehmann@uksh.de)
The employees of these organizations consist of representatives of the member states. Each member state has, according to its economic power, weighted votes and is so well able to accommodate their own economic interests [2]. The international standardization is also worked out by the three different organizations:
ISo (International organization for Standardization),
IEC (International Electrotechnical Commission) and ITU (International Telecommunication Union) [3].
The cooperation between ISo and the European Committee for Standardization (CEN) is regulated by the Vienna Agreement [4].
B. The general process of the conformity assessment
The aim of the CE marking is to guarantee a minimum level of health and safety standards for the consumers and ensure a Europe-wide competition without national boarders. Each manufacturer or authorized representative, who wants to bring a product on the European market, must prove that the product complies with the relevant European standards.
The proof of the so-called conformity is carried out with the conformity assessment procedures. Therefore, the responsible person can use the technical standards which are mentioned above [5]. Upon the successful completion of the conformity assessment procedure, the responsible manufacturer have to affix the CE mark (Fig. 1) to the product [6]. The requirements for medical products to be sold
illustration not visible in this excerpt
Fig. 1. Guideline-compliant illustration of the CE marking
on the single European market are defined in the CoUNCIL DIRECTIVE 93/42/eec of 14 June 1993 concerning medical devices [6]. Furthermore, the EU Directive on in vitro diagnostic medical devices (98/79/EEC) and the EU directive on active implantable medical devices (90/385/EEC) could be relevant [7].
The Medical Device Directive (93/42/EEC) and the Directive about active implantable medical devices (90/385/EEC) received with the Directive 2007/47/Ec an update [8].
In Germany is the implementation of these three directives ruled by the ”Medizinprodukte Gesetz” (MFG) (medical product law).
The manufacturer or an authorized representative have to check, which European (or equal national) directives apply for the product.
Afterwards he has to prove, that he observes all condition. For this purpose, it is possible to use harmonized standards, which suggests a presumption of conformity [5]. Furthermore, the manufacturer or the authorized representative is required to document the entire process and keep the records for an audit.
For the conformity assessment procedure of a software can be relevant the following harmonized standards (in Germany):
- DIN EN ISO 14971:2009 Medical devices Application of risk management (German version)
- DIN EN ISO 62304:2006 Medical device software Software life-cycle process (German version)
II. Material and Methods
A. The conformity assessment procedure
The aim of this work, was to carry out the conformity assessment procedures for the TTI (Transamerican Technologies International [9]) Imaging software from EYETEC. The most important document was the DIRECTIVE 93/42/eec. To ensure a complete risk management the harmonized standard DIN EN ISO 14971:2009 was used. The standard DIN EN ISO 62304:2006 was not used because it requires a monitoring of each lifetime phase and at the time of the conformity assessment the software was already finished and a recapitulation of the programming process not possible.
1) TTI Imaging: TTI Imaging is a picture processing software to be used by ophthalmologist, optometrist and laboratory personnel in the ophthalmology.
The software works in combination with a on a Slit-lamp mounted USB-Camera and supports the user recording photos and videos and save these in a database.
Furthermore, the software delivers a lot of different picture and video processing tools like an assortment of filters (median, Laplace, etc.) threshold operations, marker and the possibility to do several measurements (length, volume, etc.) or separate specific picture objects.
For easier patient management affords TTI Imaging a implemented SQL database. Taken photos and videos will be automatically transferred in the database. The user can load the photos and videos from the database, process them in any way and put them afterwards back in the database.
2) The conformity assessment procedure of TTI Imaging: According to [Il], the TTI Imaging software is considered to be a medical device because:
1) in combination with other hardware, the software will be used to diagnose various eye diseases. In this context, the software can be used for:
- detection and monitoring disease and
- the study of physiological processes in the patients eye
2) The principal intended action in or on human body is neither pharmacologically or immunologically or metabolically achieved.
According to [l2], the TTI Imaging is classified as a Class I medical device.
Furthermore, according to [8], the TTI Imaging is classified as an active medical device.
illustration not visible in this excerpt
Fig. 2. Risk matrix example. During the risk analysis every risk must be assigned to a tuple from severity of hazard and probability of occurrence. Risk entries above the main diagonal must be reduced. The color code and the drawn ’’borderline” allow a quick overview.
B. Directive 93/42/EEC Annex VII
For Class I medical devices, the manufacturer or the representative has to follow the procedure referred to Annex VII of the directive 93/42/eec [13] and draw up the EC declaration of conformity.
Annex VII describes the act of EC declaration of conformity. In particular, it explains the structure of the technical documentation. It must include:
- a general description of the product, including any planned variants,
- the results of the risk analysis and a list of the standards referred to in Article 5, applied in full or in part, and descriptions of the solutions adopted to meet the essential requirements of the Directive if the standards referred to in Article 5 have not been applied in full,
- the solutions for annex I [14],
- the test reports and, where appropriate, clinical data in accordance with Annex X,
- the label and instructions for use.
C. The risk analysis
The most important part of the technical report is the risk analysis. To guarantee a complete analysis, the standard DIN EN ISO 14971:2009 was used.
The aim of the risk analysis is to carve out all risks, evaluate them according to the severity of hazard and the probability of occurrence.
Afterwards you have to figure out measures to reduce the risk and iterate the risk analysis. This process must be repeated until all risks have reached an acceptable level [15].
For easier risk assessment, we used a risk matrix as shown in Fig. 2. The X-axis shows the probability of occurrence and the Y-axis the severity of hazard.
The axis intervals can be arbitrarily chosen by the manufacturer, but in case of an incident it has to withstand a review. The scale from the hazard and the probability have to cover a reasonable area. Normally the probability goes from 10 (i to 10 :î. The arrangement of the hazard scale normally happen in fonn of examples (light hazard for the patient without clinical intervention or a bad database error etc).
Every carved out risk must be, according to its hazard and probability, allocate to one matrix field. Each risk with an entry above the main diagonal must be reduced through one or more measures. This can be done, either by reducing the severity of hazard or the probability of occurrence or both. The risk analysis is complete and successful if all risk entries are below or on the main diagonal.
D. The clinical evaluation
The second important part is the clinical evaluation. The clinical evaluation provides, that all requirements for the product are fulfilled under nonnal conditions. The adequacy of the clinical data must be based on: either a compilation of the relevant scientific literature currently available on the intended purpose of the device and the techniques employed as well as, if appropriate, a written report containing a critical evaluation of this compilation or the results of all the clinical investigations made [16].[3]
Furthennore, the risk was divided into two or three risks when different groups of people (user, patient, third person) may be affected.
Ultimately, 40 risks, 30 causes, 10 categories and 3 groups of potentially victims were worked out. The risk distribution to the user groups is:
- 23 risks for patients
- 15 risks for users
- 2 risks for third persons
Summarized, prior to the risk minimization, risk distribution was as shown in Fig. 3.
illustration not visible in this excerpt
Fig. 3. Risk distribution before the risk minimization. Some risk entries are above the main diagonal and must be reduced for a successful conformity assessment.
B. The risk minimization of TTI Imaging
The risk minimization takes place in form of various measures which have to be developed and applied. For the risk minimization of TTI Imaging 26 measures have been devised. Efficacy was assessed in all measures as ’’effective”. For three hazards were no measures for reduction of the risk found. But in that case, probabilities were so low that no problems have been occurred.
After the risk minimization, risk distribution was as shown in Fig. 4.
How the risk minimization looks in detail is shown as an example in affix I of this document.
C. The clinical evaluation of TTI Imaging
The clinical evaluation was, referring to annex X from Directive 93/42/EEC section l.ld (German version), exposed and not listed in the technical report for the TTI Imaging Software.
Directive 93/42/EEC section l.ld allows this if the decision is well-founded.
- The software TTI Imaging has no direct or indirect contact with the patient, user or a third person.
- All thinkable risks were documented and minimized as far as possible.
- After the risk analysis all risks are in an acceptable range.
illustration not visible in this excerpt
Fig, 4. Risk distribution after the risk minimization. All risk entries are on or below the main diagonal and have not to be reduced anymore.
- The conditions under which the risk analysis was made do not differ from the daily hospital routine.
1) measures:
- Ml: Implementation of a request, whether there is already a database entry with an equal name and an equal birth date.
- M2: Advise the user to the dangers of confusing database entries.
2) Estimation of effectiveness:
- Ml: effective
- M2: effective
B. Risk after risk minimization Risk for the patient:
- probability of occurrence ! column (l)
- severity of hazard ! line (d)
This result is a acceptable risk for the patient. The risk have not to be reduced further.
Thats how we proceed with all identified risks. The steps are documented in tables and the results are summarized in form of matrix entries at the end of the sections.
IV. Conclusions
Because of the risk assessment and the special probabilities of the product do we come to the following conclusion:
- All identified risk have been reduced as far as possible or necessary.
- The software TTI Imaging do not present a unjustifiable danger for the patient, the user or a third person.
After the successful completion of the CE certification process we issued the certification of conformity and reported the Medical product to the authority. EYETEC is now allowed to sell TTI Imaging on the European market.
Acknowledgment
My thanks goes to the managing partners of EYETEC: Burkhard and Enno and all employees: Kathrin, Heike, Noppi, Bernd and Enrico.
A special thanks goes to Mrs. Dr. Luhmann who has agreed to look after my internship.
affix I
Risk minimization example:
Category: operation
Phase of life: input of a new patient in the database.
safety objective: confusion of the patient with an already
existing patient with the same name. Causes: input of two equal patients is possible.
A. Risk before risk minimization Risk for the patient:
- probability of occurrence ! column (3)
- severity of hazard ! line (d)
This results in a not acceptable risk for the patient. The risk must be reduced.
References
[1] Europa summaries of EU legislation, Internal market, Avaiilable: http://www.europa.eu/legislation_summaries/internal_market/index_en.htm, Jan 2012.
[2] Europäische Kommision, Allgmeine Leitlinie, Available:
http://www.umweltbundesamt.de/produkte/dokumente/ Allgemeine_Leitlinien_CEN_CENELEC_ETSI_und_EU_Komm.pdf, Jan 2012
[3] World summit on the information society, International Standards
Organizations (ISO, IEC, ITU and UNECE), Available:
http://www.itu.int/dms_pub/itu-s/md/03/wsispc2/c/S03-WSISPC2-C- 0045MPDF-E.pdf, Jan 2012.
[4] International Organization for Standardization, The Vienna Agrement, Available: http://www.iso.org/iso/about/the_iso_story_vienna_agreement.htm, Jan 2012
[5] Amt fur Amtliche Veröffentlichungen der Europäischen Gemeinschaft, Leitfaden für die Umsetzung der nach dem neuen Konzept und dem Gesamtkonzept verfassten Richtlinien, Available: http://europa.eu.int/comm/enterprise/newapproach.htm, Nov 2011.
[6] Richtlinie 93/42/EWG des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993.
[7] M. Norr, K. Tittel, MERKBLATT CE-Kennzeichnung von Medizinprodukten, Industrie und Handelskammer für München und Oberbayern, Available: http://www.muenchen.ihk.de/mike/ihk_geschaeftsfelder/innovation/ Anhaenger/CE-Kennzeichnung-von-Medizinprodukten2.pdf, Nov 2011.
[8] Richtlinie 2007/47/EG des Europäischen Parlaments und des Rates, Sep 2007, p. 53.
[9] TTI Medical Transamerican Technologies International, Available: www.ttimedical.com/index.php, Feb 2012
[10] Institut der deutschen Wirtschaft - REHADAT, CE- Kennzeichnung bei Medizinprodukten und Hilfsmitteln, Available: http://www.rehadat.de/rehadat/Download/Hilfsmittel/ CE_Kennzeichnung.pdf, Nov 2011.
[11] Richtlinie 93/42/eWg des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993, p. 6.
[12] Richtlinie 93/42/Ewg des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993, p. 57.
[13] Richtlinie 93/42/Ewg des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993, p. 52.
[14] Richtlinie 93/42/Ewg des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993, pp. 26-34.
[15] DIN EN ISO 14971:2009, Anwendung des Risikomanagement auf Medizinprodukte (German Version), 2009.
[16] Richtlinie 93/42/EWG des Rates vom 14. Juni 1993 ber Medizinprodukte, June 1993, p. 62.
EMG-based estimation of wrist kinematics using Fisher’s linear discriminant analysis
Nina Rudigkeit, Liliana P. Paredes Calderon
Abstract-The paper proposes an approach of using Fisher’s linear discriminant analysis (LDA) for simultaneous and proportional control of multiple degrees of freedom (DOFs). We assumed that linearly transformed electromyography (EMG) data could be used to estimate performed movements in free space including wrist flexion/extension, radial/ulnar deviation and forearm pronation/supination. A mirrored-bilateral training strategy was used. Angular displacement estimates for the three DOFs from Fisher’s LDA were compared to kinematic data. The coefficients of determination were 72.2% for flexion/extension, 67.9% for radial/ulnar deviation, and 67.6% for pronation/supination. This experiment showed the feasibility of using Fisher’s LDA to estimate wrist kinematics.
I. INTRODUCTION
Modern upper limb prostheses only restore a small amount of the functionality provided by a natural limb. While advanced prosthesis can have 22 DOFs and more and provide functionality such as hand opening and closing, flexion and extension of the hand, rotation of the wrist [1], they still lack control schemes enabling amputees to use them intuitively. Recent research activities focus on pattern recognition algorithms that have achieved high classification accuracy under laboratory conditions but are not robust enough or failed in clinical applications.
Current commercially available prostheses use EMG signals from residual muscle sites to control multiple DOFs in a sequential manner. For a typical fitting, a below-elbow amputee has to contract wrist flexor and extensor muscles to control terminal device closure and opening. The range of motion is proportional to the amount of the exerted force. If the hand provides more types of motion (e.g. hand flexion and extension), switching techniques such as muscle cocontraction are commonly used to carry out hand flexion instead of hand closure and hand extension instead of hand opening (i-limb ultra [2], bebionic v2 hand [3], Michelangelo hand [4]). These switching techniques and sequential activation are unnatural, require extensive training and consequently have a low degree of acceptance when a large number of DOF has to be controlled. To achieve higher acceptance, systems that provide the ability of controlling multiple DOFs simultaneously are needed.
N. Rudigkeit, Medizinische Ingenieurwissenschaft, Universität zu Lübeck; the work has been carried out at Otto Bock Healthcare GmbH, Strategic Technology Management, Max-Näder-Str. 15, 37115 Duderstadt, Germany (e-mail: Nina.Rudigkeit@miw.uni-luebeck.de).
L. P. Paredes Calderon is with Otto Bock HealthCare GmbH, Strategic Technology Management, Max-Näder-Str. 15, 37115 Duderstadt, Germany (telephone: +4955278483416; e-mail: Liliana.Paredes@ottobock.de).
The simultaneous control of multiple DOFs movements was analyzed by [5-7] who proposed a modified non-negative matrix factorization approach to estimate force functions and joint angles using a mirrored bilateral training strategy. Other approaches by contrast suggested the use of artificial neural networks [8].
This study is part of a larger project which focuses on advanced signal processing for improved myoelectric control of prosthetic devices. The goal is to intuitively control multiple DOFs simultaneously and proportionally. This paper proposes an approach of using Fisher’s LDA for estimating wrist kinematics. With prosthesis control using Fisher’s LDA good classification results were received during previous studies [9]. This promising outcome led to the modified implementation proposed in this paper.
II. Methods
A. Subjects
One normally limbed subject participated in the experiment. As the subject had no known neuromuscular disorders and an intact muscular system he provided a best-case scenario for investigating the performance of the algorithm.
B. Procedures
According to the possible movements of the wrist three DOFs could be activated individually or jointly. Opposed movements were assigned to one DOF, i.e. the first DOF referred to wrist flexion and extension, the second one to radial and ulnar deviation and the third one to forearm pronation and supination. As listed in Table I, the protocol included six single DOF movements and four combined ones.
The subject performed a series of mirrored bilateral dynamic wrist contractions. This training strategy was chosen to investigate the clinical applicability of the algorithm for unilateral amputees. Each contraction had a total duration of approximately 6 s including a resting phase of approximately 3 s between consecutive movements. Every contraction started from the resting position which was defined as the subject standing, elbows flexed at 90° and palms facing inwards. The subject was instructed to reach the target position in 1 s (second dynamic phase), maintain it for 1 s (static phase), and then return to the resting position within 1 s (first dynamic phase).
TableI Movements included in the Protocol
illustration not visible in this excerpt
To capture the kinematics of the subject during the recording session a motion capture system (Qualisys Track Manager, Qualisys AB) with eight infrared digital video cameras (ProReflex MCU, Qualisys AB) was used. The cameras were mounted on tripods and placed in a circular pattern around the subject at various heights. The exact positions were optimized during preliminary tests in order to have a view of all markers. seven reflective spherical markers were symmetrically placed on the right arm (Fig. 1). one was positioned on the prominent point of the scapular acromion, two parallel to the medial (МЕР) and lateral (LEP) epicondyles of the humerus, two at the distal styloid processes of radius (STR) and ulna (STU) and the last two at the radial (RMC) head of the second metacarpal bone and ulnar (UMc) head of the fifth metacarpal bone. The marker trajectories were digitally saved in a 3D coordinate space.
illustration not visible in this excerpt
Fig. 1. Placement of reflective markers and coordinate system for joint angle calculation. LEP: Lateral epicondyle of humerus. МЕР: Medial epicondyles of humerus. STU: Styloid process of ulna. STR: Styloid process of radius. RMC: Radial head of second metacarpal bone. UMC: Ulnar head of the fifth metacarpal bone. Marker at scapular acromion (shoulder) not shown. E: Elbow, mid-point between LEP und МЕР. O: Origin, mid-point of STU and STR. H: Hand, mid-point of UMC and RMC. The z-axis points from O to E, the x-axis from O to STR and the y-axis is the normal vector of the xz- plane.
Surface EMG-signals were recorded with six pairs of Ag- AgCl surface discrete electrodes in single differential mode (Ambu NeuroLine 720) which were placed equally spaced around the forearm. To ensure good contact the skin was shaved and lightly abraded prior to electrode placement. The arrays were connected to an EMG-amplifier (EMG-USB, ОТ Bioelettronica) with which the data was amplified at 500, sampled at 2048 Hz and digitized with 12 bit precision.
C. Signal analysis
The kinematics data and the EMG signal were synchronized to eliminate the delay between the motion capture system and the EMG recording. Afterwards, the signal was band-pass filtered using a third order Butterworth digital filter (pass-band 20-250 Hz) to remove DC offset, motion artifacts and high- frequency noise.
The filtered data was divided into windows of 200 samples, corresponding to approximately 100 ms, with an overlap of 75% between consecutive windows. For each channel features were extracted from every window. For computational simplification only time-domain features, namely root mean squares (RMS), waveform lengths (WL), sign slope changes (SSC) and zero crossings (ZC), were considered. The obtained features were used as the input data for further processing.
Two sessions consisting of the continuous execution of the joint movements (1 to 10 from Table I) with resting periods in between (movement 11, Table I) were recorded. The data from the first session was used for training the transformation matrices for each DOF independently; the data from the second session was used for the testing phase. Both training and testing were performed offline.
Class separation and simultaneous dimensionality reduction of the feature data were achieved by using a multiple class approach of Fisher’s LDA. Fisher’s LDA was extended to also estimate joint kinematics.
1) Fisher’s LDA for multiple classes
In general, the multiclass Fisher LDA proceeds in the following way:
The feature matrix X is transformed as follows:
illustration not visible in this excerpt
where W denotes the weight matrix which columns assign a certain weight to every feature according to the magnitude of its influence on the target output. To obtain the weight matrix the following steps have to be performed:
Let X = (xv ...,xn) be a matrix with containing n feature vectors x# of length m with known class affiliation C1,..., ,
where ' indicates the number of classes. The number of points belonging to class % is denoted by n(. The mean values for each class μ( and for the entire sample μ are then given by (2)
The scatter matrix Si is defined as (4)
The sum of all scatter matrices is called within-scatter matrix because it gives information about the scattering of
Student Conference on Medical Engineering Science 2012
data within each class in contrast to the between-scatter matrix SB which provides information about the scattering of data between the classes.
illustration not visible in this excerpt
The idea that the optimal projection direction is the one in which the variances between the classes are maximized while the ones within the classes are minimized can mathematically be formalized by maximizing to objective function J(W) [10]:
illustration not visible in this excerpt
In this case SB stands for the projected between-scatter matrix while Sw denotes the projected within scatter-matrix. The criterion for optimality is:
illustration not visible in this excerpt
With the help of (8) the columns w; of an optimal W turn out to be the generalized eigenvectors. The eigenvalues can be obtained by solving
illustration not visible in this excerpt
Afterwards, the eigenvectors can be found using
illustration not visible in this excerpt
The transformation matrix w is composed by sorting and merging the eigenvectors in descending order of the joint eigenvalues:
illustration not visible in this excerpt
2) Training and testing
During training, transformation matrices were calculated for each DOF independently because it was assumed that the optimal two directions of projection were different for every DOF. The key assumption was that if the dimensionality of the feature space was reduced to two then the directions of the eigenvectors referred to the opponent movements encoded in each DOF. This assumption developed from observations taken by visualizing the results of Fisher’s LDA.
In order to perform Fisher’s LDA discrete target values were necessary. Therefore, the minimum and maximum values for each DOF of the ideal target data were determined. If a certain rate was exceeded the gesture was either classified as + 1 (e.g. flexion) or —1 (e.g. extension). Below these thresholds the gesture was categorized as 0 (resting position). Additionally, for each DOF the periods of no activity corresponded to the resting positions between opponent movement of the corresponding DOF and during the activation of the other DOFs. The magnitude of the chosen threshold experienced to have a great impact on the classification results. Both overall and individual thresholds for each DOF have been tested. Individual thresholds have not shown improved performance. The results discussed in this paper have been obtained with a threshold of 0.8 for all DOFs.
Dimensionality reduction was realized by composing W only of the vectors corresponding to the two greatest eigenvalues. Once the transformation matrices were trained, we proceeded to test with the data from the second session. This data was projected to every two dimensional space given by each transformation matrix. The summation of the transformed input data across its features could be interpreted as an estimation of the intended wrist kinematics (Fig. 2).
illustration not visible in this excerpt
Fig. 2. Flowchart of Fisher’s LDA.
Estimation performance was calculated using the coefficient of determination R[2] for each DOF:
illustration not visible in this excerpt
The estimated angle is denoted asä; a is the measured angle; Cov(·) denotes the covariance between both angles and Var(·) the variances. The measured angles were calculated from the kinematic data as described in [8].
III. Results and Discussion
In a first run testing was performed with the same data used for training. The coefficients of determination were 34.7% for flexion/extension, 54.3% for radial/ulnar deviation and 31.7% for pronation/supination. Preliminary tests were done limiting the kinematics bandwidth from 0 Hz to 6 Hz which is limited by the physiological ability of neuromuscular system, arguably lower than 5 Hz. [11]. Low-pass filtering at 6 Hz led to the following R[2]-values: 57.0% for flexion/extension, 73.0% for radial/ulnar deviation and 64.1% for pronation/supination. Performance increased when testing was performed with the data of the second session. The coefficients of determination were 72.2% for flexion/extension, 67.9% for radial/ulnar deviation and 67.6% for pronation/supination. This might be the result of higher amplitudes of the EMG-signals in the second data set. In further studies tests will be taken on more data sets.
illustration not visible in this excerpt
Fig. 3. Angles measured with QTM and estimated with Fisher’s LDA. In this example, the coefficients of determination for this run were 34.7% (flexion/extension), 54.3 % (radial/ulnar deviation) and 31.7%
(supination/pronation).
IV. Conclusions
This study demonstrated that our approach using Fisher’s LDA is capable of estimating continuous and simultaneous joint kinematics, with ň[2]-values of approximately 72.2% for flexion/extension, 67.9% for radial/ulnar deviation and 67.6% for supination/pronation in nondisabled subjects using features extracted from EMG signals obtained from muscles that would be still present after transradial amputation. These results indicate that the signals contain a significant amount of information related to arm movements. With further development, it may be possible to achieve better performance.
Further research will focus on optimization of the threshold for gesture assignment, different assumptions of the no activity for each DoF, change of the dimensionality of the feature space to three to investigate the effect of the resting position on the estimation results and even classification and joint kinematics estimation for all the DoFs at once.
Acknowledgements
We like to thank Dr. B. Graimann and Dr. N. Jiang who supported this work.
References
[1] J. M. Burck, J. D. Bigelow, S. D. Harshbarger, Revolutionizing prosthetics: Systems engineering challenges and opportunities, Johns Hopkins APL Technical Digest, Vol. 30, No. 3 (2011)
[2] i-limb ultra, http://www.touchbionics.com/products/active-prostheses/i- limb-ultra/, 2012-27-01
[3] bebionic - The Hand, http://www.bebionic.com/category/the-hand/, 2012-27-01
[4] Otto Bock, http://www.ottobock.de/
[5] N. Jiang, K. B. Englehart, and P. A Parker. Extracting simultaneous and proportional neural control information for multiple degree of freedom prostheses from the surface electromyographic signal. IEEE Transactions on Biomedical Engineering, 56(4): pp. 1070-080, 2009.
[6] J. L.G. Nielsen, S. Holmgaard, N. Jiang, K. B. Englehart, D. Farina, and P. A. Parker. Simultaneous and proportional force estimation for multifunction myoelectric prostheses using mirrored bilateral training. IEEE Transactions on Biomedical Engineering, 58(3): pp. 681-688, march 2011.
[7] S. Muceli, N. Jiang, and D. Farina. Multichannel surface EMG-based estimation of bilateral hand kinematics during movements at multiple degrees of freedom. In Engineering in Medicine and Biology Society (EMBC), 2010 Annual International Conference of the IEEE, pp. 60666069, 31 2010-sept. 4 2010.
[8] S. Muceli, D. Farina, Simultaneous and proportional estimation of hand kinematics from EMG during mirrored movements at multiple degrees- of-freedom, in IEEE transactions on neural systems and rehabilitation engineering, 2011 Dec 13. [Epub ahead of print]
[9] Chu, J.-U., Moon, I. & Mun, M.-S., 2006. A supervised feature projection for real-time multifunction myoelectric hand control. Conference Proceedings of the International Conference of IEEE Engineering in Medicine and Biology Society, 1, p.2417-2420, 2006
[10] R. O. Duda, P. E. Hart, D. G. Stork, Pattern classification, John Wiley & Sons, pp. 117-124, 2000
[11] D. A. Winter, Biomechanics and motor control of human movement. Wiley-Interscience, 1990.
Material compatibility with different sterilization procedures
Yannik Schröder, Johannes K.-M. Knobloch
Abstract-Sterilization procedures that ensure a sterile product for the next usage of reprocessable medical devices have not only an effect on biological organisms but also on the product itself. To assess the material changes produced by the sterilization, tests with different sterilization procedures (ozone based, hydrogen peroxide based and formaldehyde based) have been performed. The results show that the extent of changes is highly dependant on the used procedure. However, some materials seem to be more sensitive to the effects of sterilization than others. Especially adhesives were affected. Changes in the adhesive power as well as in colour and texture were observed in all tests. Also corrosion occurred locally on some samples.
I. INTRODUCTION
Many medical products such as endoscopes or surgical instruments are designed to be used multiple times. After use, the instruments are soiled with blood or biological tissue and contaminated with germs. To avoid infecting the next patient these instruments are used on, reprocessing is necessary. The goal is a clean and sterile state so that no blood and germs remain on the product. To achieve sterility the soiled instruments need to be cleaned first. This coarse cleaning should remove any blood and soil to a certain extent. In many cases water and a cleaning agent together with manual methods like brushing is sufficient. After cleaning the products need to be sterilized. There are many different procedures available for sterilization, for example autoclaving, ethylene oxide sterilization or hydrogen peroxide treatment. Every procedure has its advantages and disadvantages (see Table I). Therefore it is difficult to determine the best even though steam sterilization is the most common one [1] and well regulated (see [2] for further reading).
Table I Advantages and disadvantages of different sterilization PROCEDURES [3]
illustration not visible in this excerpt
All of these reprocessing procedures have in common that they stress the instruments that are sterilized. In a worst case, the instruments get damaged and cannot be used anymore
Y. Schröder, Medical Engineering Science, University of Lübeck (yannik. schroeder @ web.de).
J. K.-M. Knobloch, Department of Medical Microbiology and Hygiene, University Medical Center Schleswig-Holstein, Campus Lübeck.
without putting the well-being of the patient at risk. Therefore an assessment of the changes or damages that occur while reprocessing is necessary, since the manufacturer of reusable medical devices is committed to provide information about how their products can be reprocessed [4]. In this work the impact of different sterilization methods on different materials, such as metals, plastics, elastomers and adhesives, and medical devices is evaluated.
II. Material and Methods
The sterilization procedures that are evaluated in this work are: two processes of ozone sterilization, hydrogen peroxide sterilization and formaldehyde sterilization. The sterilization is to be tested solely; no prior cleaning or disinfection is performed. The most standardized sterilization procedure - steam sterilization - is no content of this work. Since most of the exact mechanisms of actions are not disclosed only the basic principles shall be touched in the following section.
The ozone based sterilization procedure uses hydrogen peroxide (H2O2) and ozone (O3) as sterilization agents. Free radicals are formed first by the hydrogen peroxide and then by the reaction of ozone with hydrogen peroxide, water and other radicals. Those radicals can oxidize other materials and organic compounds and thus kill the germs.
The hydrogen peroxide based sterilization procedure only uses hydrogen peroxide for sterilization. Similar to the ozone based process free radicals are created which can kill the germs.
Lastly, the formaldehyde based sterilization uses saturated steam and formaldehyde as sterilization agents. The germicide effect is based on the alkylation of proteins. Germs are changed in a way that they are not able to breed afterwards. This combination of thermo physical- and chemical processes ensures a sterilizing effect.
All of these methods are low temperature procedures which are suitable for heat sensitive instruments and materials. The exact testing process is described in the following.
First of all, prior to the test, the testing parameters need to be defined. This was done individually for each sterilization process test by creating a test plan. The test plan defines which samples are to be tested (A.), what parameters are to be tested (B.), which machines and programs are to be used (data not shown due to patent restrictions), how long the test shall last and in which intervals the items shall be inspected (C.).
A. Test samples
The choice of test samples needs to be adapted to the kind of sterilization procedure. In this work only general sterilization methods applicable for many kinds of items (like endoscopes, cables and surgical instruments) are evaluated. To get comparable results, similar test sets were used for most of the tests. These sets consist of test plates made of different materials, (parts of) surgical instruments, light guide cables (see Fig. 1a) and cables used for high frequency (HF) surgery (see Fig. 1b), endoscope dummies, adhesive test plates and screw plates. For details see Table II.
Table II Items used for testing
illustration not visible in this excerpt
EPDM: ethylene propylene diene monomer, PTFE: Polytetrafluoroethylene (Teflon), PEEK: Polyether ether ketone, PPSU: Polyphenylsulfone, LCP: Liquid-crystal polymer, PPS-GF40: Polyphenylene sulfide reinforced with 40% glass fibres, Adh. : adhesive, SS: stainless steel, HF: high frequency, PDD: photodynamic diagnosis.
The test plates are used to obtain information about how the sterilization process affects specific materials. While most of the instruments are made of more than one material, information about a certain material can be worthwhile in the process of developing a new product.
illustration not visible in this excerpt
Fig. 1. a) A light guide cable before sterilization, b) A HF-Cable before sterilization
Complete- or parts of surgical instruments (like handles or sheaths) are integrated in the test to observe the changes to a manufactured instrument. These instruments often consist of a combination of plastic, steel, adhesive and other sealing materials and can therefore describe a good comprehensive package of materials.
Endoscope dummies and light guide cables are tested to observe the influence of the sterilization procedure on their light transmittance ability. These procedures can affect the adhesive between the optical fibres or the fibres itself and make them vulnerable to abrasion, which can then happen in the required cleaning process prior to the sterilization.
Adhesive test plates consist of a steel plate with adhesive on it. They represent a worst case of adhesive placement, since the adhesive can be affected by the sterilization process from almost any side. This is not possible in a real application. Different adhesives have been tested.
The screw plates consist of plates with glue reinforced screws bolted to them. The screw sizes are (going by the ISO metric screw thread) M4, M3 and M2. Based on these plates the change in adhesive power can be observed.
B. Tested parameters
At the specified inspection intervals (see C.) visual tests were fulfilled. The aim is to check if the sterilization procedure affected the samples in any way. This may be leaked adhesive, corrosion, discolourations, damaged labellings or residues. For documentation purposes photographs were taken from conspicuous areas. Additionally transmittance tests have been performed with the optical dummies and the light guide cable to detect changes in the ability to conduct light originated from the sterilization. Furthermore dielectrical strength tests with the instruments used for high frequency surgery have been performed to check the function of insulations. The conductivity of the same instruments has been monitored by a resistance measurement. As a last test the screw torque needed to loose the screws out of the screw plate has been measured. Since this is a destructive test enough screw plates for a test at the required inspection intervals have been integrated into the test.
All of these tests have been performed prior to the sterilization processes as well for reference purposes.
C. Test length and inspection intervals
The total test length is set to 400 sterilization cycles with intermediate checks at 50, 100, 200 cycles and a final test at 400 cycles. Since some of the tests are still in progress and have not been completed yet, the results refer to the intermediate check that was performed after 50 cycles if not stated otherwise.
III. Results and Discussion
Test results for the various samples and processes have been achieved. This chapter focuses on the most important visible and measurable changes.
A. Ozone sterilization I
Many of the test items have been affected by the reprocessing procedure.
Chrome-plated metal showed many small spots of corrosion (see Fig. 2). It seems the plating is affected by the procedure and therefore makes the metal vulnerable to oxidation processes. The metal tube coated with Tinnol 3000 got affected even heavier. Instead of local corrosion spots as seen on the chrome-plated test samples the coating corroded completely. The electrode loop instrument showed major oxidization and thus the testing process for it was stopped after the third cycle.
illustration not visible in this excerpt
Fig. 2. Corrosion spots on the chrome-plated metal 2.0401 after
treatment with 50 cycles of ozone sterilization
Surfaces of glass fibre containing polymers got roughened up and the glass fibres got exposed. The gilded connectors on the handles corroded locally. Dramatic colour changes were observed on the outer tubes of the HF- and light guide cables. These changed from grey to white and brownish respectively. The bend relief on the light guide cable turned from black to transparent. Sealing ring materials, such as EPDM or Viton had major visible and haptic texture changes and thus those samples have been removed from the test before the 50 cycle mark. The functional tests, like transmittance measurement and electrical safety, were passed by all items without issues.
However, since the overall result showed major alterations/damages it was decided to stop the test after the check at 50 cycles.
B. Ozone sterilization II
Because of the unsatisfying test results of the first ozone based sterilization the device manufacturer optimized the procedure and thus a new test was started.
While some of the metal test samples still had some spots of corrosion, the magnitude was much smaller than in the first test. The plated metals 2.0401 and 2.4360 (plated with Tinnol 3000) showed no signs of corrosion whatsoever, only the metal 2.4360 plated with chrome had small spots as well as the steels 1.4021 and 1.4301. Also, the major colour changes of the cables as observed on the first ozone test did not occur. This time however, there was a white local coating on some metal test samples and especially on the electrode loop, which needed to be cleaned with ethanol for the resistance measurement. This precipitation most likely resulted from residual agents which accumulated during the process. In a real reprocessing situation with intermediate cleaning this accumulation would not occur.
Some laser labellings were partially decoloured and had a “worn-out” look, while they were still readable without problems. Plastics and sealing materials weren’t affected. The adhesives on the test plates did not detach but some turned dull and discoloured, especially on the edges. Changes were also observed in the adhesive power, most notable on adhesive no. 20 which bonding power roughly halved. But no general conclusion can be made for all adhesives, since some of them actually gained power through the sterilization process.
However, the tolerances of the measuring process have to be taken into account. Again, all of the functional tests were passed.
Generally the results of this test were much better than those of the first ozone test. The alterations to the test samples weren’t grave so the test was continued. Final conclusions will be drawn after the end of the test.
C. Hydrogen peroxide sterilization
Plated metals and steel were in an impeccable condition after 50 processed cycles. No corrosion or other degradations were found on any of the test samples including instruments. However, adhesives proved as problematic. On some instruments like the optical dummy and the light guide cable the adhesive leaked or got brittle. Also the adhesive 41 separated completely from the two test plates, adhesive 31 separated from one. The other adhesives on the test plates showed signs of beginning separation on the edges. Discolourations and texture changes (dullness) were visible on some of the adhesives of the test plates. Changes in adhesive power were not consistent. Again, adhesive no. 20 had the biggest loss in adhesive power through the sterilization process which almost halved. Adhesive no. 41 lost power too, but not to the extent of adhesive no. 20. Adhesive 31 gained power, while the power of other adhesives stayed roughly the same.
Other noteworthy conspicuities were titanium nitride coatings which got affected by the process and showed light discolourations. Also, some test samples had some white stained areas which looked like water marks. The same white coating/debris was most distinctive on the noose of the electrode loop. Once again, all functional tests were passed.
Since the test for the hydrogen peroxide based sterilization has already been completed (400 performed cycles) some final results shall be presented in the following. Most plated metals and steels were still in a very good condition and showed no signs of corrosion. Only the 1.4301 steel had very few spots of corrosion. Some labels were brittle and/or discoloured, while most of them were still readable without problems. However, more instruments leaked adhesive and titanium nitride coatings were clearly more affected as they showed disintegrated areas. Lastly the glass fibre containing polymer PPS-GF40 has been bleached slightly.
D. Formaldehyde sterilization
While most of the metal test plates showed no spots of corrosion, some of them had widespread light brownish stains on them, which looked like water marks. Brownish discolourations were also observable on the crimped area of the electrode loop and the preinduced crack on the 2.0401 metal test plates.
Furthermore the jaw mobility of a few instruments (like the needle holder) was worse than at the beginning of the test. This effect is related to the oiling of the hinge and can be reversed. Adhesive no. 41 separated from both test plates, while the others adhesives only showed discolourations if any. Labels were in a flawless condition and once again, all functional tests were passed.
IV. Conclusions
The first ozone based procedure showed the worst results of all tests. Many instruments and materials got affected. The damage ranged from colour alterations to serious corrosion. Some materials for example the sealing materials were rendered useless as the texture changed. Following these results the test was stopped and the procedure was optimized.
The results of the optimized procedure were much better. The occurring corrosion was reduced significantly and no damages to plastics were detectable. This shows that a procedure is not necessarily unsuitable per se. The used parameters like temperature, the amount of agent or cycle length have a big influence on the outcome of a test as well.
None of the procedures had an influence on the transmittance of the test dummies and light guide cables. This shows that the optical fibres in the instrument/cable and on the connector pieces as well as on the ends of the endoscope were not affected by the procedures.
The same results for the electrical safety. The insulation and other critical parts regarding the safety have not been affected by any of the sterilization processes.
The occurring separation of some adhesives from the test plates should not be overemphasized, since they represent a worst case scenario that is unlikely to happen in the design of an instrument. The main purpose of these plates is to see potential colour and texture changes. They are not suited for an assessment of the adhesive bonding power. More relevant results can be gained from the screw plate tests, since the adhesive is applied in a more realistic way. To show the different impact on adhesives of different procedures and count of cycles, the screw plate test results (screw size M4) of the hydrogen peroxide based and the refined ozone based test are depicted in Fig. 3.
illustration not visible in this excerpt
No general conclusion can be made, since all of the tested sterilization methods have their advantages and disadvantages.
The procedure most suitable for plated metals and steels was the hydrogen peroxide sterilization. After 50 cycles the tested samples were almost unchanged. No corrosion was visible, only some white stains were observed on some samples. The procedure, however, proved as problematic for the treatment of adhesives, since some of them got brittle and leaked out of some instruments.
The procedure most suitable for adhesives was the second ozone based sterilization. None of the adhesive got separated from the test plates and no adhesive leaked. Nevertheless, the adhesives were not unaffected, as some of them got dull and started to separate from the test plates. Generally spoken adhesives are problematic in sterilization processes. None of the tested procedure had totally satisfying results. Especially the adhesives no. 20 and 41 lost their power due to the sterilization.
None of the tested procedures had an influence on the light transmittance ability of endoscopes and light guide cables. The electrical safety was not affected either.
As already mentioned it is not possible to determine the best sterilization process for all instruments. Based on the used materials it needs to be assessed individually which procedure shall be used for sterilization.
Since most of the tests have not been completed, the final assessment of the results of the testing at 400 cycles is a possible subject for future works.
References
[1] L. Keir, B. Wise, C. Krebs, C. Kelley-Arney, Medical Assisting: Administrative and Clinical Competencies, 6th ed., Thomson Delmar Learning, 2007, p.520.
[2] DIN EN 285:2009-08, Sterilization - Steam sterilizers - Large sterilizers.
[3] L.A. Feldman, H.K Hui, Compatibility of Medical Devices and Materials with Low-Temperature Hydrogen Peroxide Gas Plasma, Med. Dev. Diag. Indust. 1997
[4] DIN EN ISO 17664:2004-07, Sterilization of medical de-vices - Information to be provided by the manufacturer for the processing of resterilizable medical devices
The figure shows that adhesives of the same type are affected in a similar way, relatively independent on the type of the sterilization procedure. A possible reason is that the tested sterilization types are similar. All of them use low temperature and gaseous or sputtered sterilization agents with oxidative properties. Another possible explanation for the result is the different chemical composition of the adhesives themselves.
Biomedical Engineering II
Electrical Impedance Tomography Image Reconstruction with EIDORS
Julia Henschel, Steffen Kaufmann, Aram Latif, Windy C. Saputra, Tanner Moray and Martin Ryschka
Abstract- Electrical Impedance Tomography (EIT) is a functional real-time imaging technique which obtains images of the impedance distribution of an object under test. It is based on the injection of small well known alternating currents (AC) into the object under test and the measurement of resulting voltages on the boundary surface of that object. The measured voltages in combination with the known AC are subsequently used to reconstruct the impedance distribution of the object under test. Due to the soft-field behavior in EIT, image reconstruction is a non-linear, heavily ill-posed and ill-conditioned inverse problem. This work describes methods and challenges in EIT to obtain measurement data and reconstruct spatial impedance distributions of an object under test. For that purpose three different EIT example applications are presented. These applications are: intracranial EIT, EIT for micro-vessel studies and EIT for irreversible electroporation (IRE) online feedback.
I. INTRODUCTION
Electrical Impedance Tomography is a noninvasive, functional imaging technique, based on the injection of small known alternating currents (AC) and the measurement of the resulting voltages on the boundary surface of the object under test. The measured voltages and the known AC are used for a subsequent image reconstruction.
EIT is real-time with frame rates up to 50 Frames per Second (FPS), has no known hazards to the patient and is compared to other imaging techniques low cost [2]-[4]. Furthermore typical EIT systems are very compact and suitable for mobile usage.
During the measuring process a current source and one or more voltmeters are rotated around the object under test to obtain all possible combinations of transfer impedances [1]. Fig. 1 illustrates a common EIT test arrangement.
Julia Henschel is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5400, e-mail:
henschel.julia@gmx.de).
Steffen Kaufmann is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5400, e-mail: Kaufmann@fh-luebeck.de).
Aram Latif is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5400, e-mail:
aram.latif@gmx.de).
Windy C. Saputra is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5400, e-mail: saputraw@gmx.de).
Tanner Moray is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5400, e-mail:
tannermoray@hotmail.de).
Martin Ryschka is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: + 49 451 300 5026, e-mail: ryschka@fh-luebeck.de).
illustration not visible in this excerpt
Fig. 1. Test arrangement for the EIT measuring process. The measurement currents are injected via a pairs of electrodes into the object under Test (Ω). The resulting voltages are subsequently measured between the remaining electrode pairs. After all voltages have been measured, the current source moves to the next pair of electrodes. This algorithm continues until all independent transfer impedance measurements are recorded.
This work describes methods and challenges in EIT to obtain measurement data and reconstruct the spatial impedance distribution of an object under test. For that purpose three different EIT example applications are presented.
The first step in the reconstruction is to calculate the expected boundary voltages of the object under test with help of the Finite Element Method (FEM). For that purpose a 3d FEM Model is created with the mesh generator NETGEN. Subsequently the reconstruction is done with EIDORS. Due to the fact that EIT image reconstruction is a non-linear, heavily ill-posed and ill-conditioned inverse problem, a regularization algorithm with a-priori information is used. Regularization counteracts the ill-posed behavior of the image reconstruction problem and support stability of the solution.
II. Applications
This work uses three different applications to illustrate methods and challenges in EIT. These applications are: intracranial EIT, EIT for micro-vessel studies and EIT for irreversible electroporation (IRE). All systems have the same centralized embedded EIT system in common. The developed EIT system employs 16 current electrodes and 16 voltages electrodes. The system is cascadable, works in real-time in a frequency range of 10 kHz to 250 kHz and is also able to measure amplitude and phase of the object under test. For data transmission and configuration of the system also a high speed USB link is available. Figure 2 shows a photograph of the developed embedded EIT system.
All example applications will be first tested with phantoms, afterwards ex-vivo and animal studies are planned. When these studies provide useful results first measurements with human patients can be scheduled.
illustration not visible in this excerpt
Fig.2. A photograph of the Central EIT V1.00 prototype. The prototype employs 16 voltage and 16 current electrode connections (see the right hand side of the PCB), works in a frequency range of 10 kHz to 250 kHz and is able to measure amplitude and phase of the object under test. The measurement data will be transmitted via a high speed USB link to a host PC for further data processing and the actual image reconstruction.
A. EIT for intracranial applications
Multi-frequency EIT for intracranial applications allows functional real-time monitoring and diagnosis of the brain activity, the brain pressure and possible bleedings. Besides it is conceivable to create detailed impedance maps of the brain to get further comprehension in brain build-up and functionality [12].
For these purposes two micro-electrode arrays are placed into or onto the brain, through a partial skull opening. Due to this application high contact impedances of the skull and the skin are bypassed, thus allowing much more accurate measurements. Each micro-electrode array is equipped with 16 electrodes and is connected via short cables to the centralized embedded EIT measurement system. The data acquisition system is small sized and can be placed in close vicinity to the patient. Therefore cables can be kept very short thus stray capacitances will be kept small. Fig. 3 shows the system architecture.
illustration not visible in this excerpt
Fig. 3. Block diagram of the intracranial EIT system. The brain is connected via a microelectrode-array to the embedded Field Programmable Gate Array (FPGA) based System on Chip (SoC). The SoC controls data acquisition and pre-processing of the measurement data in real-time and manages the data transmission to the host.
B. A Micro EIT system for vessel studies
The EIT system for micro vessel studies enables measurement and reconstruction of impedance distributions of the containments of a small vessel. The vessel is equipped with 16 equally spaced micro-electrodes. The vessel itself is made of a non conductive material and is filled with a conducting solution, which allows an electrical contact with the containments. The size of the electrodes is about 1 mm in diameter; the inner diameter of the vessel is about 10 mm. Fig. 4 shows a principle drawing of the developed micro-electrode vessel.
Disturbances
illustration not visible in this excerpt
Fig. 4. Principle drawing of the micro-electrode vessel. The vessel is filled with a conducting medium to enable contact to the different objects (disturbances) which should be measured, detected and reconstructed.
The micro EIT system can be used i.e. for tissue or organ characterization, but the application area is not limited to life science. It can also be used in industrial or chemical settings e.g. to analyze diffusion processes or for material characterization.
C. EIT for Irreversible Electroporation (IRE)
The third application is a feedback system for IRE. IRE is an ablation technique, based on the destruction of targeted tissues by a localized electric field, with amplitudes up to 1 kV/cm [7]. These electrical fields are applied directly via two stainless steel needles into the tissue and leading to cell death which changes the permeability of the tissue permanently [5][6]. EIT will be used as a real-time monitoring system to enable online feedback to the surgeon during IRE.
An actual electroporator device consists of a high voltage pulse generator, cables and application needles. Fig. 5 shows a principle drawing of the EIT system in connection with the electroporator device.
B. Forward problem
The conductivity distribution of the object under test is determined by boundary measurements. For that purpose small alternating currents are injected in the object under test and resulting voltages are measured. With the measured voltages and the known AC the impedance distribution is reconstructed. To solve this inverse Problem, the potential distribution inside the body has to be calculated for the given current injection pattern. This step is called forward problem, which will be calculated by use of the FEM. Fig. 6 shows the forward problem in principle, if the injected currents and the impedance distributions inside the object under test are known (or estimated) and the boundary voltages are calculated.
For the EIT measurement electrode plates will be placed around the tissue under surgery. A relays allows a disconnection of the system and the electroporator device to protect the EIT system against the electrical fields of the IRE.
The contact impedances are assumed to be very small, because of the direct tissue contact. Thus a two electrode configuration for the impedance measurement can be used instead of the common four-electrode configurations. The use of a two electrode configuration, employing measurements on current carrying electrodes, leads to more independent measurements and thus to image quality improvements.
III. Reconstruction Methods
A. Ill-posed nature
EIT and almost all tomography modalities have an ill-posed nature [8][9]. According to Hadamard a mathematical model of a physical problem is ill-posed, if one or more of the following criterions are violated [10]:
1) For all admissible data, a solution exists.
2) For all admissible data, the solution is unique.
3) The solution depends continuously on the data.
The existence of a solution can be assumed, because every object has an impedance distribution. But mostly the solution is not uniquely determined, because of the finite measurements in contrast to the number of unknown conductivities. The number of independent measurements has a big impact on the quality of the reconstruction. The more independent measurements are acquired, the better the reconstruction quality. Unfortunately the number of independent measurements is limited by electrode size and voltage levels. Due to the instable nature of the reconstruction problem, also measurement noise and a limited knowledge of the object shape can lead to distortions and artifacts in the reconstructed image.
illustration not visible in this excerpt
Fig. 6. Illustration of the forward problem. The task is to calculate the boundary voltages on the objects surface based on known injected currents and the estimated impedance distribution of the object.
The solution of the forward problem is the first step in the reconstruction process. There are two methods to solve the forward problem: the analytical method, which is limited due to the geometric properties of the object under test and a second more widely used method, the numerical method.
The numerical method can be solved via the usage of the FEM. FEM breaks a continuous body down into a finite number of interconnected elements. An example of a FEM model of a cylinder is given in Fig. 7.
Cylindrical FEM Model using NETGEN and EIDORS
illustration not visible in this excerpt
Fig. 7. FEM-Model of a 3d saline Tank, generated with NETGEN and EIDORS.
C. Inverse Problem
The inverse problem describes the process of the reconstruction of the impedance distribution with known currents and measured boundary voltages. Fig. 8 shows the principle workflow.
illustration not visible in this excerpt
Fig. 8. Illustration of the inverse problem. The task is to figure out the electrical impedance distribution of the object under test based on the injected current and the measured boundary voltages.
The inverse problem is based on the forward problem. The solution is obtained by the solution of the according optimization problem, which tries to minimize the error between the calculated boundary voltages and the measured boundary voltages. The first step is to calculate boundary voltages with an estimate of the impedance distribution via the forward problem. These voltages are compared with the measured voltages. As long as the norm of the residua is bigger than a certain threshold the impedances are adjusted.
Because of the ill-posed nature of EIT, regularization methods are needed to stabilize the optimization problem [9]. In this work Tikhonov regularization is used. Equation (1) shows the minimization problem with Tikhonov regularization in which the damped minimization problem'*’“ (“ J is given.
illustration not visible in this excerpt
The basic idea of this regularization is to use prior information about the image. Tikhonov regularization tries to fit the data but punishes possible solutions with large gradients in the impedance distribution. In this connection 8r. equates to the measured voltages, 3n describes the calculated voltages. The parameter α is the regularization factor, which controls the strength of the penalty term. A disadvantage of regularization is the caused resolution decrease through image smoothing.
IV. EIDORS/NETGEN
The implementation of the FEM, as well as the reconstruction is accomplished by a Mathworks Matlab based open source framework named EIDoRs (Electrical Impedance and Diffuse optical Tomography Reconstruction software) in combination with the FEM mesh generator NETGEN. EIDoRs provides reconstruction and display for EIT data and contains interfaces for different existing EIT systems [12]. Furthermore EIDoRs has an active community and represents a common standard for EIT research groups.
This works presents three different and very promising applications of EIT to illustrate methods and challenges. Besides the well known measurement of the impedance magnitude, also the phase will be measured and used for reconstruction. This approach promises higher resolution and better contrast of the reconstructed images compared to the state of the art. The actual EIT image reconstruction will be implemented with EIDORS and NETGEN and is 3d based, for a better physical modeling and an improved visual impression.
Thus EIT is a promising low-cost real-time monitoring technique with no known hazards. Furthermore EIT allow e.g. online feedback, tissue characterization and functional imaging.
Acknowledgements
This work is financed by the program for the Future- Economy out of the European Regional Development Fund (ERDF).
illustration not visible in this excerpt
References
[1] Chongqing, V., Multifrequente Impedanztomographie zur Darstellung der elektrischen Impedanzverteilung im menschlichen Thorax, Universität Stuttgart, PhD thesis, February 2000
[2] Dräger-Medical . PulmoVista500, technical data sheet, Dräger, 2011
[3] Holder, D. S.: Electrical Impedance Tomography of brain function, available online, visited on the 11/11/2011 http://www.ucl.ac.uk/silva/medphys/research/eit/pubs/brain_EIT_overvi ew.pdf
[4] Manwaring P. K. , Halter R.J. and Borsic Q., Hartov A.: A modified Electrode Configuration for Brain EIT, In: J. Physics Conference Series 224, 2010
[5] Granot Y, Rubinsky B (2007) Methods of optimization of electrical impedance tomography for imaging tissue electroporation. Physiol. Meas. 28(2007) Π35-1147
[6] Davalos R V, Mir L M, Rubinsky B Tissue Ablation with Irreversible Electroporation Annals of Biomedical Engineering, Vol. 33, No. 2, February 2005 pp. 223-231
[7] Granot Y, Ivorra I, Maor E, Rubinsky B (2009) in vivo imaging of irreversible electroporation by means of electrical impedance tomography phy. Med. Biol. 54 (2009) 4927-4943
[8] Burger M.; Brune C. Methoden der Biomedizinischen Bildgebung , Uni Münster, 2010
[9] Brühl,M.&Hanke-Bourgeois,M. Kann Mathematik der elektrischen Impedanztomographie zum Durchbruch verhelfen? Universität Mainz, 2001
[10] Holder, D. S. (editor): Electrical Impedance Tomography - Methods, History and Applications. first. Institute of Physics Publishing, 2005
[11] A. Adler and W. R. B. Lionheart. Uses and abuses of EIDORS: an extensible software base for EIT. Physiological Measurement, 27:S25- S42, 2006.
Hofmann, U. G.; Hertleit, F.; Knopp U.; Langer E.: On the design od intracranial multi-site microelectrodes for electro impedance tomography. In: Biomedizinische Technik (2000), Vol. 45, Nr. E1, S. 167-168
Development and implementation of a method for producing directional solidified, electrospun hybrid structures as nerve guidance channels
Christopher Janssen, Stephan Klein, Birgit Glasmacher, Soenke Wienecke, Tanmay Chakradeo
As an alternative to the use of autologous grafts nerve guidance channels (NGC) are developed. These three-dimensional constructs are intended to realize an orientated re-growing of the axon to the correct target during the nerve regeneration. Especially for injuries with big gabs, NGC could become an alternative. The aim of this research was to produce an NGC for bridging gabs of 10mm. The NGC consists of an inner chitosan- scaffold with a special pore structure. Through these pores the axon sprouts are meant to channel to the correct target. The scaffold is produced with directional solidification, whereby different pore morphologies were produced by changing the freezing rate. To realize an easy saturation between the two nerve stumps the scaffold gets an overlapping electrospun fiber cover. For this purpose a special two-step method was developed. Before spinning the scaffold two different procedures to neutralize the chitosan scaffolds were tested.
I. Introduction and background
Nearly a few hundred thousand people are suffering from serious injuries of the peripheral nerves [1]. To subdivide these injuries the classification of Sunderland is applied [2]. From one degree to the next degree always one more neural structure is damaged (Fig. 1). As soon as the continuity of the axon is interrupted (second degree) the regeneration process of the peripheral nerves (Waller Degeneration) starts.
During this process first the myelin sheath breaks down and macrophages decompose the leftovers. In this way the macrophages are supported by the Swan-Cells. These
C. Janssen Luebeck University of Applied Sciences and Institute for Multiphase Processes, Leibniz University Hannover (telephone: 0176/63171777, email: christopher-janssen@t-online.de). S. Klein Luebeck University of Applied Sciences (email: klein@fh-luebeck.de).
B. Glasmacher Institute for Multiphase Processes, Leibniz University Hannover (email: glasmacher@imp.uni-hannover.de).
S. Wienecke Institute for Multiphase Processes, Leibniz University Hannover (email: wienecke@imp.uni-hannover.de).
T. Chakradeo Institute for Multiphase Processes, Leibniz University Hannover (email: chakradeo@imp.uni-hannover.de).
processes and the transmitted mitogens have a stimulating influence on the Swan-Cells (SC), so that the SCs differentiate and proliferate in the defective area [1], [3]. After 12 days the SCs still proliferate and begin to place tubularly. This tubular structure is called bands of Büngner and conduces as a natural guidance channel for the axon sprouts during the nerve regeneration [1], [3]. One of the biggest problems that interrupt the regeneration process is a too big gap between the nerve stumps. In this case an orientated growth of the axon is not possible. As a consequence neuromas can develop. Furthermore scar tissue can grow in the injured area [1] [4]. For these (two) reasons it is necessary to repair the nerve surgically [4]. Therefor the gold standard method is the use of autologous grafts, which are harvested for example at the lower leg. The advantage of this is the existence of the SCs and the basallamina, which produce neurothrophes factors [1], [5]. But there are also several disadvantages. So the harvest and the transplantation require two surgical interventions which result in a substance loss at the sampling location and a loss in function. Furthermore neuromas can grow in that area and only 50 % of the grafts are useful [1], [4], [5]. Lots of efforts have been made to find a functional alternative (e.g. nerve guidance channels)
II. NERVE guidance channels
NGCs are meant to support nerve regeneration, especially for injuries with big distances that must be bridge [1], [6].
Using NGCs creates conditions which are similar to the natural nerve regeneration of less serious injuries [1]. One positive aspect of NGCs is that problems like availability or negative immune responses can be avoided. In general NGCs are tubular to emulate the basallamina and to make an easy saturation possible (Fig. 2). In this way an orientated growth to the proximal end should be realized [1], [6]. Furthermore the tension of the connection is reduced and the tubular form supports also an accumulation of neurotrope and neurotrophe factors. Another important fact is that NGCs isolate the regeneration process from the wound healing process so that scar tissue does not interrupt the nerve regeneration [1], [5].
illustration not visible in this excerpt
Fig. 2: Principle of a NGC [17]
Because of the importance of the Swan-Cells for the nerve regeneration the scaffold should have a 3D-structure which is capable for cell colonization [1]. Apart from other facts like the material or a good porosity an ideal NGC is characterized by a few more features (Fig. 3) [1], [6].
III. Material and Methods
The developed nerve guidance channel of this research consists of an inner 3D-structure, which is the functional scaffold and an outer fiber cover which is meant to support a saturation between the two nerve stumps. The Scaffolds material is made of chitosan, while the fiber cover is made of polycaprolactone. Chitosan, a derivative of chitin, is a biopolymer, which is characterized by its good biocompatibility, biodegradability and anti-bacterial qualities. It cannot be solved in any organic and neutral dilutions. It is only soluble in acid dilutions with a pH-value < 6 [7], [8], [10]. Apart from that it could be shown, that Chitosan can well be used for a colonization and proliferation of SCs. For this reason it might be very well-suited for neural applications [5].
To produce the scaffolds with its defined pore structure, the directional solidification (Power-Down) with a subsequent freeze-drying was used. By this process it is possible to solidify a solution with directional ice crystal morphology [10], [11]. The directional solidification is characterized by a constant local and temporal temperature difference [10]. So the sample is frozen from one side and the solidification grows through the sample. The pure water of the solution solidifies at first whereby dendritical ice-crystals are formed (Fig. 4, A). The dissolved substances concentrate between the ice-crystals.
illustration not visible in this excerpt
Fig. 4: Principles of freeze-drying [11]
Then the ice crystals are removed by freeze-drying whereby cavities occur (Fig. 4, B) [10], [11], [12]. These cavities are the pores of the scaffold. The used solution consists of chitosan and ascetic acid (1 %) at the rate of 15:1.
After freezing the scaffolds still consist of chitosan acetate. A contact with water (e.g. air humidity) could result in a dissolving of the samples because of a back-formation of acetic acid. Thus the pH value will be decreased greatly and the scaffolds will dissolve. For this reason it is necessary to neutralize the scaffold by removing the remaining acetate. During this research two different procedures were tested. In the first method the scaffold is gradually brought into lower concentrated ethanol, placed in vacuum for a specific time and subsequently placed on a laboratory shaker to be sure that the whole scaffold is filled with ethanol. Finally the scaffolds are freeze-dried again. As the first method was not successful, a different method was used. Instead of bi-distilled water in method 1 a phosphate buffered salt solution (PBS) was applied. PBS is especially characterized by its very good stability of the pH-value. Compared to the first method, the scaffold is not placed into 100 % ethanol until the final stage.
The fiber cover is made of polycaprolactone which is one of the first synthesized polymers and is generated by a ringopenning polymerization of ε-caprolactone. It is a hydrophobic, semi-crystalline polymer, with a good biocompatibility and adaptable degradation kinetics [4], [9]. The fiber cover is produced by electrospinning. This process enabled the production of fibers at nano scale [14]. The basic principle is based on an electrical field between a grounded collector and a syringe or pipette with a metallic needle, filled with a polymer solution. If the polymer solution drops into the electrical field, electrical field forces lead to a deformation of the drop. The drop of the polymer solution becomes conical and is called Taylor-Cone. As the voltage is high enough, the viscosity and surface tension are exceeded and a fluid jet comes out of the drop and is accelerated towards the collector. During this process the solvent evaporates and consequently one fiber is laid onto the collector [13], [14], [15].
To produce the scaffold with the directional solidification it is necessary to design a form for the solution by which also the form of the scaffolds is fixed. For this reason the final form has tubular bores with a diameter of 2 mm. Due to boundary effects it is further necessary to keep tolerances at the top and bottom sides in mind. To cut the scaffold to the correct length a three-layer form was developed. After the freezing process the top and bottom layer can be removed by which the scaffold is fitted of the correct length (10 mm). The centerpiece consists of copper to realize an ideal orientation of the isotherms, which is highly important for a correct growth of the ice crystals. To protect this procedure from external temperature influences the form is isolated by a cover of Teflon.
After freezing it appeared that both procedures to neutralize the scaffold did not work. For this reason the scaffolds were spun without neutralizing. The developed method is based on a two-step principle. The scaffold is firstly placed on a support between two pins and is spun over a defined period of time (Fig. 6).
illustration not visible in this excerpt
Fig. 6: Supported scaffold
Then the support is removed and a 90° rotation is realized whereby the next side is spun. This is repeated until all four sides have been spun. The pins are also used to realize the overlapping ends of the fiber cover. Then (second step) this construct is spun during constant rotation over a defined period of time. Therefor a rotation apparatus was developed and built. As the chitosan scaffold has a very low solidity it cannot be fixed axially. At the end the complete NGC can be removed (Fig. 7).
illustration not visible in this excerpt
Fig. 7: NGC after spinning
V. Experiments A. Directional solidification
The chitosan scaffolds were produced with a variation of the freezing rates to get different pore structures. The freezing rate was changed from 1 to 10 K/min, while the temperature difference, the initial and final temperature remained constant. By such a variation it is possible to produce different pore morphologies. At small freezing rates the pores are placed irregularly and have different dimensions. By rising up the freezing rate, the pores get a longer form and begin to point inwards (Fig. 8). Furthermore an open structure at the lateral face which is probably caused by outer boundary effects can be found.
illustration not visible in this excerpt
Figure 8: Pore structure of a scaffold (A: 1 K/min, B: 7 K/min)
The reason for the form as well as for the inordinate orientation of the pores is the coincidence anisotropy of the ice crystals at the beginning of the solidification process [10]. Regarding the width of the pores, it is noticeable that these become smaller by rising the freezing rate. At a freezing rate of 1 K/min the average pore width is 68 pm, by a rate of 10 K/min only 31 pm. The complete development can be found in Fig. 9.
illustration not visible in this excerpt
The surface's structure of the scaffolds shows a clear axial orientation of the pores, which is a further indication for a directional growth and could mean that a similar inner pore structure exists (Fig. 10).
illustration not visible in this excerpt
Fig. 10: surface's structure of the scaffold (B = 3 K/min)
Furthermore it was recognized that after removing the top and bottom layer of the form, a different pore structure exists. It is unclear why 3 mm are not enough to get a wide open pore structure.
B. Neutralization
As mentioned in the introduction both methods were not successful. After neutralization with ethanol and bi-distillated water the structure of the scaffold is deformed and becomes useless. Furthermore it was recognized that the rate of the deformation is regulated by the pore structure. Scaffolds, produced at a high freezing rate are damaged more seriously.
illustration not visible in this excerpt
Fig. 11: Neutralized Scaffold (first method); A: 1 K/min, B: 10 K/min
The consequences of the second method were much more serious than those of the first one. The structure was dissolved completely (Fig. 11). Another negative result of both procedures is that the scaffolds lost their sizes (Fig. 12).
illustration not visible in this excerpt
Fig. 12: Sizes of the scaffold after neutralization C. Electrospinning process
After finding the ideal process parameter, various wall thicknesses were spun. To analyze the quality of the fiber cover, it was cut before and behind the scaffold. These crosssections were finally measured at 8 points and the average size of each side was calculated. By comparing the values the symmetry could be analyzed.
The rotation time was changed from 1 to 10 minutes. In the range of 5 minutes the best result was achieved (Fig. 13). The average difference between the left and the right-hand sides of the cover is in this case 7, 02 pm, while the average wall thickness after 5 minutes spinning was 214 pm.
illustration not visible in this excerpt
Fig. 13: Fiber cover (5 minutes)
If the spinning time is shorter, the wall becomes thinner and is not reproducible. A spinning time of more than 5 minutes results in a thicker wall with an acceptable symmetry.
Fig. 14 shows the increase of the wall thickness in relation to the spinning time.[5]
one positive fact is that a good saturation during the first tests was possible. Further connective tissue had grown at the connection [17]. The production of the Chitosan-scaffolds with the Power-Down freezing-device and the developed form is realized well as the evaluation shows. Especially the three- layer system seems to be a good choice. However, it is still not clear why removing 3 mm at the bottom side is not enough. So, more experiments will be necessary. Because of the significance of the pore size for successful nerve regeneration, more parameter studies are needed. In this context all parameters of the directional solidification should be changed to examine their influence on the pore morphology.
The biggest difficulty during this research was to neutralize the scaffold after freezing. Because both procedures were not successful, an alternative method must be found. Perhaps it is necessary and helpful to crosslink the scaffolds before. The biggest problem during development of the rotation apparatus was the low solidity of the chitosan scaffold, which had a big influence on the developed clamping technique. As the experimental part shows a useful method has been achieved.
To improve the developed two-step method, the process should be automated, because some faults resulted from inaccuracies through the manual proceedings. These improvements are also especially important because of the low stability of the electrospinning process.
VII. References
[1] Bruns, S., “Screening neuer Gerüstmaterialien und ihre Modifikation für den Einsatz im Tissue Engineering peripherer Nerven“, 2007
[2] Woischneck, D.; Bardenhauer, M.; Antoniadis G., “Operative Nervenläsionen bei othopädischen und unfallchirurgischen Eingriffen“ In: Orthopädie und Unfallchirurgie, 2009, pp. 181 - 192
[3] : Marco Mumenthaler et. al., “Läsionen peripherer Nerven und
radikulärer Syndrome“, 8. Aufl.: Thieme, 2003
[4] : Deumens, R et. al., “Repairing injured peripheral nerves: Bridging
the gap”, Elsevier Ltd., In: Progress in Neurobiology, 2010
[5] : Ciardelli, G.; Chiono, V., “Materials for Peripheral Nerve Regenera
tion” Wiley - VCH Verlag, In: Macromol. Biosci, 2005, pp. 13 - 26
[6] : Huang, Y.; Huang, Y., “Tissue Engineering for Nerve Repair”, In: Biomedical Engineering- Applications; Basis & Communications, 2006, pp 100 - 110
[7] : Ondruschka, J.; Trutnau; M.; Bley, T.; “Gewinnung und Potentiale des Biopolymers Chitosan:”, Wiley - VCH Verlag, In: Chemie Ingenieur Technik, Biopolymere, 2008, pp. 811 - 820
[8] : Pillai, C. K. S.; Paul, W.; Sharma, C. P., “Chitin and chitosan polymers: Chemistry, solubility and fiber formation”, Elsevier, In: Progress in Polymer Science, 2009, pp. 641 - 678
[9] : Woodruff, M. A., Hutmacher, D. W., “The return of a forgotten polymer - Polycaprolactone in the 21st century”, In: Progress in Polymer Science, 2010, pp. 1217 - 1256
[10] : Heß, U., “Erstellung und Chrakterisierung 3-dimensionaler Chitosan- Hydroxylapatit-Strukturen für das Knochen-Tissue Engineering”, 2010
[11] : Schoof, H., “Verfahren zur Herstellung gefriergetrockneter Kollagenschwämme mit definierter Porenstruktur”, Dissertation, RWTH Aachen, 2000
[12] : Kuberka, M., “Gezielte dreidimensionale Zellkultivierung auf strukturierten lyophilisierten Kollagenträgern”, RWTH Aachen, 2007
[13] : Reis, R. L. et. al., “Natural-based polymers for biomedical applictions; Woodhead Publishing”, 2008
[14] : Ramakrishna, S.et. al., “An Introduction to Electrospinning and Nanofibers” World Scientific, 2005
[15] Gualandi C., “Porous Polymetric Bioresorbable Scaffolds for Tissue Engineering”, Springer, 2011
[16] Green, W., “Netter's Orthopaedics”, Saunders, 2006
[17] Institute of Neuroanatomy, Hannover Medical School
Spectral light modulation using a digital micromirror device (DMD) for the calibration of pulse oximetry sensors
St. Marx, B. Weber, B. Nestler, H. Gehring
Abstract-Optical sensors for patient monitoring (e.g. pulse oximeters) exploit the interaction of light and the investigated tissue. Since these interactions cannot fully be described theoretically calibration with volunteer-studies is required [1], being expensive and time-consuming and not applicable for legal metrological controls.
A calibration concept for optical sensors is described, based on detection of the sensor’s optical signals, processing according to previously recorded measurements and re-emission to the sensor to be calibrated. Core part of this concept is a light source with controllable spectrum. It is composed of a white light source and a diffraction grating separating the light into its spectral components. The spectrum is imaged onto a micromirror device, enabling to control the amount of reflection depending on the optical wavelength. This allows the selection of desired wavelengths for re-emission.
Spectral light modulation and selection in the wavelength range from 550 nm to 950 nm was achieved successfully.
I. Introduction
Pulse oximetry is one of the standard monitoring systems in emergency and intensive care units and during surgery. It delivers information about the patient’s oxygen saturation. Until today these sensors can only be calibrated by experimental desaturation studies with volunteers. In the course of a controlled desaturation study, the inspiratory oxygen content is decreased to reduce the arterial oxygen saturation from nearly 100% to 75%. Parallel the raw signal from the sensor is recorded and arterial blood samples are drawn. The oxygen saturation from the blood samples is assessed by laboratory blood gas analysis (CO-oximetry) as reference. Finally the relationship between the sensor’s raw signal and the reference values is statistically analyzed and a calibration curve is generated.
Obviously such studies are very complex and not suitable for legal metrolo gical controls, as they are required for other medical devices measuring physiological values (e.g. non invasive blood pressure, body temperature) [2].
During the late 1990’s a pulse oximeter calibrator prototype was developed in an EU project by the University of Luebeck
St. Marx (tel: +49(0)451 140 3691, e-mail: stefan.marx.83@gmail.com), B. Nestler (tel: +49 (0)451 300 5524, e-mail: nestler@fh-luebeck.de) and B. Weber (tel: +49(0)451 300 5520, e-mail: weber@fh-luebeck.de) are with Luebeck LTniversity of Applied Sciences, Medical Sensors and Devices Laboratory.
H. Gehring (tel: +49 (0)451 500 6200, e-mail: gehring@uni-luebeck.de) is with the Department of Anesthesiology, University Medical Center of Schleswig-Holstein, Campus Luebeck.
in cooperation with the ISO and FDA as well as multiple medical centers across Europe and pulse oximeter manufacturers. The approach was based on the spectral modulation of the pulse oximeter’s optical signals using a digital micromirror device (DMD) and prior spectral decomposition with a diffraction grating [3, 4]. The project lead to satisfying results, yet limited to a minority of existing pulse oximeters due to technological barriers. Because of great technological innovations in this field (e.g. DMD performance improvements, developments of comfortable development environment, enhancement of grating efficiencies), a new approach was now started to develop a pulse oximeter calibrator.
To calibrate the sensors it is proposed to detect and analyze the pulse oximetry probes light signals spectrally and temporally. These signals are processed using the transmission spectra (which are recorded previously during desaturation studies with volunteers) to calculate the light signals which would be emitted from the patient’s tissue after the transmission of the original light signals. Finally a broadband white light source is used to generate these light signals physically, by the spectral modulation using the DMD which would be re-emitted to the pulse oximetry probe under calibration. The basic principle of this spectral light modulation will be described in the following section followed by first results of the system characterization.
II. Material and Methods
A. Digital micromirror device (DMD)
The core part of the spectral modulation setup is the DMD (ViALUX GmbH, Germany), an array of tiltable micromirrors. Fig. 1 shows a photograph and a microscope close-up of the DMD, Table 1 summarizes the specifications.
illustration not visible in this excerpt
Fig. 1. Photograph of the DMD-array with partially tilted mirrors (arrow) and microscope close-up of single mirror elements
Each mirror can be tilted bistable ±12° along its diagonal axis giving the possibility to control the direction of the reflected light into two different directions (ON and OFF). In Fig. 1 a small portion of the mirrors is tilted into another direction than the rest, which can be seen as a small dark bar indicated by the arrow.
A control software (screenshot see Fig. 2) was programmed in C# using Microsoft Visual Studio 2010. This custom software enabled to control the reflections of different DMD areas to conduct a basic characterization of the system.
B. Opto-mechanical setup
The opto-mechanical setup (see Fig. 3) is based on the spectral decomposition of white broadband light emitted by a conventional 10 W tungsten halogen lamp (AvaLightHAL, Avantes BV, Netherlands). The light is guided by a fiber to a lens imaging the fiber end onto an adjustable vertical slit (M- SV-0.5, Newport Spectra-Physics GmbH, Germany). An iris is placed directly in front of the first lens to adjust the light intensity.
A second lens behind the slit images the slit into the image plane. A transmittive diffraction grating (600 l/mm, 900 nm, Wasatch Photonics, USA) is located between lens and image plane, decomposing the incident light into its spectral components. The DMD is positioned in the image plane of the first diffraction order, so a certain DMD area is covered by a corresponding spectral range (e.g green on the left, red in the center and infrared on the right).
The light reflected into the ON-direction is collected by a third lens and a second fiber which is connected to a VIS/NIR spectrograph (Shamrock SR303i with DU420-BR-DD CCD camera, Andor Technology, USA) to analyze the modulated spectrum.
C. System characterization
After assembling the opto-mechanical setup, the performance was characterized regarding the following aspects:
- Wavelength range and selection,
- Intensity attenuation.
The spectral range of the whole setup was directly measured with the spectrograph when all DMD mirrors were turned ON. Then the DMD was partially turned OFF, e.g. only the left or the right half or only one single column (see Fig. 4) to investigate the possibility of selecting only certain spectral ranges.
The requirements for the achieved spectral range can be deduced from the emission spectra of commercial pulse oximeter probes. Measurements of different probe models from various manufactures showed that especially the emission in the infrared varied from 885 nm to 935 nm [5]. This circumstance emphasizes the need of a light source which can emit a flexible light spectrum.
Student Conference on Medical Engineering Science 2012
illustration not visible in this excerpt
Fig. 4. DMD patterns to investigate the possibility of wavelength selection. A: All mirrors on the left side are turned OFF, only the right half of the DMD contributes to the reflection. B: Only one single column in the DMD center is turned ON, hence reflecting only a small part of the total spectrum.
For the calibration of optical sensors it is not only necessary to emit the correct spectrum but the intensity of the emitted
light has to be modulated as well. To investigate the possibility of the DMD to attenuate the reflected light intensity the complete DMD was turned into ON-mode. Then the lines were turned to OFF one after the other, always from top and bottom at the same time. This scheme is shown graphically in Fig. 5.
illustration not visible in this excerpt
Fig. 5. DMD patterns to attenuate the intensity of the reflected light. The mirrors are consecutively turned from ON to OFF, starting from top and bottom (A), thus slowly narrowing the reflecting DMD area (B)
III. Results and Discussion A. Wavelength range and selection
Fig. 6 depicts the detected spectrum for the following cases (see also Fig. 4):
- White light source directly measured, i.e. source directly in front of the spectrograph,
- Full DMD set ON,
- Only left half set ON,
- Only right half set ON,
- Only one single column set ON.
The spectrum covered by the whole DMD ranges from 550 nm to 950 nm. As one half of the DMD is turned ON only the corresponding part of the spectrum is detected. For just one single column turned ON only a small peak is detected.
illustration not visible in this excerpt
Fig. 6. Reflected spectrum at different amounts and positions of DMD columns. The directly acquired (i.e. source directly in front of the spectrograph) spectrum of the white light source is shown on top. The columns turned ON are given top right at each graph. It is possible to select only small portions from the total spectrum.
The full DMD spectrum exhibits a dip in the center and rather flat parts to the sides. Due to the very small mirror dimensions (see table 1) interferences and maybe diffraction effects might be one reason for this shape. Yet the interpretation of the spectrum shape is not easy since many other factors have to be considered, i.e. spectrum of tungsten halogen lamp (shown top in Fig. 6), transmission characteristics of all optical elements, grating efficiency. Additionally it has to be pointed out that the spectrograph used for analysis is not calibrated in terms of intensity distribution over wavelength. Thus the spectral transfer function of this system has to be considered as well.
The narrowest line width achieved by the test setup (only one single column ON) of approximately 25 nm (Full width at half maximum, FWHM) is to broad to simulate the emission of a conventional pulse oximeter probe LED which itself has already a FWHM of ca. 25 nm. To improve the smallest achievable line width, the optical setup has to be optimized, i.e. imaging quality of the slit to the DMD plane.
B. Intensity attenuation
The relationship between reflected light intensity and amount of DMD lines turned ON is shown in Fig. 7. The original measurements contained the intensity of the complete spectrum, the graph shows only the signal intensity for one wavelength (760 nm) as an example. All other measurements followed the same graph shape.
Amount of reflecting DMD lines
Three sections can be distinguished in Fig. 7. In section A the intensity does not change with decreasing amount of reflecting DMD lines. In section B the light intensity decreases with decreasing number of reflecting DMD lines with a slope of (0.28 ±0.07) %/line while in the third section C the light intensity decreases as well but with a steeper slope of (0.71 ±0.05) %/line.
This effect can be explained with a not exactly centered image of the spectrum on the DMD surface and a spectrum that does not illuminate the whole DMD area as shown in
Fig. 8.
illustration not visible in this excerpt
decreasing amount of reflecting lines, the intensity decreases. Only in section C the spectrum is affected from both sides resulting in a steeper decrease of intensity with decreasing number of reflecting lines.
This shows clearly, that an exact alignment of the spectrum along the DMD axis is important. Also it is recommended to illuminate the complete DMD area in order to maximize the dynamic range for the intensity control.
IV. Conclusions
This work shows that is possible to perform spectral light modulation with a DMD combined with a simple setup for spectral decomposition of a broadband light source. The results so far are not directly applicable for the calibration of pulse oximeters regarding especially the efficiency of spectral selection.
The opto-mechanical setup has to be improved to optimize the imaging characteristics and to achieve a better illumination of the DMD area. Additionally the DMD control software has to be developed aiming not only to display static patterns but to display dynamic patterns which is inevitable for the calibration of pulse oximeters.
References
[1] International Organization for Standardization, ISO 80601-2-61:2011 “Medical electrical equipment - Part 2-61: Particular requirements for basic safety and essential performance of pulse oximeter equipment”, 2011
[2] German Federal Ministry of Justice, „Medizinprodukte
Betreiberverordnung - MPBetreibV, §11, Abs. 1 & Anlage 2, 2009
[3] Ph. Knoop, “Investigation of a novel method for the calibration of pulse oximeters”, Dissertation Universität Lübeck, Tectum Verlag, Marburg, 1999
[4] H. Matz, “Die Möglichkeit der in-vitro Kalibration und Validierung von Pulsoximetern mit Hilfe von zeitaufgelösten Transmissionsspektren”, Dissertation Universität Lübeck, Shaker Verlag, 2004
[5] B. Weber, B. Nestler, J.-F. Zhang, H. Gehring, „Calibration of optical sensors for patient monitoring - A novel concept“, Biomedizinische Technik/Biomedical Engineering, Volume 56, Issue s1, Pages 1-22, 2011
Fig. 8. A spectrum that does not illuminate the whole DMD area and additionally is not centered along the vertical DMD axis leads to different intensity attenuation results (see Fig. 7). A: No change of intensity; B: Partial change of intensity; C: Maximum influence on intensity
In section A the spectrum is not affected by the decreasing number of reflecting lines. In section B only one side of the spectrum (in Fig. 8 the lower part) is affected by the
Multi-frequency Electrical Impedance Tomography for irreversible Electroporation
Windy C. Saputra, Steffen Kaufmann, Tanner Moray, Aram Latif, Julia Henschel and Martin Ryschka
Abstract— Electrical Impedance Tomography (EIT) is a functional real-time imaging technique, mainly used in medical applications. EIT is based on impedance measurements of an object under test and the reconstruction of its spatial impedance distribution. EIT is painless, low-cost, works without ionic radiation and has no known hazards.
Irreversible Electroporation (IRE) is an ablation technique, based on the permanent permeability change of the cell plasma membrane due to an applied electric field.
This work presents a novel EIT system designed for measurement, visualization and feedback of IRE. The developed system employs two electrode plates surrounding the target tissue for data acquisition. It enables impedance spectroscopic measurements in real-time, is modular arranged and based on a Field Programmable Gate Array (FPGA) System on Chip (SoC) that is connected to host PC. The embedded system is in charge for data acquisition, preprocessing and transmission, whereby the host PC provides further filtering, data conditioning and the subsequent image reconstruction and display.
I. Introduction
Electrical Impedance Tomography (EIT) is a medical imaging technique, based on measurement and reconstruction of spatial impedance distributions within an object under test. For the measurements small known alternating currents (AC) are injected via electrodes into the object under test. The resulting boundary voltages are subsequently measured. Based on these measured voltages and the known currents an image of the spatial impedance distribution can be reconstructed. EIT has no known hazards and allows real-time measurements without ionic radiation [1] [6] [8]. Beside the usage in medical settings EIT is also used in industrial applications e.g. for pipe and tank monitoring [13-14] and in in large scale applications in geophysics, for example in soil composition [12] and geological layer analysis [11].
Electroporation is the significant increase of the electrical impedance of cell plasma membranes, caused by external applied high voltage pulses. Depending on the magnitude and duration of the applied voltage pulse the effect of electroporation can be a temporarily called reversible electroporation or permanently called irreversible electroporation (IRE) [1] [5].
Discovered in the late 1960s electroporation is now used for gene therapy [2] and drug therapy in the reversible mode [3] and for ablation of tissue in the irreversible mode [4]. For IRE electric fields up to 1000 V/cm are used, which are causing a subsequent cell death [7]. Currently amplitude and duration of the IRE are solely based on the experience of the surgeon, caused by a lag of suitable real-time feedback systems [7].
This work presents a novel EIT system which is designed to provide functional real-time impedance spectroscopic images of the tissue under electroporation to enable online feedback to the surgeon.
The presented system is based on a Field Programmable Gate Array (FPGA) System on Chip (SoC). FPGAs are powerful electronic logic devices, which can be individual programmed by use of a Hardware Description Language (HDL). In difference to a microcontroller, a FPGA can process tasks in parallel and allow therefore a much more powerful task handling.
II. System architecture
To obtain a good image resolution and a large image area, 64 electrodes will be used. For that purpose two 32 electrode EIT systems will be linked together. Figure 1 shows the principle build-up.
Windy C. Saputra is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5400, e-mail: saputraw@ gmx. de)
Steffen Kaufmann is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5400, e-mail: steffen.kaufmann@fh-luebeck.de)
Tanner Moray is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5400, e-mail: tannermo- ray@hotmail.de)
Aram Latif is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5400, e-mail: aram. latif@ gmx. de)
Julia Henschel is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5400, e-mail: juliahenschel@web.de)
Martin Ryschka is with Laboratory for Medical Electronics, Lübeck University of Applied Sciences (telephone: +49 451 300 5026, e-mail: ryschka@fh-luebeck.de)
illustration not visible in this excerpt
Fig 1. Principle system architecture of the developed system. The object under test is surrounded by two electrode plates. The electrode plates are connected to two EIT systems to allow a higher number of electrodes for better image resolution.
The EIT systems are organized in a master slave paradigm for measurement synchronization. The developed solution will be build around an existing commercial elec-
The electroporator consists of a high voltage pulse generator, cables and two small stainless steel needles. The needles have a diameter of about 400 ^m and will be stabbed directly into the tissue during the operation. The electrode plates of the developed EIT system will be placed around the tissue under surgery. To protect the system against high voltage pulses (electronic) relays in series with the plates allow a disconnection of the electrode plates in case of the detection of an active electroporation. For that purpose the cables of the electroporator are monitored permanently via voltage probes. The actual embedded EIT system consists mainly of a control unit, a voltmeter and a current source as well as a multiplexer to connect the blocks to the different electrodes. The embedded System is connected via USB to a host PC for data transmission and control. For patient safety issues an appropriate galvanic isolation is also established in the USB connection. The host PC implements further signal conditioning and filtering, as well as the actual image reconstruction. [6]
illustration not visible in this excerpt
Fig 3. Electrode array setup on the plates - in grey voltage electrodes, in black current electrodes.
In contrary to other EIT applications, in electroporation the electrodes can be placed directly on the tissue, omitting the high contact impedance of the human skin. Therefore it can be assumed that the use of the simpler two-electrode method is trouble-free, as long as the electrodes are used in an appropriate frequency range and operated with a low current density.
To allow measurements at different scales, two different sets of electrode plates have been manufactured - one set with 5 mm spacing and one set with 10 mm spacing. Figure 4 shows a photograph of a prototyped electrode plate.
illustration not visible in this excerpt
Fig 4. Manufactured electrode plate PCBs with connectors (right) and electrodes (left). The electrodes are aligned on a 5 mm grid.
IV. Hardware
The embedded system is based on a SoC FPGA architecture. The FPGA (LFXP2 from Lattice Semiconductor) governs data acquisition, data pre-processing and data transmission in real-time. It is easy programmable and reconfigurable to adapt to varying conditions as common in research settings. Figure 5 shows a detailed block diagram of the EIT system.
The generation of the required AC is realized with a Digital to Analog Converter (DAC, LTC1668 from Linear Technology) with 16 bit resolution and up to 50 MSPS in combination with an interpolation filter and a voltage to current source. The DAC is able to generate arbitrary waveforms by usage of Direct Digital Synthesis (DDS) techniques. The excitation frequency is adjustable in a range from 10 kHz to 250 kHz. Beside conventional sinusoidal signals also linear chirps or any other signal overlay can by generated in order to minimizing the required data acquisition time for broadband impedance measurements [9]. To maintain an optimal current density at the electrodes a Programmable Gain Amplifier (PGA, AD8250 from Analog Devices) is used. The current can be adjusted in four steps in a range of 500 _μΑ to 5 mA according to the IEC 606011.
As shown in Figure 5 a multiplexer is used to connect the current source to the electrode plates. Because of the phase shift and attenuation of the multiplexer the output current is falsified. To increase the measurement accuracy an additional low side current shunt is employed to measure the actual excitation current.
The voltage measurements are realized using a two channel 40 MSPS (mega samples per second) Analog to Digital Converter (ADC, LTC2297 from Linear Technology) with a resolution of 14 bit in combination with band pass filters and PGAs. The PGAs have four different amplification factors (1, 2, 5 and 10) to allow an optimal match to the ADC input voltage range and therefore to an optimal Signal to Noise Ratio (SNR). To maintain a high SNR band pass filters will be used to attenuate noise and interferences outside the interesting frequency range. Because of the known limitations of analog filters, inside the FPGA an additional Finite Impulse Response (FIR) filter is implemented to enable high filter orders. Contingent by the high sample frequency of the ADCs multi-rate signal processing in terms of oversampling and decimation is possible. This multi-rate approach enables an elegant signal processing and reduction of the measurement data.
The actual signal demodulation is done digitally with a Fast Fourier Transformation (FFT) based approach, which allows amplitude and phase measurements over a broad frequency range.
Figure 6 shows a photograph of the manufactured prototype of embedded EIT system PCB.
illustration not visible in this excerpt
Fig 6. Assembled prototype of the developed embedded EIT system PCB.
The data transmission between PC and embedded system is realized with a powerful USB 2.0 interface in high speed mode, capable of data rates up to 40 MByte/s. The claimed galvanic isolation is achieved with an optically coupled USB hub (USB 2 Isolator STD 2224 from Baaske Medical).
The implemented firmware is designed modularly, while data acquisition and preprocessing are implemented directly on the FPGA. The actual image reconstruction is done on the host PC with Mathworks MATLAB in connection with EIDORS [10]. This modular build-up allows maximum flexibility in terms of exchangeability and extensibility to provide ideal conditions for research purposes. Figure 7 show an overview of the FPGA Firmware.
illustration not visible in this excerpt
ate arbitrary waveforms and measures transfer impedances in a frequency range of 10 kHz to 250 kHz. The system promises high data quality, furthermore it can be expected that data collection will contribute to a better understanding of the electroporation and will provide a real-time feedback for electroporation.
Before first in vivo tests can be done the system has to be fully operational in term of software development and further hardware tests. Afterwards first phantom tests and ex vivo tests have to be carried out.
By now as a future improvement a new electrode setup is planned to allow real 3d measurements of object under test. For that purpose up to four embedded EIT systems will be interconnected to increase the amount of electrodes.
Acknowledgment
This work is financed by the program for the Future- Economy out of the European Regional Development Fund (ERDF).
zuKuiMFTsprogramm
Wirtschaft
Investition in Ihre Zukunft
financed by the European Union,
European Regional Development Fund (ERDF)
References
1. Granot Y, Rubinsky B (2007) Methods of optimization of electrical impedance tomography for imaging tissue electroporation. Physiol. Meas. 28(2007) 1135-1147
2. Gehl J (2003) Electroporation: theory and methods, perspectives for drug delivery, gene therapy and research Acta Physiol. Scand. 177 434-47
3. Mir L M, Belehradek M, Domenge C, Luboniski B, Orlowski S, Belehradek J, Schwaab B, Luboniski B and Paoletti C (1991) Electrochemotherapy, a novel antitumor treatment: first clinical trial c.R. Acad. Sci. III 313 613-8
4. Rubinsky B, Onik G, Mkus P (2007) Irreversible electroporation: new ablation modality-clinical implications Technol. Cancer Res. Treat. 6 p. 37-48
5. Davalos R V, Mir L M, Rubinsky B Tissue Ablation with Irreversible Electroporation Annals of Biomedical Engineering, Vol. 33, No. 2, February 2005 pp. 223-231
6. Holder, D S (editor): Electrical Impedance Tomography. Methods, History and Applications. Institute of Physics Publishing, 2005
7. Granot Y, Ivorra I, Maor E, Rubinsky B (2009) in vivo imaging of irreversible electroporation by means of electrical impedance tomography phy. Med. Biol. 54 (2009) 4927-4943
8. Thiel F Bioimpedanz-Analysator zur nichtinvasiven Funktions- und Zustandsanalyse von Geweben und Organen PhD thesis,University of Hannover, Germany, 2003
9. Min M, Land R., Paavle T, Parve T, Annus P and Trebbels D, Broadband spectroscopy of dynamic impedances with short chirp pulses. In: J. Physiological Measurement (32) 2011, S. 945-958
10. Adler A, Lionheart W R B: Uses and abuses of EIDORS: an extensible software base for EIT. In: J. Physiological Measurement (27) 2006,p.25-42
11. A. Kemma. Tomographic Inversion of Complex resistivity. PhD Thesis, Ruhr Universität, 200
12. M. Noel and B. Xu. Archaelogical investigation by electrical resistivity tomography: a preliminary study. Geophysical Journal International, 107:95-102, 1991
13. F. Dickin and M. Wang. Electrical resistance tomography for process applications. Meas. Sci. Technol. 7(1996) 247-260.
14. H. S. Tapp, A. J. Peyton, E. K. Kemsley, and R. H. Wilson. Chemical engineering applicatios of electrical process tomography. Sensors and Actuators B: Chemical, 92: 17-24, 2003
Filtering cardiac artefacts from transdiaphragmai pressure for the validation of a non-invasive method to assess work of breathing
M. Strutz, K. Lopez-Navas and U. Wenkebach
Abstract-To evaluate a new method to assess patient’s work of breathing towards the adjustment of the ventilation settings, it is necessary to measure the transdiaphragmai pressure. This paper deals with the effect of heartbeat artefacts on this procedure. The heartbeat is filtered off the data from three test subjects. The filtered and unfiltered data is analyzed and then compared by the resulting values of resistance, compliance, correlation coefficient (R[2]) and goodness of fit (chi_ok). The result is an increase in the values for R2 and chi_ok, which means that the filter has a positive effect without adulterating the data, being a useful tool to estimate the heartbeat artefacts from the measured signal, however, the filter can be improved if especially designed to fit to the heart rate of each test person.
I. INTRODUCTION
Nowadays respirators are used at every hospital, intensive care areas, nursing homes and even at the patient’s own place. In the last case however, the availability of trained personal as a nurse or a doctor to continuously check the adjustment of the respirator is limited.
This fact may have a negative effect on the patient’s state of health. Therefore it is indispensable to make possible that the respirator adjusts itself in reaction to the spontaneous breaths of the patient adaptating to his/her requirements. For this purpose the work of breathing (WOB) must be known to the respirator. A common and effective method for this is the measurement of transdiaphragmal pressure (Pdi), which requires the insertion of a balloon tipped catheter in the body and is therefore unpleasant and risky. In contrast, we currently develop a non-invasive method for the assessment of Pdi, which is described in detail in [2]. For the validation of this method we started a study with healthy volunteers. Unfortunately we observed soon a problem in the measured signals because the pressure caused by the heartbeat, also called cardiogenic oscillations, was seen as an artefact on the measured Pdi (see fig. 1). Because there is no opportunity to measure the transdiaphragmal pressure completely without heartbeat, it was necessary to identify if these artefacts have any effect on the validating process. According to the results it could be worth to implement modifications on the designed method, but the initial analysis is specific for the validation.
M. Strutz is with FH Lübeck - Tel: 0451 3005673 - merle.strutz@stud.fh-luebeck.de
K. Lopez-Navas is with FH Lübeck- Tel: 0451 3005626 - lopez-navas@fh-luebeck.de
U. Wenkebach is with FH Lübeck - Tel: 0451 3005501 - wenkebach@fh-luebeck.de
II. Material and Methods
A. Data
All analyzed data was obtained from our study with test persons. Thin catheters with two small latex balloons at the distal end were placed by a doctor to measure Pdi. One of the balloons was placed in the lower third of the oesophagus and the other in the stomach. The difference in pressure was calculated to get the transdiaphragmal pressure [1]. Established by the location of the balloons the heartbeat has an effect on the measurement which cannot be avoided. Respiratory flow (V’) and airway pressure (Paw) were also measured and recorded with a sampling rate of 200Hz. These signals are used by our alternative method to produce a non invasive estimation of Pdi which is not affected by the heart waves [2]. Since the validation of our method bases on the comparison between measurement and estimation of Pdi it is indispensable to get a clear measured signal.
illustration not visible in this excerpt
Fig. 1. Cardiogenic oscillations on measured transdiaphragmal pressure.
In addition to that the test persons were connected to a respirator by a face mask. Between the respirator and the test person there were a few sensors, which were linked to a measurement box connected to a computer to save all data.
The measurement was subdivided in three phases. In the first phase the test person breaths normally. In the second phase the anatomical dead space is augmented forcing the test person to use more work of breathing to compensate the extra resistance. In the third phase the respirator supports the test person by giving a fixed pressure while he/she still determines the time and length of a cycle, but the respirator takes the work from the diaphragm. The stetting of the respirator in the last phase is named Assisted Spontaneous Breathing (ASB). It works to support the inspiration by increasing the pressure at the beginning of each breathing cycle [3].
[...]
-
-
-
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X. -
-
¡Carge sus propios textos! Gane dinero y un iPhone X.