In this thesis, an existing non-contact dermatoscope will be further developed on the basis of knowledge and experience, and established as a new prototype for dermatoscopy at the Hannover Institute of Optical Technologies (HOT).In this system, the light generated by a white LED is collimated and polarized by a lens system, and generates a homogeneous light spot at a distance of 60cm. By cross polarization, the light reflected directly onto the skin surface can be suppressed, so that only light reflected in deeper skin layers can pass through the analyzer, and contributes to the image information. Due to the difficult handling of the original device, the further developed (advanced) system was compactified and automated, taking into account the basic principle of non-contact dermatoscopy. The illumination unit used in the original non-contact dermatoscope was replaced with a newly constructed reflector in order to improve the brightness, and the homogeneity of the light spot in the target area. These two reflectors were measured with a near field goniophotometer to characterize the illuminance distribution. The conducted tests included the definition of an ideal setup of the lens system, both in practice, and in optical systems simulations by using Zemax. It could be shown that the reflectors improve the illuminance, and generates a homogeneous light spot in the target area, which homogeneously illuminates the image area of the camera. Furthermore, this system has been completely automated by providing automatic focus as well as adjustment of one of the polarizers (analyzer) used. For this purpose, a automatic focus lens was integrated on the existing objective and a mid range infrared distance sensor was installed into the system. By various tests, such as the determination of the resolution with the modulation transfer function, the new camera system was characterized. Based on this tests, the highest possible resolution was determined and the work area could be defined. In this work area structures of 30 μm can be resolved sharply. In addition to the automation of the focus, a stepper motor has been installed to control the analyzer. A program was written in LabVIEW, which controls all components, automates the image acquisition, and provides the possibility of image processing (blood contrast enhancement). Subsequently, the entire system has been mounted on a variably adjustable swivel arm, in order to improve the handling for the dermatologist.
CONTENTS
ABSTRACT
1 INTRODUCTION
1.1 State of the art
1.1.1 Dermatoscopy devices by HEINE
1.1.2 Dermatoscopy devices by FotoFinder
1.2 Contact dermatoscopy
1-3 The original prototype
1-4 Research question and aims
2 THEORY
2.1 Design and functionality of a white-light-emitting diode ...
2.1.1 The functionality of an LED
2.1.2 Energy and momentum conservation at a pn-junction
2.1.3 The functionality and realization of a phosphor-based WLED
2.2 Focus tunable lenses
2.2.1 Functionality of the focus tunable lens
2.3 Basics of image formation and image quality
2.3.1 RGB-camera
2.3.2 Criteria for assessing image quality
2.3.3 Process for determining the image quality
3 CHARACTERIZATION OF THE CCD-CAMERA WITH AND FOCUS TUNABLE LENS
3.1 Connection of the focus lens with the objective
3.2 Resolution
3.2.1 Determination of the subjectively perceivable resolution
3.2.2 Determination of the modulation transfer function to characterize the objective resolution
3.3 Depth of field
4 CHARACTERIZATION OF THE SIMULATED AND REFLECOTR
4.1 The simulation with Zemax
4.1.1 The simulated reflector
4.1.2 Design and construction of the simulated reflector . .
4.2 Near field goniophotometer
4.2.1 Measurement with the near field goniophotometer . .
5 THE PROGRAM
5.1 The programming language Lab VIEW
5-2 Program structure
5.2.1 The program flow
6 THE FURTHER DEVELOPED DERMATOSCOPY DEVICE
6.1 Specifications of all commercially available components of the further developed prototype
6.1.1 Specification of the CBT-90 white LED
6.1.2 Specification of the camera Fkai 2.8 MP Color GigE Vision
6.1.3 Specification of the (automated) focus tunable lens EL- 16-40-TC
6.1.4 Specification of the mid range distance sensor DT35 ...
6.2 Design and operation of the further developed dermatoscopy device
7 SUMMARY AND OUTLOOK
BIBLIOGRAPHY
ABSTRACT
In this thesis, an existing non-contact dermatoscope will be further developed on the basis of knowledge and experience, and established as a new prototype for dermatoscopy at the Hannover Institute of Optical Technologies (HOT). The original device was first used by Günther et al. (2013) to study melanoma skin cancer and is particularly useful for the early detection of skin cancer and the differentiation of inflammatory skin diseases. In this system, the light generated by a white LED is collimated and polarized by a lens system, and generates a homogeneous light spot at a distance of 60 cm. By cross polarization, the light reflected directly onto the skin surface can be suppressed, so that only light reflected in deeper skin layers can pass through the analyzer, and contributes to the image information. Due to the difficult handling of the original device, the further developed (advanced) system was compactified and automated, taking into account the basic principle of non-contact dermatoscopy.
The illumination unit used in the original non-contact dermatoscope was replaced with a newly constructed reflector in order to improve the brightness, and the homogeneity of the light spot in the target area. These two reflectors were measured with a near field goniophotometer to characterize the illuminance distribution. The conducted tests included the definition of an ideal setup of the lens system, both in practice, and in optical systems simulations by using Zemax. It could be shown that the reflectors improve the illuminance, and generates a homogeneous light spot in the target area, which homogeneously illuminates the image area of the camera. Furthermore, this system has been completely automated by providing automatic focus as well as adjustment of one of the polarizers (analyzer) used. For this purpose, a automatic focus lens was integrated on the existing objective and a mid range infrared distance sensor was installed into the system. By various tests, such as the determination of the resolution with the modulation transfer function or with the USAF-1951 test chart, the new camera system was characterized. Based on this tests, the highest possible resolution was determined and the work area could be defined. In this work area structures of 30 pm can be resolved sharply. In addition to the automation of the focus, a stepper motor has been installed to control the analyzer. A program was written in Lab VIEW, which controls all components, automates the image acquisition, and provides the possibility of image processing (blood contrast enhancement). Subsequently, the entire system has been mounted on a variably adjustable swivel arm, in order to improve the handling for the dermatologist and make the examination as swift and comfortable as possible for the patient. This new system must now be tested in practice by dermatologists.
INTRODUCTION
Malignant melanoma (black skin cancer) is a type of cancer that can be successfully operated on early diagnosis. The most common type of examination is dermatoscopy, which is a non-invasive visual examination method for the diagnosis of pathological changes of the skin. The examination is made by visual inspection of the suspicious mole of by dermatologist, and a subsequent characterization. This characterization is based on the so-called ABCDE rule, which allows a classification of the mole due to its asymmetry, border, color, diameter, and evolution. This visual examination is not only suitable for the investigation of skin cancer, but also for inflammatory skin diseases, which are correlated with an optically visible change in the skin (such as psoriasis). In order to ensure characterization according to the ABCDE rule, the dermatologist can use a variety of aids, to magnify and better represent the considered skin area. Dermatoscopy allows such an enlargement and clearer presentation of the pathologically altered skin parts. This will enable dermatologists to more easily evaluate suspicious structures using the visual criteria such as the ABCDE rule. Optical dematoscopy devices are special reflected-light microscopes, which illuminate the skin area homogeneously with additional lighting. These reflected-light microscopes, on the one hand, can be simple magnifying glasses for enlarging the skin area to be examined, or, on the other hand, more complex reflected-light microscopes, which magnify the skin areas and store them digitally. An advantage of dermatoscopes with digital cameras over the magnifying glass is that the temporal evolution can be considered, because of the documentation of the image. A disadvantage of these complex reflected-light microscopes is, that many of them are placed onto the skin areas and mechanically deform the examined skin area causing secondary effects on the image and possible pain for the patient. Mechanical deformations change important parameters such as surface topography, or, color perception. For example, a mole that is stretched by the dermatoscope may appear paler, possibly leading to false diagnosis. Furthermore, the contact greatly affects the reproducibility of the viewed image, whereby the temporal development of the mole or an inflammatory skin disease is hampered. These disadvantages can be avoided by a non-contact imaging method, such as the non-contact dermatoscope. As the name suggests, there is no contact with the patient. As a result, the skin regions to be examined are not mechanically deformed, and, thus, do not distort the visual impression.
1.1 State of the art
This section will present selected commercially available dermatoscopy systerns by HEINE Optotechnik (Herrsching, Germany) and FotoFinder Systerns GmbH (Bad Birnbach, Germany), in order to provide an overview of the available devices. The systems presented here are both contact-based, and non-contact, as well as digital, and analog.
1.1.1 Dermatoscopy devices by HEINE
In the following, the two dermatoscopes NCI and DELTA20 T by HEINE will be presented. In both cases, it is a hand-held dermoscope (figure 1.1 shows both dermatoscopy devices).
Abbildung in dieser Leseprobe nicht enthalten
Fig. 1.1: Representation of the NCI (left) and DELTAK) T (right) dermatoscopy device by HEINE. Both devices can generate polarized and non-polarized illumination. By using polarized light in combination with a 90'-' twisted analyzer, the light reflected directly on the skin can be filtered. (NCI ; DELTA20)
With these handsets, the dermatologist is forced to refer to proven diagnostics criteria, i.e. the ABCD rule. The ABCD rule is a common guideline for dermatologists to differentiate between benign nevus and malignant melanoma. Attention is paid to irregularities in the shape of the pigment, as well as its color, boundary, and diameter. This combination can be used to evaluate abnormal changes in pigment and skin lesions. (Argenziano et al., 1998)
HEINE® NCI DERMATOSKOP
The dermatoscope NCI by HEINE is an analog incident light microscope. It can be used in both contact, and non-contact mode, where the possible magnification is different. In the non-contact investigation, a magnification of up to a factor of 6 can be realized, in the contact-based (with contact disc) a 9-fold magnification is possible. The optical system provides a good sharpness and high resolution. The LED illumination offers true and natural colors. (NCI; Heine®)
HEINE® DELTA20 T DERMATOSKOP
The dermatoscope DELTAK) T presented here can be used as a classic contact dermatoscope using an immersion fluid. It can also be used as incident light dermatoscope without immersion liquid with the addition of polarization. The user is allowed to switch between polarized and non-polarized LED illumination. Due to the combination of high-resolution optics and polarized light, vascular structures can be assessed well. It can be decided whether a skin tumor is present or not. (DELTA20; Heine®)
1.1.2 Dermatoscopy devices by FotoFinder
In the following, the two dermatoscopes medicam 1000 and bodystudio ATBM by FotoFinder will be presented (figure 1.2 shows both dermatoscopy devices). The difference to the previous models by HEINE is the digital pro-
Abbildung in dieser Leseprobe nicht enthalten
Fig. 1.2: Representation of the medicam 1000 (left) and body studio ATBM (right) dermatoscopy devices by Fotofinder, (medicam 1000; ATMB)
cessing of images. Both medicam 1000 and bodystudio ATBM are aided systerns and equipped with special software. Hence, the examined skin areas can be saved to a PC, and used for later comparison. This makes it possible, in addition to proven evaluation criteria (see above), to supplement computer-assisted analysis (indicating a risk ranking) to the dermatologist, (medicam 1000; ATMB)
FotoFinder medicam 1000
The dermatoscopy device medicam 1000 by FotoFinder is a digital incident light microscope that can be used to perform video documentation and fluorescence diagnostics. The system has an additional spacer, which ensures uniform measurement distances. The spacer causes contact with the patient while the recordings of the observed body parts occurs over the hand-held device medicam 1000. 140-fold magnifications of the image sections are possible. Connecting the medicam 1000 to a PC makes the images visible for both practitioner and patient on an external monitor. By connecting to a PC there is the possibility of image storage, which has the advantage of image comparability in future examinations, (medicam 1000)
FotoFinder bodystlidio ATBM
With the so-called Automated-Total-Body-Mapping-Method (ATBM®) whole body documentation is possible within a few minutes. With this system, the camera is automatically positioned with integrated auto focus and scans the entire body surface independently. This body-scan should visualize new or altered moles. As soon as the new images are saved, the bodystudio ATBM has already automatically compared them to the images from the last scan. This can help the dermatologist to find skin changes and identify suspicious lesions more easily. (ATMB)
1.2 Contact dermatoscopy
In the previous section 1.1, some commercially available dermatoscopy systerns were presented. For image acquisition employing with contact der- matoscopes, a contact plate with immersion liquid may be pressed onto the skin, whereby it is deformed, and the camera is located behind this plate. The immersion liquid fills the space between cover glass and objective. The use of immersion liquid raises the refractive index to 1.5- 1.8 (air 1.0), whereby Fresnel-reflection is reduced. This is because the space is now filled with an optically denser medium, so that total reflection is avoided. The Fresnel-reflection quantitatively describes the reflection and transmission of a plane, electromagnetic wave at a planar interface. Besides the reduction of total reflection, the larger refractive index of the immersion medium n reaches a larger numerical aperture NA = n · sin (a). A larger NA means that more light can be captured, thus the information content, and the resolution is improved.
However, the application of a contact plate also causes disadvantages. As a result of the appearing skin deformation, the reproducibility of the recording is hampered, whereby the documentation of the development of the skin disease is impaired. Because of this skin deformation no conclusions can be drawn on the surface morphology of the skin. By stretching the skin, the pigmentation seems paler. Furthermore, lesions on uneven places are difficult to capture, and can cause pain for the patient. Also, the vascularization of the skin lesions appears different compared to reality.
All these disadvantages are eliminated by the original prototype at the
Hanover Center for Optical Technologies (HOT) according to Günther et al. (2013); Meinhardt-Wollweber et al. (2017).
1.3 The original prototype
Abbildung in dieser Leseprobe nicht enthalten
Fig. 1.3: Schematical representation of the original prototype of a non-contact dermatoscopy device at the beginning of this master thesis, according to Günther et al. (2013); Meinhardt- Wollweber et al. (2017). The undirected light by the white LED is approximately focused onto the lens system by a commercially available plastic reflector. The lens system intents to form the necessary collimation, so that at 60 cm distance a homogeneously illuminated light spot/image field is created. This light first passes the polarizer, and then hits and interacts with the target (skin). The reflected light reaches the CCD-camera through an analyzer and an objective. Depending on the desired representation the light reflected directly on the skin surface can be suppressed by the analyzer. By means of the CCD-camera, the image can subsequently be reconstructed, and displayed on the monitor. This image can then be saved on the PC.
The original prototype (see fig. 1.3) uses a 22501m CBT-90 white LED by Luminus, Inc. (Sunnyvale, USA) as a illumination source (CBT90). The undirected radiation is captured and directed by a reflector type TYRA-S by Ledil Oy (Salo, Finland), which reduces the losses due to the wide radiation angle of the LED. The subsequent lens system consists of three lenses by Thorlabs Inc. (New Jersey, USA) (from right to left):
- Focal length 100 mm (S/N: LA1050-A)
- Focal length 60 mm (S/N: LB1723-A)
- Focal length 60 mm (S/N: LA1401-A)
These three lenses direct and limit the divergence and form the necessary collimation, such that at 60 cm distance a homogeneously illuminated image field is created. After thepropagating through lens system, the light passes through a polarizer, and reaches the target. The reflected light falls onto the CCD-camera type Flea3 2.8 MP Color GigE Vision by FLIR Integrated Imaging Solutions Inc. (Richmond, Canada) through an analyzer and the objective MVL75M1 by Thorlabs Inc. (New Jersey, USA). The smallest structures relevant to dermatoscopic diagnostics have a diameter of 30 pm- 180 pm (Meinhardt-Wollweber et al., 2017). In combination with the objective and the CCD-camera, structures of 19.7 pm size can still be resolved, hence, no information is lost due to poor resolution. Another challenge in dermoscopic imaging is the suppression of disturbing (unwanted) skin surface reflections. Due to the high intensity of light reflected directly at the surface, information from deeper layers is superimposed and lost. To avoid this, the so-called cross-polarization was installed in the pototypes. Figure 1.4 shows the difference between a cross-polarized and a non-cross- polarized dermatoscopic image.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 1.4: Difference between a non-cross-polarized (left) and a cross-polarized (right) dermato- scopre image. In the left image, the light reflexion is clearly visible. The pictures were taken with the original prototype. (Meinhardt-Wollweber et al., 2017)
In order to suppress the skin surface reflections, a certain polarization angle is predetermined by the polarizer. This polarized light with a defined polarization angle illuminates the skin. However, this usually leads to reduction of the light intensity. Part of the light penetrates the skin. The light, which penetrates the skin changes the polarization through interaction with the different skin layers and becomes depolarized. The light reflected directly on the skin surface retains its polarization. The use of another polarizer, a so-called analyzer which is rotated by 90° with respect to the first polarizer blocks the disturbing reflections of the skin surface. The depolarized light from deeper skin layers can pass through the analyzer, and contributes exclusively to image information. To avoid motion blur during image acquisition caused, e.g., by respiration, the camera has a global shutter. The image can then be displayed on the monitor and saved on the PC. (Günther et al., 2013; Basu et al., 2015; Meinhardt-Wollweber et al., 2017)
1.4 Research question and aims
Based on the achieved results of the original prototype, which was published by Meinhardt-Wollweber et al. (2017), and the gained knowledge of the importance of this device, the prototype should be further developed and expanded. For the intended dermatoscopy device, a profile of requirements with corresponding solution variants was created in advance:
Direct contact of the dermatoscopes with the patient may cause pain due to malignant changes in the skin. A non-contact dermatoscopy device, such as the presented prototype, avoids contact and thereby caused no pain. Another important aspect is the unwanted skin surface reflection, which is avoided with the cross polarization, and is integrated in the original prototype. The cross polarization must to be implemented in any new version of the device. Furthermore, a light-intense, homogeneously illuminated light spot, which irradiates the skin region, is required. The more light falls onto the desired skin surface, the more skin information can be gained. In order to collimate the light of the LED homogeneously onto the target area, the developed lens system should be replaced by a custom-designed reflector (see fig. 1.3). On the one hand, the reflector should collimate more light on the target area, and, on the other hand, the construction should become more compact. In order to represent all blood vessels, a resolution of about 30 pm is needed. Since this resolution was achieved with the previous prototype, both objective and camera (as discussed in sec. 1.3) may also be implemented within the new prototype. Recorded images should be saved in order to enable comparison in later investigations. For the improvement of the handling of the device for the dermatologist during the investigation, a program will be developed. To acquire new knowledge considering the blood contrast in the image, an analysis possibility is to be implemented within the program. In addition, the response time of all parameters should be reduced to make the examination as comfortable and swift as possible for the patient. The emphasis is on finding the optimal focus because it takes most of the examination time. For this purpose, an automatic focus lens is to be integrated into the new system.
2 THEORY
2.1 Design and functionality of a white-light-emitting diode
A light-emitting diode (LED) is a semiconductor component whose electrical properties correspond to those of a diode. If current flows along the pass direction of the diode, light of a certain wavelength is emitted depending on the semiconductor material and its type. This light is produced in the pn- junction of the semiconductor material due to electron-hole recombination.
The use of an LED rather then a conventional light source has some major advantages in dermatoscopy:
- Small compact design
- High efficiency
- High luminous flux
- Long lifetime
This variety of advantages arise from the construction of an LED which consists of a semiconductor material with a pn-junction. A special form of the LED is the white-light-emitting diode (WLED). As the WLED СБТ-90 white LED by Luminus, Inc. (Sunnyvale, USA) has been integrated in the original and further developed experimental set-up, the functionality and the realization of a phosphor-based WLED will be explained in detail. Before discussing a WLED, the fundamental mechanism of an LED is explained.
2.1.1 The functionality of an LED
Ligure 2.1 graphically shows the band model of a semiconductor. The energy states of the electrons are not sharply separated in semiconductors, but they are rather comprised in broad bands. Electrons form electron-hole pairs with the defect electrons. Holes or defect electrons describe the missing valence electron in the semiconductor. An electron-hole pair can be bound as well as unbound (free charge carriers). Lor the unbound case, energy must be absorbed, whereby the electron passes to the excited state (conduction band) and leaves back a hole in the valence band. Such a separate electron-hole pair is shown in figure 2.1 in the intrinsic and p-type semiconductor.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.1: Representation of the band model for semiconductors. Here, Eg stands for the band gap energy, Ед for the excitation energy of the defects, VB for the valence band and CB for the conduction band. On the left, the intrinsic semiconductor is represented. If additional foreign atoms are introduced into the semiconductor material, which have one more electron in the valence band than the pure semiconductor, an impurity level is formed near the lower energy of the conduction band (see middle picture). If additional foreign atoms are introduced into the semiconductor material that have one electron less in the valence band than the pure semiconductor, an impurity level near the valence band edge is formed (see right picture). These electrons and holes can move freely and independently of each other in the material. (Meschede, 2008)
The current is carried by the electron in the conduction band and by the holes in the valence band. The state of energy E occupied by an electron is described by the Fermi distribution as a function of temperature T and Fermi energy £p (Meschede, 2008):
Consequently, for the distribution of the holes fh results:
Abbildung in dieser Leseprobe nicht enthalten
At T = 0, all energy states below the Fermi energy are fully occupied (see fig. 2.2 left). Characteristically, the Fermi energy is located in the middle of the band gap in an intrinsic semiconductor. This fact means that the occupation probability of the states can be approximately described by the Boltzmann distribution (Meschede, 2008):
Abbildung in dieser Leseprobe nicht enthalten
Typically, the band gap energy is in the order of eV, such that there are only a few electrons in the conduction band at room temperature. Because of the material-dependent band gab, the Fermi energy £p can be shifted towards the band edges by doping (i.e., the introduction of charge carriers, such as electrons and holes). Depending on the type of diode, a distinction is made in p- and n-type. Due to the type, the conductivity is increased enormously. If additional holes are generated as charge carriers in a semiconductor, the Fermi energy shifts downwards the acceptor level (see fig. 2.2 right). This doping is called p-type semiconductor, and can be found, e.g., in Aluminum (III. Main Group) -type silicon crystal (IV. Main Group). The
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.2: Representation of the Fermi distribution in intrinsic, n- and p-type semiconductors. Here, ε Ţ stands for the Fermi energy, VB for the valence band and CB for the conduction band. The dark gray-shaded areas represent the population distribution at T = 0. At T = 0, all energy states of the intrinsic semiconductor below the Fermi energy are fully occupied (left). In n-type semiconductors, electrons are situated in the conduction band (middle), while in p-type semiconductors, holes are situated in the valence band (right). (The distribution of the holes results from the reflection of the distribution of the electrons at the respective Fermi energy.) (Meschede, 2008)
middle of figure 2.2 shows an n-type semiconductor, such as phosphorus (V. Main Group) -type silicon crystal. Here, electrons are generated as charge carriers in the semiconductor, and the Fermi energy shifts upwards the direction of the donor level. These additional charged carries are movable, thus, improved conductivity is observed at room temperature. Figure 2.3
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.3: Representation of the charge carriers at a pn-junction. Left: No applied voltage is represented. Middle: The pn-junction is electrically driven in the reverse-biasing. Right: The pn-junction is electrically driven in the forward direction. (Meschede, 2008)
shows the basic mechanisms that occur at a pn-junction with and without applied voltage. A pn-junction refers to a material transition in semiconductors, where two oppositely doped semiconductor materials are combined. If p- and n-type semiconductor materials are combined without external influ- enees (such as an applied voltage), a so-called space-charge region results. This space-charge region is caused by the free charge carriers generated due to the doping. The free electrons diffuse into the p-type region and the free holes into the n-type region. The electrons recombine with the stationary holes in the p-type region and the free holes with the stationary electrons in the n-type region, respectively. As a consequence of the diffusion and recombination of the majority charge carriers with the minority charge carriers, the boundary layer depletes in charge carriers, and an electric field with a potential AV is created. The majority charge carriers describe the free charges generated by the doping, and the minority carriers the fixed charges. This field exerts a force on the remaining charge carriers and counteracts further diffusion. The space-charge zone is dependent on the type of atoms and their type density. If a voltage is applied in the reverse direction (see fig. 2.3), the electric field is enhanced, and the space-charge region increases. If a voltage is applied in the forward direction, the space-charge region and the electric field is reduced, a current will flow across the transition. As a result, electrons and holes can recombine in the pn-junction, and recombination radiation is generated depending on the band gap energy. This recombination radiation forms the basis for the LED radiation. (Meschede, 2008; Eichler, 2015)
2.1.2 Energy and momentum conservation at a pn-junction
The most commonly used semiconductor materials belong to the III. and V. main group of the periodic system of the elements. The basic material for light-emitting diodes is a direct semiconductor, usually a gallium compound doped with an element of the V. main group, such as phosphorus. By different combinations of specific elements of these main groups and by different mixing ratios, the band gap, and, thus, the band gap energy can be varied. Among other things, the band structure of the semiconductor determines the behavior of the energy transfer during the transition of an electron from the conduction band into the valence band and vice versa. The size of the band gap determines the wavelength of the LED. A combination of the III. and V. main groups can generate very short wavelength (UV range), which forms the basis of diodes emitting white light. To generate
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.4: Representation of a direct and an indirect semiconductor. Left: The density of elec- tronie states and the dispersion relation for the direct semiconductor GaAs. Right: The indirect transition by means of the Si semiconductor is represented. The direct transition is not allowed with Si semiconductors. This is recognized by the crossed-out arrow. The gray-colored areas in the dispersion diagram symbolize occupied electronic states in equilibrium. Due to the desired equilibrium of forces, the bands have a parabolic shape. This shape corresponds approximately to an harmonic oscillator. (Meschede, 2008)
recombination radiation, free electrons and holes must be present. The recombination of an electron and a hole requires conservation of energy and momentum. Otherwise, the emission of light is not possible. Due to the con-
servation of energy, the emitted radiation has the wavelength of the order of the band gap. The reason for this is the so-called (interband) relaxation. Relaxation describes the transition of a system into its basic or equilibrium state. In this case, the excited electron looses energy in the form of heat through the interaction with phonons (lattice vibrations). As a result, the wavelength only approximately coincides with the band gap energy. In addition to energy conservation, the recombination radiation must also fulfill the conservation of momentum for the electron-hole pair (hke1,hkh) as well as for the emitted photon (hkph):
Abbildung in dieser Leseprobe nicht enthalten
Direct transitions can only be made if the lowest electronic state is directly above the highest hole state (see fig. 2.4). Here, the momentum conservation is satisfied because the к-values are identical. For G a As, an electron from the conduction band can recombine at the position к = 0 with a hole in the valence band. In this case, a photon is emitted with approximately the band gap energy. Otherwise (e.g., Si) the momentum conservation is not fulfilled, because the к-values are different. This situation forbids a direct transition. The element Si is the prominent counterexample to the direct transition and describes the indirect transition. Indirect transition can be accomplished with the help of phonons. The indirect transition will not be considered in more detail, because it does not apply to the case of LEDs. (Meschede, 2008)
2.1.3 The functionality and realization of a phosphor-based WLED
The first LEDs emitted red light, and were useless for phosphor based white light LEDs, as in general, long wavelength light could not convert to short wavelength light. Over the years, LED technique has been further developed, such that shorter and shorter wavelength down to the ultraviolet range could be generated. By converting a part of the short-wave (blue) radiation into long-wave light in the yellow range, it was possible to realize LED-based white light sources. (Shinde et al., 2012)
The functionality of a phosphor-based WLED
The short-wave radiation of an LED is converted by a phosphor layer into long-wave radiation of the yellow wavelength range (see fig. 2.5). This process, converting absorbed light of shorter wavelength into higher wavelength, is called fluorescence. The most common variant of a phosphor- based WLED is a yellow phosphor (YAG : Ce[3]+; Yttrium-Aluminum-Garnet doped with Cerium) coated InGaN-LED (Indium-Gallium-Nitride). InGaN- LEDs emit blue radiation. (Nakamura et al., 2013; Shinde et al., 2012) The Yttrium-Aluminum-Garnet is referred to as crystalline host lattices and the
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.5ะ Spectrum of a typically white LED.
Cerium atoms as the doping element. The Cerium atoms are the optically active component of this compound. By exciting this layer in the region of 460 nm, depending on the Cerium concentration, green to yellow radiation is emitted. A typical spectrum of such a WLED is shown in figure 2.5. (Nakamura et al., 2013)
Realization of a phosphor-based WLED
There are several methods to manufacture a phosphorous-based WLED. In order to avoid inaccuracy, such as different layer thicknesses and consequently a different conversion, the phosphor has to be applied homogeneously and with a defined layer thickness onto the LED-chip. This is an important part in the manufacturing process of an LED.
Phosphor Blue LED
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.6: Widespread realization of a high-performance WLED. On the LED-chip, a defined layer thickness of the phosphorescent substance is applied. Below the blue LED, the heat sink is located, which dissipates the heat generated during the radiation process.
2.2 Focus tunable lenses
In order to improve the handling of the presented dermatoscopy device, we were looking for a solution to adjust the focus of the lens system automatically. A focus tumble lens EL-16-40-TC by Optotune Switzerland AG (Dietikon, Switzerland) was implemented into the system. This automatic focus lens was designed on the basis of the human eye. The eye is the oldest but most successful system to bring objects into focus automatically. The eye's lens is neither moved forwards nor backwards. Here, only the elastic lens material of the eye is bent for focusing. Hence, the lens is reshaped continuously. (Blum et al., 2011; EL-16-40-TC)
2.2.1 Functionality of the focus tunable lens
The Optotune focusable lenses are based on a combination of optical fluids and a polymer membrane. A container which is filled with an optical fluid and surrounded by a thin, elastic polymer membrane, represents the heart of the Optotune lens. This elastic polymer membrane seals the container. The deflection of the membrane, and, thus, the radius of curvature of the lens, can be altered by pumping fluid into or out of the container. This is done by pressure on the outer ring. The pressure difference can be controlled, e.g., mechanically, electromechanically, pneumatically, etc.. In the presented figure 2.7, the electromechanical method is used. Piezo elements were integrated into the outer ring, which perform a mechanical movement when applying an electrical voltage via the so-called piezoelec- trie effect. With increasing voltage, the pressure is increased to the outer ring, thereby the fluid is pumped into the center of the lens. Depending on the voltage, the shape of the lens can be changed. (Blum et al., 2011; EL-16-40-TC; Muller et al., 1992)
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.7: Realization of a focus tunable liquid lens according to Blum et al. (2011). Left: The voltage is zero. Right: When the voltage is increased, the radius of curvature of the lens increases due to the piezo elements. These piezo elements cause mechanical downward movement (ere- ating pressure on the outer ring), which pumps the liquid into the lens.
2.3 Basics of image formation and image quality
The camera used in this work is the RGB camera Flea3 2.8 MP Color GigE Vision by FLIR Integrated Imaging Solution Inc. (Richmond, Canada). The following section explains the basic functionalities of the most common and commercially available (charge-coupled-device and complementary metal oxide semiconductor sensors). The main component of a digital camera is the camera sensor. With its technical specifications, the sensor in combination with the objective decides about the image quality of the recording. The incident light is converted into electrical signals by the camera and afterwards further processed into a digital signal. From the data obtained, an image can be reconstructed.
CCD-camera
The charge-coupled-device (CCD) sensor consists of pixels arranged in a grid. The number of pixels in combination with the objective provide information concerning the possible resolution. The structure of a pixel is described in the following and graphically represented in figure 2.8.
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.8: Structure of a pixel of a CCD sensor according to Hűik (1998); Nakamura (2016). One pixel consists of a semiconductor substrate, which is mostly realized by silicon. Two optically transparent electrical conductors (electrodes) are situated on this substrate, such that a voltage can be applied. By means of an optically transparent, electrically isolated layer (on top of the p-type semiconductor) the positive electrode is electrically isolated from the substrate by a thin silicon dioxide layer (Si02־layer).
Applying voltage in the transmission direction to an electrode, an area is formed below the isolated layer on the surface of the semiconductor and is referred to as a potential well. In this region of the potential well, the charge carrier concentration of the majority charge carriers is very small (depletion area). When a photon is impinging on a pixel, the light is detected by a photodiode. Due to the inner photo effect, the light is converted into electrical charge. Photons with energy greater than the band gap of the semiconductor lift electrons from the valence band into the conduction band (see sec. 2.1.1). Thus, electron-hole pairs are created in the semiconductor. The generated electrons migrate to the positive electrode within the potential well, while the defect electrons reach the negative electrode (cathode). The charge in the potential well can be integrated over a certain time. The stored charge is proportional to the amount of incident light.
Incident photons Uph(A), whose energy is greater than the band gap of the semiconductor, lift ηει(λ) electrons from the valence band into the conduction band. The efficiency of this process is described by the quantum efficiency. The quantum efficiency, QE(A) describes the ratio between emitted photons and excited electrons (Nakamura, 2016; Art, 2006):
Abbildung in dieser Leseprobe nicht enthalten
[Abbildung in dieser Leseprobe nicht enthalten] defines the photon flux falling onto one pixel, with τ as the exposure time of the camera. The product of photon flux and exposure time corresponds to the number of photons falling onto the camera sensor:
Abbildung in dieser Leseprobe nicht enthalten
The quantum efficiency is different for various wavelengths, and can be described using the spectral sensitivity of the camera sensor. The free electrons are stored up to a so-called full-well-capacity within a potential well. The amount of stored charges is proportional to the photon flux of the incident light. The full-well-capacity of individual potential wells can be saturated when the exposure time τ of the CCD sensor is too long. This allows electrons to excite adjacent pixels, resulting in a bright spot around the local overexposure (so-called blooming). In the captured image, the adjacent pixels are no longer to be differentiated from each other. (Nakamura, 2016; Erhardt, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.9: Interline transfer according to Allen and Triantaphillidou (2012). The collected charge carriers are vertically shifted out of the image area into a light-insensitive and shaded storage area. The storage areas are arranged as rows just next to the pixels, and shift the collected charge down by a row at the same time into the serial read out register.
For the determination of the number of electrons stored in the potential wells, each pixel must be read out. The CCD camera is named by the read out principle, so-called charge-coupling-device. Each pixel in a CCD is subdivided by three photogates (electrodes). The charge shifts in the direction in which the potential is applied. This principle is called "bucket chain principle". After exposure, the collected charge carriers are vertically shifted into a light-insensitive and shaded storage area, so-called shift register. The shift registers are arranged as rows just next to the pixels (see fig. 2.9). (Allen and Triantaphillidou, 2012)
The serial read out register is able to shift the charge in each pixel towards the amplification and analog-to-digital converter (ADC) circuit. There, the amount of charge is first converted into a voltage signal, and afterwards into a digital one. Due to the shielding strips (storage area) between the light- sensitive imaging area, this process is called interline transfer. (Erhardt, 2008; Nakamura, 2016; Allen and Triantaphillidou, 2012)
The further procedure up to the reconstruction of an image is visualized in figure 2.10. After a voltage signal has been read out from a pixel in
Abbildung in dieser Leseprobe nicht enthalten
Fig. 2.10: Read out process of a CCD camera according to Hűik (1998). Light shines onto the camera sensor, and lifts a number of electrons corresponding to the quantum efficiency in the conduction band. These electrons shift into the potential well, where the stored charge quan- titles of the individual pixels are successively read out according to the bucket chain principle. By temporal variation (tļ < t2 < t3) of the voltages (Vo < V| < V2) at the metal contacts, the charge packets are shifted by field-supported diffusion. Three of the shown processes correspond to the shift by one pixel.
the read out register, the signal can be amplified by means of the signal
amplification g, the so-called gain. The gain has a linear effect on the input signal lininei), which depends on the number of separated electrons:
Abbildung in dieser Leseprobe nicht enthalten
The gain can be changed manually by the user, such that the brightness of the image increases or decreases. From the resulting signal Igain, a value is determined in ADU (amlog-to-digital units) or DN (digital numbers) via an analog-to-digital converter (ADC). The digital numbers are the digital values after the conversion of the voltage signal. The conversion of the analog to digital signal can be described in digital units per electron ADU/e~ or DN/e~ by the system gain GSystem of the camera (Erhardt, 2008):
The system gain Gsystem indicates the conversion (without electronic amplification) of electrons to DN, and, thus, provides information about the efficiency of the system, such as sensitivity, system losses (e.g.: the higher the number of bits of an ADC, the finer the gradation of the digital values, which results in a higher sensitivity). The digital value is converted into brightness values, so-called gray values, before the image is reconstructed. Using the system gain Gsytem, equation 2.10 can also be written as (Nakamura, 2016; Mullikin et al., 1994):
The number of bits defines the storage depth, hence, the maximum number of gray values. For example, an 8-bit mono-camera has 2s = 256 gray values. The gray value GW corresponds to the output signal. If the y-value is set to 1, there is a linear correlation between the input signal Iįn and the output signal Iont· From the calculated gray values, the image can be reconstructed.
Due to the structure of a CCD-sensor, the pixels cannot be read out simultaneously. Therefore, some pixels will be exposed longer than others, which can eventually lead to the blooming effect. In order to be able to read out the pixels simultaneously, the so-called global shutter is used. The basic requirement for this process is the interline transfer architecture. With the global shutter, the deletion and subsequent exposure of all sensor lines take place simultaneously. As previously described, at the end of the exposure, all lines are simultaneously shifted to a shaded and light-insensitive sensor area next to the pixel columns. The individual lines are read out separately. The simultaneous exposure of all lines has the advantage that the image of a moving object is reproduced distortion-free. (uEye, 2016)
CMOS-camera
A complementary metal oxide semiconductor (CMOS) sensor is basically built like a CCD-sensor. A CMOS-sensor, which basically also consists of pixels arranged in a grid, is composed of a photodiode and a potential well. The fundamental difference to the CCD-sensor is that a transistor is mounted directly onto individual pixels. This means that the detected optical signals are directly converted to the corresponding voltage signals on the spot. However, less space for the photosensitive area remains in one pixel, and therefore the light efficiency per pixel is lower compared to the CCD-sensor. As a result, the photosensitive area has less space in one pixel. (Allen and Triantaphillidou, 2012)
2.3.1 RGB-camera
A black-and-white camera can only detect intensity differences, but no differences in the wavelengths of the incident light. To selectively detect different colors, RGB-cameras use color filters in front of the pixels which are applied in a specifically placed pattern, such as, e.g., the Bayer pattern, which is the most commonly used filter. The pixels are thus exposed only to light of a particular color (red, green and blue). All other colors can be reconstructed from these three basic colors. A color pixel consists of four pixels, so-called subpixel, with one red, one blue and two green color filters, in order to come close to the color perception of the human eye. This is because the human eye has its highest sensitivity in the green spectral range. Since a color pixel consists of four individual subpixels, the usable pixel number is also reduced by a quarter and thus the resolution of the camera is affected the same manner. If only the individual color channels (red, green and blue) are read out, the resolution is also reduced because the pixel distance doubles. (Erhardt, 2008; Nakamura, 2016; Allen and Triantaphillidou, 2012)
2.3.2 Criteria for assessing image quality
How well structures in an image can be recognized depends on the quality of the image. The image quality is initially a subjective impression, but can be classified, and objectively studied with different criteria, which are explained in the following sections.
Image sharpness and resolution
The resolution generally describes the smallest possible distance between two structures such that they can be detected separately from one another (Delorme et al., 2012). A distinction is made between moving objects (tern- porai resolution) and fixed objects (spatial resolution). The sharper the structures are imaged in a recording, the better they can be resolved.
[...]
-
-
-
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X.