This will count as one of your downloads.
You will have access to both the presentation and article (if available).
Artificial Intelligence (AI) has found its place within the field of medical diagnostics. It can assist in feature recognition, in image segmentation and enhancement or in assessing specific structures. In the field of dermatology, there is a special interest in AI supported tools to classify skin lesions.
However, in order to train such a classifier, lots of image data is needed. The available image data on skin lesions shows a large variety in size, resolution and lighting as there is no underlying acquisition standard.
These properties make up the main components of the image data but do not contain any information on the actual subject and hence do hinder the training process. In this work, a method for cleaning image data is presented. It aims to standardize the image data in the aspects of image size, section and lighting without lowering the resolution and while maintaining the relevant structures shown.
To achieve this, the images are framed to ensure they all have the same size. In the next step, microscopic images are selected and their region of interest is determined. The lighting is adjusted by standardizing the colors. Afterwards, the final size adjustment follows.
The results were evaluated by carrying out a t-distributed stochastic neighbor embedding as well as a Principal Component Analysis. It could be shown that the presented routine improves the separability of benign and malign skin lesions which can also speed up and increase the quality of training process of AI models.
The results of measurements on a 100 nm height standard with both selected light sources have been compared. Under consideration of the coherence length of both light sources of 1.58 μm for the SC source and 1.81 m for the LDP source differences could be recorded. Especially at sharp edges, the LDP light source could record height changes with slopes twice as steep as the SC source. Furthermore, it became obvious, that measurements with the SC source tend to show edge effects like batwings due to diffraction. Additional effects on the measured roughness and the flatness of the profile were investigated and discussed.
We calculate the group velocity dispersion (GVD) of different cladding modes based on the measurement of the fiber structure parameters, the hole diameter and the pitch of a presumed homogeneous hexagonal array. Based on the scanning electron image, a calculation was made of the optical guiding properties of the microstructured cladding. We compare the calculation with a method to measure the wavelength-dependent time delay. We measure the time delay of defined cladding modes with a homemade supercontinuum light source in a white light interferometric setup. To measure the dispersion of cladding modes of optical fibers with high accuracy, a time-domain white-light interferometer based on a Mach-Zehnder interferometer is used. The experimental setup allows the determination of the wavelengthdependent differential group delay of light travelling through a thirty centimeter piece of test fiber in the wavelength range from VIS to NIR. The determination of the GVD using different methods enables the evaluation of the individual methods for characterizing the cladding modes of an endlessly single-mode fiber.
For the most precise measurement of DMD, we investigated a new type of method. It is capable of measuring the modal dispersion in two different ways. The first way is the standard transversal measurement, where the launching condition is altered by moving the radial position of the injected pulse while maintaining a zero-angle launching condition. The second way involves changing the launching angle into the fiber. This is done to get the most precise value for the DMD. Also, using a supercontinuum light source for the injection pulse, it is possible to vary the wavelength to be able to measure near the zero dispersion wavelength in order to investigate the effects of the chromatic dispersion.
In experiments on a height standard, it could be shown that the setup is capable of recording multiple height steps of 101 nm over a range of 500 m with an accuracy of about 11.5 nm. Further experiments on conductive paths of a micro-electro-mechanical systems (MEMS) pressure sensor demonstrated that the approach is also suitable to precisely characterize nanometer-sized structures on production-relevant components. The main advantage of the proposed measurement approach is the possibility to collect precise height information over a line on a surface without the need for scanning. This feature makes it interesting for a production-accompanying metrology.
Nicotinamide adenine dinucleotide (NADH) is a coenzyme naturally consumed and produced during cellular metabolic processes and has thoroughly been studied to determine the metabolic state of a cell. Measuring the fluorescence of NADH within the cell represents a non-disruptive marker for cell viability. Since the measurement process is optical in nature, NADH fluorescence also provides a pathway for sampling at different measurement depths within a given tissue sample. The measurement system we are using utilizes a special UV light source, to excite the NADH fluorescence state. However, the high energy potentially alters or harms the cells. To investigate the influence of the excitation signal, the cells were irradiated with a laser operating at a wavelength of 355 nm and examined for cytotoxic effects. The aim of this study was to develop a non-cytotoxic system that is applicable for large-scale operations during drug-tissue interaction testing.
At the beginning of a welding process, the heat-up phase of the metal is critical to the resultant weld quality. If a defined temperature range exceeded too fast, the risk of cracking is significantly increased.4 During the welding process the thermal supervision of the central processing location is decisive for a high secure weld. In the border areas as well as in connection of the welding process especially cooling processes are crucial for the homogeneity of the results. In order to obtain sufficiently accurate resolution of the dynamic heating- and cooling-processes, the system can carry out up to 500 frames per second.
GVD characteristics of two different large-mode-area double-clad fibers with defined launching pump laser power level were systematically analyzed. The dispersion parameters for different fiber designs and various doping levels are investigated over a broad spectral range in the emission area of Yb-doped fiber samples for controlled sets of operating parameters. The experiment utilizes a supercontinuum source developed within this laboratory as well as a Mach-Zehnder interferometer with a dual-channel spectral-detection system sensitive to wavelengths from 0.95 μm to 1.75 μm. Temporally resolved spectrograms recorded at distinct delay positions enable the detection of interference fringes for the equalization wavelength. By applying a Sellmeier polynomial fit to the wavelength dependent differential group-delay function, the GVD can be derived. The measured Yb-doped large-mode-area fibers show a variation of the doping concentration between 0.7 mass percent to 3 mass percent of ytterbium. The measurement of the Yb-doped large-mode-area fiber with or without optical load on the sample during the measurement was examined.
View contact details
No SPIE Account? Create one
