Melanoma is the deadliest form of skin cancer if left untreated. Incidence rates of melanoma have been increasing, especially among young adults, but survival rates are high if detected early. Unfortunately, the time and costs required for dermatologists to screen all patients for melanoma are prohibitively expensive. There is a need for an automated system to assess a patient's risk of melanoma using photographs of their skin lesions. Dermatologists could use the system to aid their diagnosis without the need for special or expensive equipment. One challenge in implementing such a system is locating the skin lesion in the digital image. Most existing skin lesion segmentation algorithms are designed for images taken using a special instrument called the dermatoscope. The presence of illumination variation in digital images such as shadows complicates the task of finding the lesion. The goal of this research is to develop a framework to automatically correct and segment the skin lesion from an input photograph. The first part of the research is to model illumination variation using a proposed multi-stage illumination modeling algorithm and then using that model to correct the original photograph. Second, a set of representative texture distributions are learned from the corrected photograph and a texture distinctiveness metric is calculated for each distribution. Finally, a texture-based segmentation algorithm classifies regions in the photograph as normal skin or lesion based on the occurrence of representative texture distributions. The resulting segmentation can be used as an input to separate feature extraction and melanoma classification algorithms. The proposed segmentation framework is tested by comparing lesion segmentation results and melanoma classification results to results using other state-of-the-art algorithms. The proposed framework has better segmentation accuracy compared to all other tested algorithms. The segmentation results produced by the tested algorithms are used to train an existing classification algorithm to identify lesions as melanoma or non-melanoma. Using the proposed framework produces the highest classification accuracy and is tied for the highest sensitivity and specificity.

Melanoma is the deadliest form of skin cancer if left untreated. Incidence rates of melanoma have been increasing, especially among young adults, but survival rates are high if detected early. Unfortunately, the time and costs required for dermatologists to screen all patients for melanoma are prohibitively expensive. There is a need for an automated system to assess a patient's risk of melanoma using photographs of their skin lesions. Dermatologists could use the system to aid their diagnosis without the need for special or expensive equipment. One challenge in implementing such a system is locating the skin lesion in the digital image. Most existing skin lesion segmentation algorithms are designed for images taken using a special instrument called the dermatoscope. The presence of illumination variation in digital images such as shadows complicates the task of finding the lesion. The goal of this research is to develop a framework to automatically correct and segment the skin lesion from an input photograph. The first part of the research is to model illumination variation using a proposed multi-stage illumination modeling algorithm and then using that model to correct the original photograph. Second, a set of representative texture distributions are learned from the corrected photograph and a texture distinctiveness metric is calculated for each distribution. Finally, a texture-based segmentation algorithm classifies regions in the photograph as normal skin or lesion based on the occurrence of representative texture distributions. The resulting segmentation can be used as an input to separate feature extraction and melanoma classification algorithms. The proposed segmentation framework is tested by comparing lesion segmentation results and melanoma classification results to results using other state-of-the-art algorithms. The proposed framework has better segmentation accuracy compared to all other tested algorithms. The segmentation results produced by the tested algorithms are used to train an existing classification algorithm to identify lesions as melanoma or non-melanoma. Using the proposed framework produces the highest classification accuracy and is tied for the highest sensitivity and specificity.

Purpose: Automatic border detection is the first and most crucial step for lesion segmentation and can be very challenging, due to several lesion characteristics. There are many melanoma border-detecting algorithms that perform poorly on dermoscopy images of basal cell carcinoma (BCC), which is the most common skin cancer. One of the reasons for poor lesion detection performance is that there are very few algorithms that detect BCC borders, because they are difficult to segment, even for dermatologists. This difficulty is due to low contrast, variation in lesion color and artifacts inside/outside the lesion. Segmentation that has adequate lesion-feature capture, with acceptable tolerance, will facilitate accurate feature segmentation, thereby maximizing classification accuracy. Methods: The main objective of this research was to develop an effective BCC border detecting algorithm whose accuracy is better than the existing melanoma border detectors that have been applied to BCCs. Fifteen auto-thresholding techniques were implemented for BCC lesion segmentation; but, only five were selected for use in algorithm development. A novel technique was developed to automatically expand BCC lesion borders, to completely circumscribe the lesion. Two error metrics were used that better measure Type II (false-negative) errors: Relative XOR error and Lesion Capture Ratio (a novel error metric). Results: On training and test sets of 1023 and 119 images, respectively, based on two error metrics, five thresholding-based algorithms outperformed two state-of-the-art melanoma segmentation techniques, in segmenting BCCs. Five algorithms generated borders that appreciably better matched dermatologists’ hand-drawn borders which were used as the “gold standard.” Conclusion: The five developed algorithms, which included solutions for image-vignetting correction and border expansion, to achieve dermatologist-like borders, provided more inclusive and therefore, feature-preserving border detection, favoring better BCC classification accuracy, for future work.

Image segmentation and image registration are two fundamental problems in computer vision and medical image processing. In image segmentation, one seeks to partition an image into meaningful segments by assigning a label to each pixel indicating which segment it belongs to. In image registration, one seeks to recover a spatial transformation that geometrically aligns two or more images, which allows downstream image analyses in which the registered images share a coordinate system. Image processing pipelines typically apply these procedures sequentially even though the segmentation of an image could improve its registration and registration of an image could improve its segmentation. With an appropriate parametrization, one can view these two tasks as an inference problem in which the spatial transformation and segmentation are latent variables. In this work, registration and segmentation are integrated through a hierarchical Bayesian generative framework. The framework models the data generating process of a set of magnetic resonance (MR) images of ischemic stroke lesioned brains. Under this framework, we simultaneously estimate a lesion tissue segmentation along with a spatial diffeomorphic transformation that maps a subject image into spatial correspondence with a healthy template image. The framework is evaluated on two-dimensional images both real and synthetic. Experimental results on real MR images show that simultaneous segmentation and registration can significantly improve the accuracy of lesion segmentation as well as the accuracy of registration near the lesion.

This thesis deals with the application of graph theory in image segmentation. There are specifically presented method utilizing graph cuts and extensions of this method. In the first chapter thera are initially explained basics of graph theory that are essential for understanding of the presented method. It is described in the second chapter, including its extensions that use shape priors. In the third chapter there is presented solution which is used for vertebrae lesion segmentation in the CT data sets. Final function is implemented into the program but it can be used also separately. Success rate is described using sensitivity and specificity in the last chapter, there are also examples of results.

Multiple sclerosis is a neurological disease causing a degeneration of myelin around the axons in the central nervous system. This process leaves traces in the form of lesions, which can be distinguished in an MRI examination. It is important to detect these at an early stage to state diagnosis and initiate medication.  In this Master's Thesis, an automatic segmentation algorithm was developed, with the purpose of segmenting possible multiple sclerosis lesions. Secondly, a progression model was developed with the purpose of estimating the state of each individual lesion. The implementation was based on synthetic contrast weighted images, segmentation maps and quantitative relaxation maps produced by SyMRI (SyntheticMR, Linköping, Sweden). The automatic segmentation algorithm has a relatively high sensitivity but low precision, causing a large number of false positives. The algorithm performed better in the cerebrum compared to the cerebellum. The large number of false positives appeared mainly due to partial volume effects, creating hyperintense artifacts in synthetic T2W FLAIR images. A larger amount of data would have been desirable to create a more robust algorithm. The progression model showed promising results, with a clear correlation to the synthetic contrast-weighted images and segmentation maps available in SyMRI. The progression model could be useful in disease monitoring, medical decisions and diagnosis of Multiple Sclerosis.