Abstract:

Since it is proved that the alternations of the cutaneous vasculature are involved in most cases of skin disorders, the segmentation and quantification of skin lesions with the aid of computer science are of great interest for medical research. In this master thesis, an optical coherence tomography angiography (OCTA) system is adopted to provide volumetric tissue and vascu- lar morphological information of human skin. Then an algorithm is developed to segment the lesion zones and extract the parameters automatically. By studying several common skin lesions such as cherry angiomas, basal cell carcinomas (BCCs) and benign nevi, OCTA is proved to be a powerful modality for the imaging of the microvascular architectures with high resolution. The feasibility of our algorithm is also verified by the results. Furthermore, the relationship among the parameters are discussed so as to find out the characteristics of skin lesions. At the end, a new sample/probe arm is assembled to improve the performance of the current OCTA system.