Segmentation of Skin Lesions’ Macroscopic Images for Cancer Diagnosis

Skin cancerous lesions are more effectively treated and the disease's spread can be prevented if they are diagnosed in the early stages. Specialized tests have demonstrated accurate skin cancer diagnosis based on images of skin lesions. So far, the majority of processing and diagnostic activities have relied on dermoscopic skin images, with high resolution and high separation accuracy, that a small group of skin and hair specialists examine. In today's societies, with widespread access to mobile phones, patients have the opportunity to take their skin photos to identify skin lesions and their probability of being cancerous. This early detection can prompt quicker treatment. In this study, we aim to apply intelligent image processing techniques for the segmentation of skin lesions in macroscopic images, which are low-quality images taken by mobile phones, to facilitate basic analysis. For cancer diagnosis, better segmentation of lesions leads to greater diagnostic accuracy. In this research, by tracing the neighboring points of the lesion border with Moore's edge detection method, we extract the lesion borders with good accuracy from the macroscopic images. The application of the proposed algorithm to macroscopic and dermoscopic images demonstrates reasonable segmentation accuracy in comparison. The proposed processing method warrants further investigation.

کلیدواژه ها:

Dermoscopic ، Macroscopic image ، Moor Neighbor tracing ، Segmentation ، Skin lesions ،