Extraction and validation of edge maps for breast thermal image segmentation using level sets

Abstract

Breast is a mammary gland that lies over musculature which encases the chest wall. Breast undergoes morphological changes due to change in the physiological conditions. Pathological condition such as breast cancer cause morphological and metabolic changes. Early diagnosis and prognosis improves the survival rate of breast cancer. Medical thermography captures thermal variations and preferred adjunct diagnostic tool for early diagnosis of breast cancer. The poor signal to noise ratio and low contrast are the inherent newlinelimitations of thermal images. The detection of breast contour is challenging task because of its amorphous nature and lack of sharp boundaries. To process the image the types of noise, its sources and the characteristics of noise need to be addressed. Segmentation of breast tissue is a means of identifying its pathological conditions. Segmentation is found to be challenging due to presence of vague and amorphous breast edges. Its complexity further increases in the region of lower breast boundaries and infra mammary folds. The extraction of edge map is essential as it provides reliable information about the breast boundaries. newline newline newline