Hardware Implementation of Epidermis Segmentation in Skin Histopathological Images

The histopathological analysis of tissue specimens is the gold standard to diagnose many kinds of cancer. Recently, a few computer-aided diagnosis (CAD) techniques have been proposed for the automated diagnosis of a skin cancer type, known as melanoma, by analyzing whole slide images (i.e., biopsy slides). The first step in the automated melanoma diagnosis is the epidermis segmentation of the biopsy images. The whole slide images are typically very large and hence require a significant amount of processing time. In this paper, we propose a hardware accelerator for the epidermis segmentation in a skin whole slide image. Experimental results show that the proposed system provides a significant speedup over a software-only implementation.