Novel hybrid hardware architecture for nuclei detection in skin histopathological images

Histopathological examination is the gold standard for many kinds of cancer diagnosis as the histopathology slides provide a cellular level view of the disease. Computer-aided automated diagnosis techniques have recently been developed based on the analysis of skin histopathological images to diagnose skin cancer. Due to the very high resolution characteristics of the skin whole slide images, the image analysis has high computational complexity. In recent years, FPGA has been widely used to implement and speed up high complexity signal and image processing techniques. In this paper, a MATLAB-FPGA hybrid implementation of the histopathological image analysis technique for nuclei detection in skin whole slide image is proposed. The nuclei detection framework is comprised of two steps: (1) epidermis segmentation and (2) nuclei segmentation. The proposed implementation is compared with the MATLAB only results for accuracy and speed-up.