Automated Melanoma Staging in Lymph Node Biopsy Image using Deep Learning

Melanoma is the most dangerous type of skin cancer with aggressive and unpredictable behavior which can spread to any part of the body. The Proliferation Index (PI) in lymph node biopsy is an important indicator to determine the spread of the cancer in the human body. In this paper, we propose a technique for automatic measurement of PI values in Ki-67 stained biopsy image using deep learning algorithm. This technique segments the nuclei and measures the PI by using a trained convolutional neural network (CNN) model. The CNN model is trained using the manually nuclei segmentation of Ki-67 images. Experimental results show that the proposed segmentation technique can robustly segment and classify the nuclei with low computational complexity and has an average error rate less than 0.7%.