Efficient Identification of Melanocytic Nuclei in Pathology Images for Melanoma Diagnosis Using A Weakly-Supervised Deep Learning Framework

In this paper, we introduce a weakly-supervised deep learning framework that detects atypical melanocytic nuclei in histopathology images for the diagnosis of Melanoma with a nuclei detection accuracy of 92% surpassing the-state-of-the-art fully automated techniques. The proposed method will support pathologists in providing an objective, accurate and reliable diagnosis of Melanoma and minimizes the missed detection rate of the tumor. This will accordingly enhance the disease prognosis and improve the treatment efficacy.