Investigations on deep learning Framework for mitosis detection in Histopathology images

Abstract

Cancer is one of the deadliest diseases and causes a high mortality rate; breast cancer is one of the types of cancer that is diagnosed in women at a rapidly increasing rate. Various screening mechanisms, such as mammography, magnetic resonance imaging, thermography imaging, and microscopy imaging, were used for the diagnosis of cancer and understanding of the severity of the disease. In modern medicine, digitized histopathology, which is a kind of microscopic imaging, is becoming a more popular and widely accepted procedure for final decision-making in disease prediction. Histopathology Whole Slide Image (WSI) analysis through machine vision techniques is necessitated to investigate any abnormality in the biological structure of the cell and is helpful for effective therapeutics. Automated computer-aided diagnosis of histopathological images is a challenging problem in the field of medical imaging. There is also a growing demand for computer-assisted automatic detection of suspected lesions in histopathology images, which helps minimize the manual procedure carried out by pathologists and reduces the time spent on biopsy slide reading and interpretation. The important prognostic marker in breast cancer diagnosis is the mitotic figure count in histopathology images newline