Mitosis detection in breast cancer

A time-saving and cost-effective AI-based computer-aided system for pathologists and doctors to effectively and accurately identify the mitoses in WSIs and thus, they can count the mitoses efficiently. Mitotic count is crucial to evaluate the tumor aggressiveness. An AI-powered mitosis detection algorithm in digital pathology minimizes inter-observer variability among pathologists.

Deep learning based object detection algorithm is utilized to develop mitosis detection system. The train data set consist of images from 3 different scanner devices such as 3D Histech, Leica, Hamamatsu. Thus, the mitosis detection model is capable of giving promising results on 3 different image formats. The model demonstrated an F1- score of 73.62%, 69.34%, 56.96%, in 3D Histech, Leica, Hamamatsu scanner devices respectively.

NVIDIA GeForce RTX 3090 is recommended or a similar/equivalent one.

Pathologists, Pathologist technicians, private and public hospitals, medical research center, R&D Laboratories, diagnostic laboratories, pathology laboratories, medical centers, scientific experts, medical universities

This model and associated code are released under the CC-BY-NC-ND 4.0 license and may only be used for non-commercial, academic research purposes with proper attribution. Any commercial usage requires Virasoft and other project partner approval.