A DEEP LEARNING PIPELINE FOR BREAST CANCER KI-67PROLIFERATION INDEX SCORING

Abstract

The Ki-67 proliferation index is an essential biomarker that helps pathologists to diagnose andselect appropriate treatments. However, automatic evaluation of Ki-67 is difficult due to nucleioverlapping and complex variations in their properties. This paper proposes an integrated pipeline foraccurate automatic counting of Ki-67, where the impact of nuclei separation techniques is highlighted.First, semantic segmentation is performed by combining the Squeez and Excitation Resnet andUnet algorithms to extract nuclei from the background. The extracted nuclei are then divided intooverlapped and non-overlapped regions based on eight geometric and statistical features. A marker-based Watershed algorithm is subsequently proposed and applied only to the overlapped regions toseparate nuclei. Finally, deep features are extracted from each nucleus patch using Resnet18 andclassified into positive or negative by a random forest classifier. The proposed pipeline’s performanceis validated on a dataset from the Department of Pathology at Hôpital Nord Franche-Comté hospital.

Tasks

Reproductions