Đánh giá chỉ số tăng sinh Ki-67 từ ảnh mô học tế bào ung thư vú

Abstract

This study employs DeepLabv3-Densenet121, FCRN, and U-Net models combined with YOLOv8 to evaluate the Ki-67 proliferation index from histological images of breast cancer cells. The SHIDCB-Ki-67 dataset, comprising 6000 images, was used. The method achieved a maximum accuracy of 98.1% in classifying and calculating the Ki-67 index, opening up potential applications in medical diagnostics.