Evaluating AI-Based Mitosis Detection for Breast Carcinoma in Digital Pathology: A Clinical Study on Routine Practice Integration

Background: /Objectives: Histopathological diagnosis of invasive carcinoma breast samples includes the scoring of mitotic activity. This is a tedious and time-consuming task with high interpathologist variability. Methods: As an assistance to pathologists, we developed a deep learning based pipeline for mitosis detection and mitotic scoring according to the Elston and Ellis grading system on Whole Slide Images (WSI) for the first time here described. Results: We present its performance on routine data through a clinical study which clearly demonstrates its value. When assisted by Artificial Intelligence (AI), pathologists show better accuracy and reproducibility on the mitotic score. Conclusions: To the best of our knowledge, this is the first study to demonstrate that AI can successfully assist pathologists for mitotic score determination in human breast WSI in routine practice.