Quality Control of Whole Slide Images using the YOLO Concept

Computational pathology applies computer vision algorithms on whole slide images. The digitization of tissue glass slides marks a significant change in the clinical diagnostic workflow. One of the challenges in digital pathology is the presence of artifacts such as tissue fold, air bubbles, and ink-markers on archived cases. These artifacts may affect the focus points in digital scanners, and their presence may negatively affect the quality of the output tissue image and the subsequent diagnosis. Manual review of whole slide images requires experts, and it is a laborious and time-consuming task. In this paper, we trained the YOLO-v4 (You-Only-Look-Once) model to detect air bubble edges, tissue folds, which can happen during slide preparation, and ink-marked tissue glass slides, which occur when pathologists highlight regions of interest on glass slides. Our method is not only fast but also highly accurate. The experiments showed 99.5 % IOU calculation (intersection over union, also called Jaccard Index) for locating artifacts.