AIRAQc: Pre-Analytical Tool for Accurate Identification and Quantification of Artefacts in Histopathology

Kuldeep Gautam, Geetank Raipuria, Nitin Singhal

Keywords: Artefact Detection, Histopathology, Quality Control

TL;DR: We present AIRAQc for detection and quantification of artefacts on Histopathology Whole Slide Images. The solution yields superior performance as compared to existing approaches, testing on common artefacts types, and a low inference time.

Abstract: Quality control in digitized Whole Slide Images (WSI) is critical to eliminate the impact of slide processing and scanning artefacts. Current artefact detection methods often lack comprehensive categorization or analyze images at a single magnification, limiting their efficacy. AIRAQc addresses these limitations through a multi-magnification approach, training on a large-scale dataset of over 40000 WSI from diverse sources. The solution yields superior performance as compared to existing approaches, testing on common artefacts types - air bubble, fold, pen mark and out-of-focus, and low inference time of 0.11 sec/ mm2 of tissue.

Submission Number: 122