Quantitative Benchmarking of Anomaly Detection Methods in Digital Pathology

TL;DR

This paper provides a comprehensive quantitative benchmark of over 20 anomaly detection methods applied to digital pathology images, addressing unique challenges and guiding future research in the field.

Contribution

It systematically evaluates classical anomaly detection methods on curated pathology datasets, highlighting their strengths, limitations, and influencing factors.

Findings

Identified key factors affecting detection performance.

Compared strengths and limitations of various methods.

Established a comprehensive benchmark for future research.

Abstract

Anomaly detection has been widely studied in the context of industrial defect inspection, with numerous methods developed to tackle a range of challenges. In digital pathology, anomaly detection holds significant potential for applications such as rare disease identification, artifact detection, and biomarker discovery. However, the unique characteristics of pathology images, such as their large size, multi-scale structures, stain variability, and repetitive patterns, introduce new challenges that current anomaly detection algorithms struggle to address. In this quantitative study, we benchmark over 20 classical and prevalent anomaly detection methods through extensive experiments. We curated five digital pathology datasets, both real and synthetic, to systematically evaluate these approaches. Our experiments investigate the influence of image scale, anomaly pattern types, and training epoch selection strategies on detection performance. The results provide a detailed comparison of each method's strengths and limitations, establishing a comprehensive benchmark to guide future research in anomaly detection for digital pathology images.