Quantitative Benchmarking of Anomaly Detection Methods in Digital Pathology
This work addresses the need for quantitative benchmarks to guide anomaly detection research in digital pathology, which is important for applications like rare disease identification, but it is incremental as it focuses on evaluating existing methods rather than proposing new ones.
The authors tackled the problem of benchmarking anomaly detection methods for digital pathology images, which face unique challenges like large size and stain variability, by evaluating over 20 methods on five datasets and providing a detailed comparison of their performance.
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.