Region and Spatial Aware Anomaly Detection for Fundus Images
This work addresses the challenge of improving diagnostic accuracy for ocular diseases by reducing false positives in fundus image analysis, representing an incremental advancement in domain-specific anomaly detection.
The paper tackles the problem of false positives in anomaly detection for fundus images, where existing methods often misclassify normal retinal structures as anomalies, and proposes a Region and Spatial Aware Anomaly Detection (ReSAD) method that significantly outperforms existing methods on two benchmark datasets.
Recently anomaly detection has drawn much attention in diagnosing ocular diseases. Most existing anomaly detection research in fundus images has relatively large anomaly scores in the salient retinal structures, such as blood vessels, optical cups and discs. In this paper, we propose a Region and Spatial Aware Anomaly Detection (ReSAD) method for fundus images, which obtains local region and long-range spatial information to reduce the false positives in the normal structure. ReSAD transfers a pre-trained model to extract the features of normal fundus images and applies the Region-and-Spatial-Aware feature Combination module (ReSC) for pixel-level features to build a memory bank. In the testing phase, ReSAD uses the memory bank to determine out-of-distribution samples as abnormalities. Our method significantly outperforms the existing anomaly detection methods for fundus images on two publicly benchmark datasets.