Hierarchical Salient Patch Identification for Interpretable Fundus Disease Localization

TL;DR

This paper introduces a hierarchical salient patch identification method for interpretable fundus disease localization, achieving superior accuracy with only image-level labels and neural network classifiers.

Contribution

The paper presents a novel weakly supervised approach combining hierarchical analysis and patch selection to improve disease localization accuracy in fundus images.

Findings

Achieved state-of-the-art localization performance on fundus datasets.

Effectively identified disease regions with only image-level supervision.

Validated the method's robustness through ablation studies.

Abstract

With the widespread application of deep learning technology in medical image analysis, the effective explanation of model predictions and improvement of diagnostic accuracy have become urgent problems that need to be solved. Attribution methods have become key tools to help doctors better understand the diagnostic basis of models, and are used to explain and localize diseases in medical images. However, previous methods suffer from inaccurate and incomplete localization problems for fundus diseases with complex and diverse structures. To solve these problems, we propose a weakly supervised interpretable fundus disease localization method called hierarchical salient patch identification (HSPI) that can achieve interpretable disease localization using only image-level labels and a neural network classifier (NNC). First, we propose salient patch identification (SPI), which divides the image into several patches and optimizes consistency loss to identify which patch in the input image is most important for the network's prediction, in order to locate the disease. Second, we propose a hierarchical identification strategy to force SPI to analyze the importance of different areas to neural network classifier's prediction to comprehensively locate disease areas. Conditional peak focusing is then introduced to ensure that the mask vector can accurately locate the disease area. Finally, we propose patch selection based on multi-sized intersections to filter out incorrectly or additionally identified non-disease regions. We conduct disease localization experiments on fundus image datasets and achieve the best performance on multiple evaluation metrics compared to previous interpretable attribution methods. Additional ablation studies are conducted to verify the effectiveness of each method.