CAMEL2: Enhancing Weakly Supervised Learning for Histopathology Images by Incorporating the Significance Ratio

Abstract

Histopathology image analysis plays a crucial role in cancer diagnosis. However, training a clinically applicable segmentation algorithm requires pathologists to engage in labor‐intensive labeling. In contrast, weakly supervised learning methods, which only require coarse‐grained labels at the image level, can significantly reduce the labeling efforts. Unfortunately, while these methods perform reasonably well in slide‐level prediction, their ability to locate cancerous regions, which is essential for many clinical applications, remains unsatisfactory. Previously, CAMEL is proposed, which achieves comparable results to those of fully supervised baselines in pixel‐level segmentation. However, CAMEL requires 1280 × 1280 image‐level binary annotations for positive WSIs. Here, CAMEL2 is presented, by introducing a threshold of the cancerous ratio for positive bags, it allows one to better utilize the information, consequently enabling us to scale up the image‐level setting from 1280 × 1280 to 5120 × 5120 while maintaining accuracy. The results with various datasets demonstrate that CAMEL2, with the help of 5120 × 5120 image‐level binary annotations, which are easy to annotate, achieves comparable performance to that of a fully supervised baseline in both instance‐ and slide‐level classifications.