Multi-instance curriculum learning for histopathology image classification with bias reduction.

Multi-instance learning (MIL) exhibits advanced and surpassed capabilities in understanding and recognizing complex patterns within gigapixel histopathological images. However, current MIL methods for the analysis of the histopathological images still give rise to two main concerns. On one hand, vanilla MIL methods intuitively focus on identifying salient instances (easy-to-classify instances) without considering hard-to-classify instances, which is biased and prone to produce false positive instances. On the other hand, since the positive tissue occupies only a small fraction of histopathological images, it is commonly suffer from class imbalance between positive and negative instances, causing the MIL model to overly focus on the majority class in training instances classifier. In light of these issues of bias learning, we propose a multi-instance curriculum learning method that collaboratively incorporates hard negative instance mining and positive instance augmentation to improve classification performance of the model. Specifically, we first initialize the MIL model using easy-to-classify instances, then we mine the hard negative instances (hard-to-classify instances) and augment the positive instances via the diffusion model. Finally, the MIL model is retrained with memory rehearsal method by combining the mined negative instances and the augmented positive instances. Technically, the diffusion model is first designed to generate lesion instances, which optimally augment diverse features to reflect realistic positive samples with post screening scenario. Extensive experimental results show that the proposed method alleviates model bias in MIL and improves model interpretability.