Bilateral Asymmetry Guided Counterfactual Generating Network for Mammogram Classification

Abstract

Mammogram benign or malignant classification with only image-level labels is challenging due to the absence of lesion annotations. Motivated by the symmetric prior that the lesions on one side of breasts rarely appear in the corresponding areas on the other side, given a diseased image, we can explore a counterfactual problem that how would the features have behaved if there were no lesions in the image, so as to identify the lesion areas. We derive a new theoretical result for counterfactual generation based on the symmetric prior. By building a causal model that entails such a prior for bilateral images, we obtain two optimization goals for counterfactual generation, which can be accomplished via our newly proposed counterfactual generative network. Our proposed model is mainly composed of Generator Adversarial Network and a prediction feedback mechanism, they are optimized jointly and prompt each other. Specifically, the former can further improve the classication performance by generating counterfactual features to calculate lesion areas. On the other hand, the latter helps counterfactual generation by the supervision of classification loss. The utility of our method and the effectiveness of each module in our model can be verified by state-of-the-art performance on INBreast and an in-house dataset and ablation studies.

GANterfactual - Counterfactual Explanations for Medical Non-Experts using Generative Adversarial Learning

“…There is existing work that uses those techniques of adversarial image-to-image translation for creating counterfactuals, but often the counterfactuals are not created for the purpose of explaining ML algorithms, but rather to improve those algorithms [28,29]. Zhao et al [30] proposed to use a StarGAN [31] architecture to create counterfactual explanation images.…”