Learning with Counterfactual Explanations for Radiology Report Generation
Lists
Only you see this
Abstract
Due to the common content of anatomy, radiology images with their corresponding reports exhibit highly similarity. Such inherent data bias can predispose automatic report generation models to learn entangled and spurious representations resulting in misdiagnostic reports. Moreover, the lack of explainability hinders the acceptance by radiologists in clinical practice. To tackle these, we propose a novel \textbf{Co}unter\textbf{F}actual \textbf{E}xplanations-based framework (CoFE) for radiology report generation. Counterfactual explanations serve as a potent tool for understanding how decisions made by algorithms can be changed by asking ``what if'' scenarios. By leveraging this concept, CoFE can learn non-spurious visual representations by contrasting the representations between factual and counterfactual images. Specifically, we derive counterfactual images by swapping a patch between positive and negative samples until a predicted diagnosis shift occurs. Here, positive and negative samples are the most semantically similar but have different diagnosis labels. Additionally, CoFE employs a learnable prompt to efficiently fine-tune the pretrained large language model, encapsulating both factual and counterfactual content to provide a more generalizable prompt representation. Extensive experiments on two benchmarks demonstrate that leveraging the counterfactual explanations enables CoFE to generate semantically coherent and factually complete reports and outperform in terms of language generation and clinical efficacy metrics.
BibTeX
@misc{li2024learning,
author = {Mingjie Li and Haokun Lin and Xiaodan Liang and Ling Chen and Abdulmotaleb El Saddik and Xiaojun Chang},
title = {{Learning with Counterfactual Explanations for Radiology Report Generation}},
year = {2024},
url = {https://openreview.net/forum?id=1Htbe2fiQU},
howpublished = {OpenReview},
note = {Submitted to ICLR 2024},
}