A Self-training Framework for Automated Medical Report Generation

Abstract

Medical report generation, focusing on automatically generating accurate clinical findings from medical images, is an important medical artificial intelligence task. It reduces the workload of physicians in writing reports. Many of the current methods depend heavily on labeled datasets that include image-report pairs, but such datasets labeled by physicians are hard to acquire in clinical practice. In this paper, we introduce a self-training framework named REMOTE (i.e., Revisiting sElf-training for Medical repOrT gEneration) to exploit the unlabeled medical images and a MedCLIPScore to augment a small-scale dataset for training the medical report generation model. Experiments conducted on the MIMIC-CXR benchmark dataset and a COVID-19 dataset demonstrate that, our REMOTE framework, using only 1% labeled training data, achieves competitive performance with previous methods that are trained on entire training data.