arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.11451cs.CV
收藏

CoMT:医学思维链减少医学报告生成中的幻觉

CoMT: Chain-of-Medical-Thought Reduces Hallucination in Medical Report Generation

Academy for Engineering and Technology, Fudan University(复旦大学工程技术研究院);Tencent YouTu Lab(腾讯优图实验室);Xiamen University(厦门大学);Institute of Meta-Medical, Fudan University(复旦大学元医学研究院);Engineering Research Center of AI and Robotics, Ministry of Education(教育部人工智能与机器人工程研究中心);Jilin Provincial Key Laboratory of Intelligence Science and Engineering(吉林省智能科学与工程重点实验室)

Yue Jiang, Jiawei Chen, Dingkang Yang, Mingcheng Li, Shunli Wang, Tong Wu, Ke Li, Lihua Zhang

AI总结:

提出医学思维链方法CoMT,通过分解诊断过程模仿医生认知,结构化放射学特征为细粒度思维链,以缓解医学报告生成中的幻觉并提升诊断准确性。

摘要:

自动医学报告生成(MRG)具有重要的研究价值,因为它可以辅助放射科医生进行临床诊断和报告撰写,因此受到越来越多的关注。尽管近期取得了进展,但由于需要精确的临床理解和疾病诊断推理,生成准确的报告仍然是一项艰巨的任务。此外,由于医学数据的可及性有限以及疾病分布的不均衡,训练数据中罕见疾病的代表性不足使得大规模医学视觉语言模型(LVLMs)容易产生幻觉,例如遗漏或捏造,这严重削弱了诊断性能,并进一步加剧了MRG在实际应用中的挑战。在本研究中,为了有效缓解医学报告生成中的幻觉问题,我们提出了一种医学思维链方法(CoMT),旨在通过分解诊断过程来模仿人类医生的认知过程。将具有不同重要性的放射学特征结构化为细粒度的医学思维链,以增强诊断过程中的推理能力,从而缓解幻觉问题并提高MRG的诊断准确性。代码和数据集已在https://github.com/FRENKIE-CHIANG/CoMT发布。

英文摘要:

Automatic medical report generation (MRG), which possesses significant research value as it can aid radiologists in clinical diagnosis and report composition, has garnered increasing attention. Despite recent progress, generating accurate reports remains arduous due to the requirement for precise clinical comprehension and disease diagnosis inference. Furthermore, owing to the limited accessibility of medical data and the imbalanced distribution of diseases, the underrepresentation of rare diseases in training data makes large-scale medical visual language models (LVLMs) prone to hallucinations, such as omissions or fabrications, severely undermining diagnostic performance and further intensifying the challenges for MRG in practice. In this study, to effectively mitigate hallucinations in medical report generation, we propose a chain-of-medical-thought approach (CoMT), which intends to imitate the cognitive process of human doctors by decomposing diagnostic procedures. The radiological features with different importance are structured into fine-grained medical thought chains to enhance the inferential ability during diagnosis, thereby alleviating hallucination problems and enhancing the diagnostic accuracy of MRG. The code and dataset have been released at https://github.com/FRENKIE-CHIANG/CoMT.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑