arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.10410cs.AI

CWCD:用于结构化医学报告生成的类别级对比解码

CWCD: Category-Wise Contrastive Decoding for Structured Medical Report Generation

The Department of Computer Science and Engineering(计算机科学与工程系);University at Buffalo, The State University of New York, NY, USA(布法罗大学,纽约州立大学)

Shantam Srivastava, Mahesh Bhosale, David Doermann, Mingchen Gao

AI总结:

本文提出CWCD框架,通过类别特定参数化和对比正常与遮蔽X光片生成结构化放射报告,提升临床效果和生成质量。

摘要:

解读胸部X光片具有挑战性,因为解剖结构和许多临床重要病变的细微表现容易重叠,使准确诊断对经验丰富的放射科医生来说也耗时。最近专注于放射学的基础模型,如LLaVA-Rad和Maira-2,将多模态大语言模型(MLLMs)置于自动化放射报告生成(RRG)的前沿。然而,尽管有这些进展,当前基础模型在生成报告时仍采用单次前向传递策略。这种解码策略减少了对视觉token的关注,并在生成过程中增加了对语言先验的依赖,从而在生成的报告中引入了虚假的病变共现。为缓解这些限制,我们提出了类别级对比解码(CWCD),一种新的模块化框架,旨在增强结构化放射报告生成(SRRG)。我们的方法引入了类别特定的参数化,并通过使用类别特定的视觉提示对比正常X光片与遮蔽X光片来生成类别级报告。实验结果表明,CWCD在临床效果和自然语言生成指标上均优于基线方法。进一步的消融研究进一步阐明了每个架构组件对整体性能的贡献。

英文摘要:

Interpreting chest X-rays is inherently challenging due to the overlap between anatomical structures and the subtle presentation of many clinically significant pathologies, making accurate diagnosis time-consuming even for experienced radiologists. Recent radiology-focused foundation models, such as LLaVA-Rad and Maira-2, have positioned multi-modal large language models (MLLMs) at the forefront of automated radiology report generation (RRG). However, despite these advances, current foundation models generate reports in a single forward pass. This decoding strategy diminishes attention to visual tokens and increases reliance on language priors as generation proceeds, which in turn introduces spurious pathology co-occurrences in the generated reports. To mitigate these limitations, we propose Category-Wise Contrastive Decoding (CWCD), a novel and modular framework designed to enhance structured radiology report generation (SRRG). Our approach introduces category-specific parameterization and generates category-wise reports by contrasting normal X-rays with masked X-rays using category-specific visual prompts. Experimental results demonstrate that CWCD consistently outperforms baseline methods across both clinical efficacy and natural language generation metrics. An ablation study further elucidates the contribution of each architectural component to overall performance.

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。

↑