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 2604.09450cs.LGcs.AIeess.IV
收藏

ECHO: 通过一步块扩散实现高效的胸部X光报告生成

ECHO: Efficient Chest X-ray Report Generation with One-step Block Diffusion

Beijing Jiaotong University(北京交通大学);Dalian University of Technology(大连理工大学)

Lifeng Chen, Tianqi You, Hao Liu, Zhimin Bao, Jile Jiao, Xiao Han, Zhicai Ou, Tao Sun, Xiaofeng Mou, Xiaojie Jin, Yi Xu

AI总结:

本文提出ECHO,一种基于扩散模型的高效视觉-语言模型,用于生成胸部X光报告,通过一步块扩散和响应不对称扩散策略,显著提高了生成效率和文本连贯性,同时在临床准确性上保持良好表现。

摘要:

胸部X光报告生成(CXR-RG)有潜力显著减轻放射科医生的工作负担。然而,传统自回归视觉-语言模型(VLMs)由于序列令牌解码而存在高推理延迟。基于扩散的模型通过并行生成提供了一种有前景的替代方案,但它们仍然需要多个去噪迭代。将多步去噪压缩到单步可以进一步减少延迟,但通常会因令牌因子化去噪器引入的均场偏差而降级文本连贯性。为了解决这一挑战,我们提出了ECHO,一种高效的基于扩散的VLM(dVLM),用于胸部X光报告生成。ECHO通过一种新颖的直接条件蒸馏(DCD)框架实现了稳定的每块一步推理,该框架通过从策略扩散轨迹中构建非因子化监督来缓解均场限制,以编码联合令牌依赖性。此外,我们引入了一种响应不对称扩散(RAD)训练策略,该策略进一步提高了训练效率,同时保持模型有效性。广泛的实验表明,ECHO超越了最先进的自回归方法,在RaTE和SemScore上分别提高了64.33%和60.58%,同时在临床准确性上几乎没有下降的情况下,实现了高达8倍的推理加速。

英文摘要:

Chest X-ray report generation (CXR-RG) has the potential to substantially alleviate radiologists' workload. However, conventional autoregressive vision--language models (VLMs) suffer from high inference latency due to sequential token decoding. Diffusion-based models offer a promising alternative through parallel generation, but they still require multiple denoising iterations. Compressing multi-step denoising to a single step could further reduce latency, but often degrades textual coherence due to the mean-field bias introduced by token-factorized denoisers. To address this challenge, we propose \textbf{ECHO}, an efficient diffusion-based VLM (dVLM) for chest X-ray report generation. ECHO enables stable one-step-per-block inference via a novel Direct Conditional Distillation (DCD) framework, which mitigates the mean-field limitation by constructing unfactorized supervision from on-policy diffusion trajectories to encode joint token dependencies. In addition, we introduce a Response-Asymmetric Diffusion (RAD) training strategy that further improves training efficiency while maintaining model effectiveness. Extensive experiments demonstrate that ECHO surpasses state-of-the-art autoregressive methods, improving RaTE and SemScore by \textbf{64.33\%} and \textbf{60.58\%} respectively, while achieving up to \textbf{$8\times$} inference speedup with negligible degradation in clinical accuracy.

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。

↑