文献速递技术与安全神经乳腺腹部MR

基于两阶段深度学习的高保真多b值DWI生成与ADC还原:多中心研究

Accurate multi-b-value DWI generation using two-stage deep learning: multicenter study.
PubMed 文献 · Eur J Radiol · 2025-10-24

该回顾性研究基于三家医院及TCIA数据库共5万张DWI图像(覆盖脑、乳腺、腹部、颈部、盆腔,b值0-1000 s/mm²),构建并验证两阶段深度学习框架DC2Anet-MineGAN用于多器官多b值DWI合成与ADC还原。合成ADC与真实值平均差异小于0.02且无统计学显著差异,所有b值SSIM大于0.81、PSNR大于74;非训练b值(75、850)分别有75%和50%图像被放射科医生评为优秀,ICC均超过0.92。作者指出存在潜在幻觉或失真可能,需进一步多中心临床验证。

要点

  • 合成ADC与真实ADC平均差异小于0.02且无统计学显著差异,各区域图像质量均未见劣于真实图像。
  • 所有b值合成图像SSIM大于0.81、PSNR大于74,结构保真度与信号保真度较高。
  • 非训练b值75时120张中90张(75%)评为优秀,b值850时120张中60张(50%)评为优秀。

核心结论

  • 两阶段DC2Anet-MineGAN框架可在多b值、多解剖区域实现准确高质量的DWI合成与ADC还原,克服临床DWI采集的局限并支持可靠定量成像。
  • 作者同时指出合成图像存在潜在幻觉或失真可能,需进一步多中心临床验证。

研究设计与样本

  • 回顾性研究,纳入三家医院2020年3月至2025年3月采集的DWI图像及TCIA数据库图像,共50,000张。
  • 数据覆盖五个解剖区域(脑、乳腺、腹部、颈部、盆腔)及b值0-1000 s/mm²,按8:2分为训练集与测试集。
  • 两阶段模型以DC2Anet按器官进行低b值到高b值合成,以MineGAN插值至任意b值。
  • 采用MSE、MAE、PSNR、SSIM及放射科医生Likert评分与ICC评估性能,并对非训练b值合成图像进行组间统计学比较。

诊断标准与关键数值

  • 合成ADC与真实值各区域平均差异小于0.02,所有p大于0.05。
  • 所有b值合成图像SSIM大于0.81、PSNR大于74。
  • 非训练b值b=75时120张中90张(75%)被放射科医生评为优秀,b=850时120张中60张(50%)评为优秀。
  • 两个非训练b值的ICC均超过0.92。
  • 合成ADC与真实ADC间无显著差异,未见任何区域图像质量劣于真实图像。

局限性

  • 原文指出合成图像存在潜在幻觉或失真可能。
  • 原文指出需进一步多中心临床验证。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

OBJECTIVES: This study aimed to develop and validate a two-stage deep learning framework (DC2Anet-MineGAN) for high-fidelity multi-organ, multi-b-value DWI synthesis and accurate ADC restoration, addressing real-world limitations of DWI acquisition.

MATERIALS AND METHODS: This retrospective study included DWI images collected from three hospitals (from March 2020 to March 2025) and the TCIA database, with a total of 50,000 images across five anatomical regions (brain, breast, abdomen, neck, pelvis) and b-values (0-1000 s/mm2). Images were split 8:2 into training and test sets. The two-stage model used DC2Anet for low-to-high b-value synthesis per organ and MineGAN for interpolation to arbitrary b-values. Performance was evaluated using MSE, MAE, PSNR, and SSIM, and radiologist Likert ratings with ICC for synthetic images at non-trained b-values. Group differences were assessed statistically.

RESULTS: Of 50,000 images, synthetic ADC values closely matched ground truth across all regions (mean difference < 0.02; all p > 0.05), with SSIM > 0.81 and PSNR > 74 for all b-values. For unseen b-values (b = 75, 850), 90/120 (75 %) and 60/120 (50 %) images were scored as excellent by radiologists; ICC exceeded 0.92 for both b-values. No significant difference was observed between synthetic and real ADC, and no regions showed inferior image quality. Recognizing potential hallucination or distortion, further multi-center clinical validation is required.

CONCLUSION: The proposed two-stage DC2Anet-MineGAN framework achieves accurate, high-quality DWI synthesis and ADC reproduction across multiple b-values and anatomical regions, overcoming clinical DWI limitations and supporting reliable quantitative imaging.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Xia L, Qi X, Liu J, et al. Accurate multi-b-value DWI generation using two-stage deep learning: multicenter study. Eur J Radiol. 2025. PMID: 41161267. doi:10.1016/j.ejrad.2025.112497

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。