文献速递腹部技术与安全MR

减少肝脏MRI屏气时间:深度学习加速增强后T1 VIBE序列的临床性能

Reducing breath-hold time in liver MRI: Clinical performance of deep learning-accelerated post-contrast T1 VIBE sequences.
PubMed 文献 · Eur J Radiol · 2026-03-25

前瞻性研究纳入99例患者,在1.5 T系统上比较标准18秒与深度学习加速10秒、6秒增强后静脉期T1加权两点Dixon梯度回波序列。标准序列在解剖可分辨性和病灶清晰度上评分更高,但加速序列所有项目均未劣于标准序列,屏气时间最多缩短67%。

要点

  • 深度学习加速序列屏气时间10秒和6秒,较标准18秒最多缩短67%,运动伪影评分无差异。
  • 标准序列解剖可分辨性评分5 vs 加速序列4和4(均p<0.001),病灶清晰度5 vs 4和4(均p<0.001)。
  • 诊断可接受性(Likert≥3)非劣效检验预设界值-5%,两个加速序列所有项目均未劣于标准序列。

核心结论

  • 基于深度学习的重建使增强后肝脏MRI显著加速,屏气时间最多缩短67%;尽管图像质量有轻微折衷,诊断价值仍可接受,支持运动稳健、更快的增强后肝脏MRI。

研究设计与样本

  • 前瞻性研究,纳入99例患者(平均年龄61.0±15.4岁,女性49.5%),在1.5 T系统上采集三种增强后静脉期T1加权两点Dixon梯度回波序列:标准序列(采集时间18秒)和两个深度学习加速序列(10秒和6秒);三名盲法放射科医师采用5点Likert量表评分,以读者间中位数定义每患者共识,采用Friedman检验和Holm校正Wilcoxon符号秩检验,诊断可接受性(Likert≥3)非劣效界值预设为-5%。

诊断标准与关键数值

  • 诊断可接受性定义为Likert评分≥3,与标准序列比较的非劣效界值为-5%;标准序列解剖可分辨性中位评分5 vs 加速序列4和4(均p<0.001),病灶清晰度5 vs 4和4(均p<0.001),整体图像质量中位数均为4 vs 4和4(p<0.001)。

临床应用与阅片要点

  • 深度学习加速增强后肝脏MRI序列(10秒和6秒)在运动伪影评分上与标准18秒序列无差异,诊断可接受性未劣于标准序列,但解剖可分辨性和病灶清晰度评分略低。

局限性

  • 原文未报告局限性。

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

BACKGROUND: Post-contrast liver MRI often requires long breath-holds, risking motion artifacts that can reduce diagnostic quality. We assessed whether deep learning-accelerated acquisitions can shorten breath-holds while maintaining diagnostic image quality.

METHODS: In this prospective study, patients underwent three post-contrast venous-phase T1-weighted two-point Dixon gradient-echo sequences on a 1.5 T system: a standard (18 s acquisition time) and two deep learning-accelerated sequences (10 s and 6 s). Three blinded radiologists rated overall image quality, motion artifacts, other artifacts, anatomic differentiability, and lesion conspicuity on 5-point Likert scales. Per-patient consensus was defined as the median across readers. Within-patient differences were tested with the Friedman test and Holm-adjusted Wilcoxon signed-rank tests. Non-inferiority of diagnostic acceptability (defined as Likert ≥ 3) was tested using a prespecified margin of - 5% compared with the standard sequence.

RESULTS: We enrolled 99 patients (mean age 61.0 ± 15.4 years, 49.5% females). The standard sequence had higher ratings for anatomic differentiability (5 vs. 4 and 4), and lesion conspicuity (5 vs. 4 and 4¸ both p < 0.001) and modestly higher overall image quality despite identical medians (4 vs. 4 and 4, p < 0.001). Motion artifact ratings did not differ across sequences. No inferiority was noted for both accelerated sequences for all items.

CONCLUSION: Deep learning-based reconstruction enabled substantial acceleration of post-contrast liver MRI, reducing breath-hold time by up to 67%. Despite minor image quality trade-offs, diagnostic value remained acceptable, supporting motion-robust, faster post-contrast liver MRI.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Rau S, Fink A, Sacalean V, et al. Reducing breath-hold time in liver MRI: Clinical performance of deep learning-accelerated post-contrast T1 VIBE sequences. Eur J Radiol. 2026. PMID: 41903373. doi:10.1016/j.ejrad.2026.112815

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。