深度学习加速3D FLAIR在多发性硬化白质病变检测中的可行性研究
本研究前瞻性纳入26例多发性硬化患者,在3T扫描仪上比较深度学习重建3D FLAIR与常规3D FLAIR。结果显示深度学习重建序列在图像质量、病变敏感性及信噪比方面均显著优于常规序列,且采集时间缩短32%。结论支持深度学习重建3D FLAIR临床应用潜力。
要点
- 深度学习重建3D FLAIR可检出更多<3 mm小病变且无假阳性
- 图像噪声和伪影更少,诊断信心更优,采集时间缩短32%
- 适用于多发性硬化白质病变的精细化评估,值得纳入常规方案
核心结论
- 深度学习重建3D FLAIR显著改善多发性硬化病变检出和图像质量,支持其整合入标准成像方案。
- 深度学习重建3D FLAIR对<3 mm病变的敏感性为99.5%,高于常规3D FLAIR的88.8%。
- 深度学习重建3D FLAIR对≥3 mm病变的敏感性为100%,高于常规3D FLAIR的97.9%。
研究设计与样本
- 前瞻性、阅片者设盲研究,纳入26例多发性硬化患者。
- 同一检查时段内在3T扫描仪上行深度学习重建3D FLAIR与常规3D FLAIR序列扫描。
- 两名神经放射科医生采用4分Likert样量表评价噪声、伪影、锐利度、整体图像质量和诊断信心。
- 病变显著性按<3 mm和≥3 mm分级,定量指标包括病变计数、表观信噪比和对比噪声比。
- 采用复合金标准计算敏感性和精确率。
关键影像表现/征象
- 深度学习重建3D FLAIR显示更低的噪声和伪影,更高的锐利度和整体图像质量。
- 深度学习重建3D FLAIR对<3 mm病变的显著性改善更明显。
- 深度学习重建图像额外检出29个病变,其中10个≥3 mm,19个<3 mm,且无假阳性。
诊断标准与关键数值
- 深度学习重建3D FLAIR对<3 mm病变敏感性为99.5%,常规3D FLAIR为88.8%。
- 深度学习重建3D FLAIR对≥3 mm病变敏感性为100%,常规3D FLAIR为97.9%。
- 深度学习重建3D FLAIR定性评分显著优于常规序列(p<0.001),阅片者间一致性高。
- 深度学习重建图像的表观信噪比和对比噪声比显著高于常规图像(p<0.001)。
- 深度学习重建3D FLAIR采集时间为3:54分钟,较常规5:44分钟缩短32%。
临床应用与阅片要点
- 深度学习重建3D FLAIR可作为多发性硬化白质病变精细化检测的优选序列。
- 阅片时应注意<3 mm小病变的显著性改善可提高早期和微小病灶的识别能力。
- 该技术在提高图像质量的同时缩短采集时间,有助于改善患者耐受性和检查效率。
- 将深度学习重建序列纳入常规方案前需结合本机构设备和后处理软件条件进行评估。
局限性
- 本研究样本量较小,仅26例患者,且为单中心前瞻性设计。
- 原文指出需在更大、更多样化队列中进一步验证深度学习重建算法的临床应用价值。
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: Deep learning (DL)-based image reconstruction (DLBIR) techniques promise accelerated MRI acquisitions with enhanced image quality. Herein, we compare the image quality of a DLBIR-based 3D FLAIR (3D-FLAIRDL) with conventional 3D FLAIR (3D-FLAIRSOC) in a cohort of multiple sclerosis (MS) patients.
MATERIALS AND METHODS: Our prospective, reader-blinded study, included 26 MS patients who underwent both sequences on a 3T scanner during the same study session over three months. Two neuroradiologists assessed noise, artifacts, sharpness, overall image quality, and diagnostic confidence using 4-point Likert-like scales. Lesion conspicuity was graded for lesions < 3 mm and ≥ 3 mm. Quantitative metrics included lesion count, apparent signal-to-noise ratio (aSNR), and contrast-to-noise ratio (aCNR). A composite gold standard was used to calculate sensitivity and precision.
RESULTS: 3D-FLAIRDL showed significantly better qualitative scores (p < 0.001) with high inter-reader agreement. Lesion conspicuity and diagnostic confidence improved, especially for < 3 mm lesions. DLBIR images detected 29 additional lesions (10 ≥ 3 mm-sized lesions and 19 < 3 mm lesions), yielding higher sensitivity (99.5% vs. 88.8% for < 3 mm; 100% vs. 97.9% for ≥ 3 mm) without false positives. aSNR and aCNR were significantly higher for DLBIR images (p < 0.001). Acquisition time was reduced by 32% (3:54 vs. 5:44 min).
CONCLUSION: DLBIR 3D FLAIR significantly improves lesion detection and image quality in MS, supporting its potential integration into standard imaging protocols. As DLBIR algorithms evolve, further validation in larger, diverse cohorts will be essential.
引用
- Rai, Chan, Nickel, et al. Deep learning-accelerated 3D flair for white matter lesion detection in multiple sclerosis: a feasibility study. Neuroradiology. 2026. PMID: 41569417. doi:10.1007/s00234-025-03889-3
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。