压缩感知用于屏气高分辨率肝胆期成像:与并行成像的图像噪声、伪影、胆道解剖评估及局灶性病变检测比较
回顾性纳入125例同时行压缩感知(CS)高分辨率与标准并行成像肝胆期(HBP)MRI的患者,比较图像质量、胆道解剖诊断信心及局灶性病变检出率。CS HBP空间分辨率更高(1×1.4×1 mm³ vs 1.3×1.8×3 mm³),屏气时间相近(15 s vs 16 s)。结果显示CS HBP肝边缘锐利度、胆管显示及局灶性病变检出率更优,但主观噪声和非呼吸伪影更明显,胆道解剖观察者间一致性更高(κ 0.67 vs 0.45)。
要点
- CS HBP可提高肝边缘锐利度和胆管显示,但图像噪声和非呼吸伪影增加。
- CS HBP胆道解剖诊断信心更高(4.70-4.74 vs 3.96-4.55),观察者间一致性更好(κ 0.67 vs 0.45)。
- CS HBP局灶性病变检出率更高(88.2%和84.5% vs 66.3%和73.4%)。
关键影像表现/征象
- CS HBP肝边缘锐利度和胆管可视化显著优于标准HBP。
- CS HBP主观图像噪声和非呼吸伪影较标准HBP更明显。
- CS HBP胆道解剖诊断信心评分4.70-4.74,高于标准HBP的3.96-4.55(p=0.002)。
诊断标准与关键数值
- 胆道解剖观察者间一致性:CS HBP κ=0.67,标准HBP κ=0.45。
- 胆管开口数目观察者间一致性:CS HBP ICC=0.860,标准HBP ICC=0.579。
- 局灶性病变检出敏感度:CS HBP读者1为88.2%、读者2为84.5%;标准HBP读者1为66.3%、读者2为73.4%。
临床应用与阅片要点
- CS HBP可改善肝边缘和胆管细节显示,适合需精细评估胆道解剖或小病灶的病例。
- 需注意CS HBP噪声和非呼吸伪影增加,阅片时应结合标准HBP或其它序列综合判断。
局限性
- 单中心回顾性设计。
- 样本量有限(125例)。
- 未报告呼吸伪影以外的其他伪影类型细节。
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
PURPOSE: To assess image quality, performance for biliary anatomy diagnosis, and focal lesion detection rate of breath-hold high-resolution 3D T1-weighted hepatobiliary phase imaging using compressed sensing (CS HBP) compared to standard HBP using conventional parallel imaging.
METHODS: This retrospective study assessed consecutive 125 patients who underwent CS HBP and standard HBP between November 2019 and July 2020. Optimized resolution and scan time for CS HBP were 1 × 1.4 × 1 mm3 and 15 s, while those for standard HBP were 1.3 × 1.8 × 3 mm3 and 16 s. Two independent radiologists evaluated qualitative indices on the clarity of liver margin, visibility of the hepatic vessel and bile duct, image noise, and artifact on a 5-point scale. Biliary anatomy, confidence for biliary anatomy diagnosis, expected number of bile duct openings, and number of focal lesions were assessed. Wilcoxon signed-rank test, Pearson chi-square test, and sensitivity for focal lesion were used for statistical analysis. Intraclass correlation coefficient (ICC) and Cohen's kappa (κ) were used to determine inter-observer agreement.
RESULTS: CS HBP showed significantly better liver edge sharpness and bile duct visualization, but greater subjective image noise and non-respiratory artifacts compared to standard HBP. CS HBP showed higher number of concordantly assigned biliary anatomy across readers (86 vs. 80), indicating greater inter-observer agreement for biliary anatomy (κ, 0.67 vs. 0.45) and the number of bile duct openings (ICC, 0.860 vs. 0.579) with significantly higher diagnostic confidence (4.70-4.74 vs. 3.96-4.55; p = 0.002). Both readers identified more focal lesions in CS HBP than in standard HBP (88.2% and 84.5% vs. 66.3% and 73.4%).
CONCLUSION: Breath-hold high-resolution CS HBP was a feasible clinical sequence providing superior liver edge sharpness, bile duct visualization, and focal lesion detection rate compared to standard HBP despite higher noise and artifact. Due to improved spatial resolution, CS HBP yielded a higher inter-observer agreement and confidence for the biliary anatomy diagnosis.
引用
- Choi MH, Kim B, Han D, et al. Compressed sensing for breath-hold high-resolution hepatobiliary phase imaging: image noise, artifact, biliary anatomy evaluation, and focal lesion detection in comparison with parallel imaging. Abdom Radiol (NY). 2021. PMID: 34591152. doi:10.1007/s00261-021-03290-7
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。