磁共振胰胆管成像图像质量最佳压缩感知加速因子探讨

摘要/Abstract

摘要： 目的探讨磁共振胰胆管成像(MRCP)图像质量最佳压缩感知(CS)加速因子。方法对2021年10月至2023年10月间于我院行MRCP检查的60例患者行回顾性分析,分别行常规3D-MRCP、CS-3D-MRCP(CS18、CS24、CS32)扫描,使用Philips图像后处理工作站进行客观图像分析,由2名经验丰富、高年资影像科医师按照5分法进行主观图像分析。应用Kappa检验分析图像质量主观评分结果一致性,比较4组胰胆管显示评分、图像锐利度、胰胆管显示情况和扫描时间。结果 CS32组SD、SNR评分为(38.4±3.0)、(40.4±2.8),均分别显著低于SENSE4[(44.3±3.5)、(43.9±3.2)]、CS18[(43.2±4.2)、(44.5±4.0)]、CS24[(43.0±3.1)、(46.2±5.2)]组;CNR评分为(8.2±1.3),显著高于SENSE4(7.4±1.4)、CS18(7.7±1.3)、CS24(7.8±1.0),差异有统计学意义(P<0.05);而后三组SD、SNR、CNR评分比较差异无统计学意义(P>0.05)。2名观察者对SENSE4、CS18、CS24、CS32主观评分一致性较好(Kappa=0.792、0.719、0.800、0.734,P<0.05)。CS32组图像锐利度、胆总管、肝总管、胆囊管、左肝管及右肝管显示评分为(3.01±0.70)、(3.40±0.45)、(3.30±0.53)、(3.15±0.53)、(2.76±0.65)、(2.85±0.60),均分别显著低于SENSE4[(3.75±0.42)、(3.80±0.35)、(3.90±0.21)、(3.75±0.43)、(3.55±0.47)、(3.50±0.47)]、CS18[(3.70±0.48)、(3.95±0.15)、(3.85±0.33)、(3.80±0.43)、(3.45±0.49)、(3.55±0.45)]、CS24[(3.65±0.44)、(3.76±0.35)、(3.85±0.35)、(3.60±0.45)、(3.45±0.48)、(3.35±0.41)]组,差异有统计学意义(P<0.05);而后三组比较差异无统计学意义(P>0.05)。SENSE4、CS18、CS24及CS32组检查扫描时间分别为186 s、23 s、14 s及12 s。结论以CS技术采集MRCP图像信息可在保证成像质量的前提下缩短扫描时间,最佳加速因子为CS24。

关键词: 磁共振胰胆管成像, 图像质量, 压缩感知, 最佳加速因子

Abstract: Objective To explore the optimal compression sensing (CS) acceleration factor for the image quality of magnetic resonance cholangiopancreatography (MRCP). Methods 60 patients underwent MRCP between October 2021 and October 2023 were retrospectively analyzed. Conventional 3D-MRCP and CS-3D-MRCP (CS18, CS24, CS32) were performed respectively, and objective image analysis was performed by Philips image post-processing workstation. Subjective image analysis was performed by two senior radiologists according to the five-point method. Kappa test was used to analyze the consistency of subjective evaluation results of image quality, and the display score and image sharpness of pancreaticobiliary duct were compared among the four groups, and the display situation and scanning time of pancreaticobiliary duct were compared among the four groups. Results The SD and SNR scores of CS32 group were (38.4±3.0) and (40.4±2.8), they were significantly lower than SENSE4 [(44.3±3.5), (43.9±3.2)], CS18 [(43.2±4.2), (44.5±4.0)] and CS24 [(43.0±3.1), (46.2±5.2)]. The CNR score of CS32 was (8.2±1.3), which was significantly higher than SENSE4 (7.4±1.4), CS18 (7.7±1.3), CS24 (7.8±1.0), respectively, the difference was statistically significant (P<0.05). There was no statistical difference in SD、SNR、CNR scores among SENSE4、CS18 and CS24 (P>0.05). The subjective scores of SENSE4, CS18, CS24 and CS32 of the two observers are consistent (Kappa=0.792, 0.719, 0.800, 0.734, P<0.05). The image sharpness, common bile duct, common hepatic duct, cystic duct, left hepatic duct and right hepatic duct in CS32 group were (3.01±0.70), (3.40±0.45), (3.30±0.53), (3.15±0.53), (2.76±0.65), (2.85±0.60), they were significantly lower than SENSE4 [(3.75±0.42), (3.80±0.35), (3.90±0.21), (3.75±0.43), (3.55±0.47) and (3.50±0 0.47)], CS18 [(3.70±0.48), (3.95±0.15), (3.85±0.33), (3.80±0.43), (3.45±0.49), (3.55±0.45)], CS24 [(3.65±0.44), (3.76±0.35), (3.85±0.35), (3.60±0.45), (3.45±0.48), (3.35±0.41)], the difference was statistically significant (P<0.05). There was no statistical difference between the last three groups (P>0.05). The scanning time of SENSE4, CS18, CS24 and CS32 groups is 186s, 23s, 14s and 12s, respectively. Conclusion Collecting MRCP image information with CS technology can shorten the scanning time on the premise of ensuring the imaging quality, with CS24 was identified as the best acceleration factor.

Key words: Magnetic resonance cholangiopancreatography, Image quality, Compression sensing, Optimum acceleration factor

王月蓉, 李可可, 景红. 磁共振胰胆管成像图像质量最佳压缩感知加速因子探讨[J]. 肝脏, 2025, 30(8): 1147-1150.

WANG Yue-rong, LI Ke-ke, JING Hong. Study on the optimal compression sensing acceleration factor for MRI image quality[J]. Chinese Hepatolgy, 2025, 30(8): 1147-1150.

/ 推荐

参考文献

[1] Idowu B M, Onigbinde S O, Afolabi B I, et al. Aetiological spectrum of pancreaticobiliary diseases in adult Nigerians on magnetic resonance cholangiopancreatography[J]. Niger J Clin Pract, 2024, 27(5):592-598.
[2] Sun K, Li M, Shi Y, et al. Convolutional neural network for identifying common bile duct stones based on magnetic resonance cholangiopancreatography[J]. Clin Radiol, 2024, 79(7):553-558.
[3] Ke C, Yang T, Huang G, et al. Investigation of the accuracy of magnetic resonance cholangiography and multi-slice spiral computed tomography in the diagnosis of cholangiocarcinoma[J]. J Gastrointest Oncol, 2023, 14(3):1496-1503.
[4] Chen Z, Xue Y, Wu Y, et al. Feasibility of 3D breath-hold MR cholangiopancreatography with a spatially selective radiofrequency excitation pulse: prospective comparison with parallel imaging technique and compressed sensing method[J]. Acad Radiol, 2022, 29(12):e289-e295.
[5] Goo E H, Kim S S. Evaluating the quality of optimal MRCP image using RT-2D-compressed SENSE (CS) turbo spin echo: comparing respiratory triggering (RT)-2D-SENSE turbo spin echo and breath hold-2D-single-shot turbo spin echo[J]. Tomography, 2022, 8(3):1374-1385.
[6] 浦仁旺, 刘爱连, 王家正, 等. 不同压缩感知加速倍数对三维磁共振胰胆管成像图像质量的影响[J]. 实用放射学杂志, 2021, 37(1):128-131.
[7] 王诗瑜, 刘爱连, 王家正, 等. 压缩感知不同加速因子对肝脏3D-mDixon序列成像质量的影响[J]. 放射学实践, 2020, 35(12):1605-1609.
[8] Yuan X, Zhou L, Liu C, et al. Comparative analysis of CT and abdominal X-ray in the diagnosis of gastric perforation[J]. Asian J Surg, 2024, 47(7):3369-3370.
[9] Xiao C H, Liu P, Zhang H H, et al. Incremental diagnostic value of virtual non-contrast dual-energy CT for the diagnosis of choledocholithiasis over conventional unenhanced CT[J]. Diagn Interv Imaging, 2024, 105(7-8):292-298.
[10] Ho P H, Lee Y C, Ng C J, et al. Assessing the clinical utility of abdominal computed tomography in sepsis patients with unknown origin: a retrospective cohort study[J]. Medicine (Baltimore), 2024, 103(20):e38114.
[11] 王益双, 路涛, 张天悦, 等. 3D-SPACE-MRCP与2D-HASTE-MRCP在磁共振胰胆管成像中的应用对比[J]. 临床放射学杂志, 2021,10(7): 1413-1416.
[12] Inès Mannes, Dallongeville A, Badat N, et al. Breath-hold compressed-sensing 3D MR cholangiopancreatography compared to free-breathing 3D MR cholangiopancreatography: prospective study of image quality and diagnostic performance in pancreatic disorders[J]. Abdomi Radio, 2020, 45(4):1082-1091.
[13] Kromrey M L, Funayama S, Tamada D, et al. Clinical evaluation of respiratory-triggered 3D MRCP with navigator echoes compared to breath-hold acquisition using compressed sensing and/or parallel imaging[J]. Mag Reson Med Scien, 2019,19(4): 318-323.
[14] Chevallier O, Escande H, Ambarki K, et al. Single-breath-hold MRI-SPACE cholangiopancreatography with compressed sensing versus conventional respiratory-triggered MRI-SPACE cholangiopancreatography at 3Tesla: comparison of image quality and diagnostic confidence[J]. Diagnostics (Basel), 2021, 11(10):1886.
[15] Blaise H, Remen T, Ambarki K, et al. Comparison of respiratory-triggered 3D MR cholangiopancreatography and breath-hold compressed-sensing 3D MR cholangiopancreatography at 1.5 T and 3 T and impact of individual factors on image quality[J]. Eur J Radiol, 2021, 142:109873.
[16] Chen Z, Sun B, Xue Y, et al. Comparing compressed sensing breath-hold 3D MR cholangiopancreatography with two parallel imaging MRCP strategies in main pancreatic duct and common bile duct[J]. Eur J Radiol, 2021, 142:109833.
[17] Suzuki T, Aonuma T, Oyama K, et al. High-resolution three-dimensional T1-weighted hepatobiliary MR cholangiography using Gd-EOB-DTPA for assessment of biliary tree anatomy: parallel imaging versus compressed sensing[J]. Eur J Radiol, 2021, 136:109515.
[18] Tokoro H, Yamada A, Suzuki T, et al. Usefulness of breath-hold compressed sensing accelerated three-dimensional magnetic resonance cholangiopancreatography (MRCP) added to respiratory-gating conventional MRCP[J]. Eur J Radiol, 2020, 122:108765.
[19] Mannes I, Dallongeville A, Badat N, et al. Breath-hold compressed-sensing 3D MR cholangiopancreatography compared to free-breathing 3D MR cholangiopancreatography: prospective study of image quality and diagnostic performance in pancreatic disorders[J]. Abdom Radiol (NY), 2020, 45(4):1082-1091.