深度学习重建辅助流动敏感黑血序列在豆纹动脉MRI的临床应用  

Deep learning reconstruction-assisted flow sensitive black blood sequences for clinical applications in MRI of lenticulostriate artery

作　　者：姚灵 段庆红 张顺 杨洁 陈振涛 YAO Ling;DUAN Qinghong;ZHANG Shun;YANG Jie;CHEN Zhentao(Department of Radiology,the Affiliated Tumor Hospital of Guizhou Medical University,Guiyang 550001,China;Canon Medical Systems(China)Co.,Ltd.,Beijing 100027,China)

机构地区：[1]贵州医科大学附属肿瘤医院影像科,贵州贵阳550001 [2]佳能医疗系统(中国)有限公司,北京100027

出　　处：《实用放射学杂志》2024年第7期1048-1051,1069,共5页Journal of Practical Radiology

摘　　要：目的探讨深度学习重建(DLR)用于流动敏感黑血(FSBB)序列在豆纹动脉(LSA)MRI的临床应用价值。方法前瞻性选取30例健康志愿者,扫描常规FSBB序列,根据是否采用DLR方法分为DLR-FSBB组和常规FSBB组,比较2组双侧LSA图像信噪比(SNR)、对比噪声比(CNR)和分支数量,并采用Likert量表(5分制)对整体图像质量进行主观评估。采用配对t检验比较2组图像质量定量指标SNR及CNR。采用Kappa系数评价2位观察者主观评分的一致性。结果颅脑双侧LSA DLR-FSBB图像的SNR、CNR显著高于常规FSBB图像(左侧SNR 34.45±8.21 vs 16.91±4.67,P<0.001,右侧SNR 31.24±6.35 vs 16.02±3.00,P<0.001;对应的CNR为19.38±6.37 vs 9.27±3.12,15.82±4.29 vs 8.01±2.27,P<0.001)。DLR-FSBB图像LSA分支数量稍多[左侧4.00(4.00,5.00)vs 3.00(3.00,4.00),P<0.001;右侧3.00(3.00,4.00)vs 3.00(2.00,3.00),P=0.003]。DLR-FSBB整体图像质量主观评分高于常规FSBB[阅片者1:4.00(4.00,4.00)分vs 3.00(3.00,3.00)分;阅片者2:4.00(4.00,4.00)分vs 3.00(3.00,3.00)分,P<0.001]。结论与常规FSBB序列相比,DLR-FSBB序列LSA图像噪声较小,血管分支显示更清晰且数量稍多。DLR改善了常规FSBB序列LSA整体图像质量。Objective To investigate the clinical value of deep learning reconstruction(DLR)for flow sensitive black blood(FSBB)sequence in MRI of lenticulostriate artery(LSA).Methods Thirty robust volunteers were prospectively selected,performed scans with conventional FSBB sequences,and divided into DLR-FSBB group and conventional FSBB group according to whether DLR method was used.The signal-to-noise ratio(SNR),contrast-to-noise ratio(CNR)and number of branches of the two groups of images for bilateral LSA were compared,and the overall image quality was evaluated subjectively using Likert score(5-point scale).A paired t-test was used to compare quantitative indicators,SNR and CNR of image quality between the two groups.The Kappa coefficient was applied to evaluate the consistency of subjective scores between the two observers.Results The SNR and CNR of DLR-FSBB images of cranial bilateral LSA were considerably higher than those of conventional FSBB images(left SNR 34.45±8.21 vs 16.91±4.67,P<0.001,and right SNR 31.24±6.35 vs 16.02±3.00,P<0.001;corresponding CNR 19.38±6.37 vs 9.27±3.12,15.82±4.29 vs 8.01±2.27,P<0.001).The number of branches of LSA was slightly higher in the DLR-FSBB images[4.00(4.00,5.00)vs 3.00(3.00,4.00)on the left side,P<0.001;3.00(3.00,4.00)vs 3.00(2.00,3.00)on the right side,P=0.003].The subjective score of overall image quality was higher in DLR-FSBB than that in conventional FSBB[reader 1:4.00(4.00,4.00)points vs 3.00(3.00,3.00)points;reader 2:4.00(4.00,4.00)points vs 3.00(3.00,3.00)points,P<0.001].Conclusion Compared with conventional FSBB sequences,DLR-FSBB sequences of LSA have lower image noise,and vessel branches are displayed more plainly and slightly more numerous.DLR can improve the overall image quality of LSA in conventional FSBB sequences.

关 键 词：深度学习重建 流动敏感黑血 磁共振成像 豆纹动脉 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] R445.2[自动化与计算机技术—控制科学与工程] R743[医药卫生—影像医学与核医学]