人工智能辅助压缩感知三维液体衰减反转恢复序列在静脉内耳钆造影中的应用  

Application of three-dimensional fluid-attenuated inversion recovery sequence using artificial intelligenceassisted compressed sensing technique in intravenous gadolinium contrast-enhanced magnetic resonance imaging of inner ear

作　　者：刘锴 王建 蒋怀礼 张澍杰 吴迪 黄新生[3] 曾蒙苏 赵梦龙 LIU Kai;WANG Jian;JIANG Huaili;ZHANG Shujie;WU Di;HUANG Xinsheng;ZENG Mengsu;ZHAO Menglong(Department of Radiology,Zhongshan Hospital,Fudan University,Shanghai Institute of Medical Imaging,Shanghai 200032,China;Department of Radiology,Juxian People’s Hospital,Rizhao 276500,Shandong,China;Otolaryngology Head and Neck Surgery,Zhongshan Hospital,Fudan University,Shanghai 200032,China)

机构地区：[1]复旦大学附属中山医院放射诊断科,上海市影像医学研究所,上海200032 [2]莒县人民医院影像科,日照276500 [3]复旦大学附属中山医院耳鼻咽喉头颈外科,上海200032

出　　处：《中国临床医学》2025年第2期212-217,共6页Chinese Journal of Clinical Medicine

摘　　要：目的探讨人工智能辅助压缩感知(artificial intelligence-assisted compressed sensing,ACS)技术应用于静脉内耳钆造影三维液体衰减反转恢复(three-dimensional fluid-attenuated inversion recovery,3D-FLAIR)序列检查的价值。方法前瞻性纳入2024年1月至11月在复旦大学附属中山医院接受静脉内耳钆造影ACS和联合压缩感知(united compressed sensing,uCS)3D-FLAIR序列(超长重复时间16000 ms,扫描时间分别为6 min 40 s、10 min 24 s)检查的患者。由两位影像医师独立通过两序列图像评价患者内淋巴积水情况。对比两序列的图像质量主观评分,信噪比(signal-tonoise ratio,SNR)、对比噪声比(contrast-to-noise ratio,CNR)。分析两序列及评价者间评级结果一致性。结果两种序列图像质量主观评分差异无统计学意义。ACS 3D-FLAIR的SNR和CNR高于uCS 3D-FLAIR(P<0.001)。两序列耳蜗、前庭内淋巴积水分级的kappa值分别为0.942、0.888(P<0.001)。两名影像医师使用ACS 3D-FLAIR对耳蜗和前庭内淋巴积水分级的kappa值分别为0.784和0.831(P<0.001),使用uCS 3D-FLAIR对耳蜗和前庭内淋巴积水分级的kappa值分别为0.725和0.756(P<0.001)。结论ACS 3D-FLAIR较uCS 3D-FLAIR可在更短的扫描时间内获得更高的SNR和CNR,更适用于静脉内耳钆造影检查;使用ACS 3D-FLAIR对内淋巴积水分级结果与使用uCS 3D-FLAIR相似。Objective To investigate the value of artificial intelligence-assisted compressed sensing(ACS)technology for intravenous gadolinium contrast-enhanced magnetic resonance imaging of the inner ear using three-dimensional fluid-attenuated inversion recovery(3D-FLAIR)sequence.Methods The patients received gadolinium contrast-enhanced magnetic resonance imaging using ACS and united compressed sensing(uCS)3D-FLAIR at Zhongshan Hospital,Fudan University from January to November 2024 were prospectively enrolled.The repetition time was 16000 ms,and acquisition time was 6 min 40 s and 10 min 24 s in ACS 3D-FLAIR and uCS 3D-FLAIR,respectively.The images on the two sequences were evaluated independently by two radiologists.The image quality of the two sequences was subjectively evaluated and compared.The signal-to-noise ratio(SNR)and contrast-to-noise ratio(CNR)were compared between the two sequences.The grading consistencies using two sequences and between the two doctors were analyzed.Results There was no statistically difference in subjective score of image quality between the two sequences.SNR and CNR of the ACS 3D-FLAIR sequence were significantly higher than those of the uCS 3D-FLAIR sequence(P<0.001).The kappa values of grades of cochlear and vestibular endolymphatic hydrops were 0.942 and 0.888 using two sequences(P<0.001).The kappa values of grades of cochlear and vestibular endolymphatic hydrops using the ACS 3D-FLAIR sequence between the two doctors were 0.784 and 0.831,respectively(P<0.001);the kappa values of grades of cochlear and vestibular endolymphatic hydrops using uCS 3D-FLAIR sequence between the two doctors were 0.725 and 0.756,respectively(P<0.001).Conclusions ACS 3D-FLAIR could provide higher SNR and CNR than uCS 3D-FLAIR,and is more suitable for intravenous gadolinium contrast-enhanced magnetic resonance imaging of the inner ear;the endolymphatic hydrops grades using ACS 3D-FLAIR is similar to use uCS 3D-FLAIR.

关 键 词：内淋巴积水 3D-FLAIR 人工智能辅助压缩感知 联合压缩感知 扫描时间 

分 类 号：R764.3[医药卫生—耳鼻咽喉科]