摘要目的:探讨三维动脉自旋标记（three-dimensional arterial spin labeling, 3D-ASL）技术对孤立性眩晕患者后循环缺血（posterior circulation ischemia, PCI）和后循环卒中（posterior circulation stroke, PCS）的预测价值。方法:回顾性纳入2022年1月1日至2024年12月31日期间在南京大学医学院附属鼓楼医院接受3D-ASL成像的孤立性眩晕患者。根据影像学检查结果，将孤立性眩晕患者分为PCI组和非PCI组；PCI组进一步分为PCS组和非PCS组。收集基线临床资料和实验室检查数据。通过3D-ASL相关参数获取后循环不同脑区脑血流量（cerebral blood flow, CBF），包括2个不同时相（1.5 s和2.5 s）标记后延迟时间（post labeling delay time, PLD）、延迟灌注CBF（ΔCBF）、多序列PLD（Multi-PLD）及动脉通过时间（arterial transit time, ATT）下的CBF，用于评估后循环灌注。应用多变量 logistic回归分析确定不同CBF值与孤立性眩晕患者PCI和PCS的相关性。应用受试者工作特征（receiver operating characteristic, ROC）曲线评估不同CBF值对PCI和PCS的预测价值。 结果:共纳入81例孤立性眩晕患者，年龄（63.0±12.1）岁，男性44例（54.3%）；58例（71.6%）为PCI，27例（25.9%）为PCS。多变量 logistic回归分析显示，PLD 1.5 s-CBF[优势比（odds ratio, OR）1.372，95%置信区间（confidence interval, CI）1.169~1.611； P<0.001]、ΔCBF（ OR 1.197，95% CI 1.072~1.336； P=0.001）和Multi-PLD-CBF（ OR 2.099，95% CI 1.257~3.504； P=0.005）为PCI的独立预测因素。ROC曲线分析显示，上述3个参数单独以及联合预测PCI的曲线下面积分别为0.962（95% CI 0.915~1.000）、0.683（95% CI 0.543~0.823）、0.944（95% CI 0.985~1.000）和0.999（95% CI 0.997~1.000）。多变量 logistic回归分析显示，小脑区域PLD 1.5 s-CBF（ OR 1.246，95% CI 1.030~2.089； P=0.002）、ΔCBF（ OR 1.153，95% CI 1.038~1.281； P=0.008）和Multi-PLD-CBF（ OR 1.388，95% CI 1.219~1.689； P=0.001）是PCS的独立预测因素。ROC曲线分析显示，上述3个参数单独及联合预测PCS的曲线下面积分别为0.956（95% CI 0.911~1.000）、0.802（95% CI 0.685~0.920）、0.972（95% CI 0.923~1.000）和0.977（95% CI 0.937~1.000）。 结论:3D-ASL可早期预测PCI和PCS，多个参数联合可提高对PCI和PCS的预测能力。

abstractsObjective:To investigate the predictive value of three-dimensional arterial spin labeling (3D-ASL) for posterior circulation ischemia (PCI) and posterior circulation stroke (PCS) in patients with isolated vertigo.Methods:Patients with isolated vertigo underwent 3D-ASL imaging at Drum Tower Hospital Affiliated to Nanjing University School of Medicine from January 1, 2022 to December 31, 2024 were included retrospectively. According to the imaging findings, the patients with isolated vertigo were divided into PCI group and non-PCI group. The PCI group was further divided into PCS group and non-PCS group. The baseline clinical data and laboratory examination data were collected. Cerebral blood flow (CBF) in different brain regions of the posterior circulation was obtained through 3D-ASL related parameters to evaluate the posterior circulation perfusion, including CBF at two post-labeling delay times (PLD) (1.5 s and 2.5 s), delayed perfusion CBF (ΔCBF), multisequence PLD (Multi-PLD) CBF, and CBF under arterial transit time (ATT). Multivariate logistic regression analysis was used to determine the association of different CBF values with PCI and PCS in patients with isolated vertigo. Receiver operating characteristic (ROC) curve was used to evaluate the predictive value of different CBF values for PCI and PCS. Results:A total of 81 patients with isolated vertigo were included, aged 63.0±12.1 years, 44 were males (54.3%); 58 (71.6%) had PCI and 27 (25.9%) had PCS. Multivariate logistic regression analysis showed that PLD 1.5 s-CBF (odds ratio [ OR] 1.372, 95% confidence interval [ CI] 1.169-1.611; P<0.001), ΔCBF ( OR 1.197, 95% CI 1.072-1.336; P=0.001), and Multi-PLD-CBF ( OR 2.099, 95% CI 1.257-3.504; P=0.005) were the independent predictive factors of PCI. ROC curve analysis showed that the area under the curve for predicting PCI using the above three parameters alone and in combination were 0.962 (95% CI 0.915-1.000), 0.683 (95% CI 0.543-0.823), 0.944 (95% CI 0.985-1.000), and 0.999 (95% CI 0.997-1.000), respectively. Multivariate logistic regression analysis showed that the PLD 1.5 s-CBF ( OR 1.246, 95% CI 1.030-2.089; P=0.002), ΔCBF ( OR 1.153, 95% CI 1.038-1.281; P=0.008), and multi-PLD-CBF ( OR 1.388, 95% CI 1.219-1.689; P=0.001) in cerebellar region were the independent predictors of PCS. ROC curve analysis showed that the area under the curve for predicting PCS using the above three parameters alone and in combination were 0.956 (95% CI 0.911-1.00), 0.802 (95% CI 0.685-0.920), 0.972 (95% CI 0.923-1.000), and 0.977 (95% CI 0.937-1.00), respectively. Conclusion:3D-ASL can predict PCI and PCS early, and combining multiple parameters can improve the predictive ability for PCI and PCS.