基于MRI自动化定量海马体积诊断颞叶癫痫患者海马硬化

目的 评估基于MRI自动化定量海马体积诊断颞叶癫痫患者海马硬化（HS）的效能。方法 回顾性分析经术后病理检查证实的64例存在HS的颞叶癫痫患者，采用AccuBrain软件对头部3D T1WI进行自动化定量分析，检测海马体积指数（HVI）；以病理结果为诊断金标准，统计2名医师目测诊断HS结果。采用受试者工作特征（ROC）曲线获得HVI诊断HS的截断值及曲线下面积（AUC），计算HVI诊断HS的敏感度、特异度和准确率，评价2种诊断方式所获结果的一致性。结果 64例中，38例左侧HS、26例右侧HS。医师目测诊断HS的敏感度、特异度、准确率分别为90.63%（58/64）、100%（64/64）及95.31%（122/128）。ROC曲线结果显示，HVI诊断HS的最佳截断值为0.185，AUC为0.936。HVI诊断HS的敏感度、特异度、准确率分别为87.50%（56/64）、93.75%（60/64）及90.63%（116/128）。医师目测与HVI诊断HS的准确率一致性较强（Kappa=0.684，P<0.05）。结论 MRI自动化定量海马体积可用于诊断颞叶癫痫患者HS，其诊断效能较高。

MRI automated quantification of hippocampal volume for diagnosis of hippocampal sclerosis in temporal lobe epilepsy

Objective To evaluate the efficacy of MRI automated quantification of hippocampal volume for diagnosis of hippocampal sclerosis (HS) in patients with temporal lobe epilepsy. Methods Sixty-four temporal lobe epilepsy patients with HS confirmed by postoperative pathology were retrospectively enrolled. AccuBrain software was used for automatic quantitative analysis of head 3D T1WI, and hippocampal volume index (HVI) was calculated. Taken pathological results as the gold standards, the visual diagnostic results of 2 physicians HS were recorded. Receiver operating characteristic (ROC) curve was used to obtain the cutoff value and area under the curve (AUC) of HVI for diagnosing HS. The consistency between visual diagnosis and HVI diagnosis of HS was analyzed. Results Among 64 patients with 128 hippocampus, left HS was detected in 38 cases and right HS in 26 cases. The sensitivity, specificity and accuracy of visual diagnosis of HS was 90.63% (58/64), 100% (64/64) and 95.31% (122/128), of HVI was 87.50% (56/64), 93.75% (60/64) and 90.63% (116/128), respectively. ROC curve showed that the optimal cutoff value for HVI diagnosis of HS was 0.185, with AUC of 0.936. There was a relatively strong consistency between visual evaluation and HVI in diagnosing HS (Kappa=0.684, P<0.05). Conclusion Automatic quantification of hippocampal volume based on MRI could be used to diagnose HS in patients with temporal lobe epilepsy, which had good diagnostic efficiency.