摘要目的:探讨基于表面的形态测量(SBM)方法在药物难治性癫痫术前评估中的应用价值。方法:回顾性分析2013年11月至2018年7月清华大学玉泉医院癫痫中心收治的51例难治性癫痫患者的临床资料。所有患者行致痫灶的手术切除或热凝损毁治疗，且术后>6个月的Engel分级均为Ⅰ级。使用SBM后处理方法对患者的头颅MRI三维T1加权成像结果进行分析，分析应用SBM方法对癫痫患者的致痫灶整体检出情况以及SBM各特征值[包括大脑的褶皱指数(GI)、沟回深度(SQ)以及皮质厚度(TH)]对于癫痫患者的病灶检出率。结果:根据其影像学特征及手术切除或热凝损毁范围，51例患者可分为磁共振成像阴性(MN)组(13例)、磁共振成像阳性区域与手术部位吻合(PCS)组(27例)、磁共振成像阳性区域包含手术部位(PIS)组(4例)以及磁共振成像阳性区域与手术部位不一致(PNS)组(7例)。其中，PCS组患者又可分为内侧颞叶癫痫型(13例)、皮质发育不良型(5例)以及脑软化或皮质发育畸形型(9例)。应用SBM方法，MN组、PCS组、PIS组以及PNS组的病灶检测阳性比例分别为3/13、21/27、2/4以及1/7。PCS组中，内侧颞叶癫痫型、皮质发育不良型以及脑软化或皮质发育畸形型的检测阳性比例分别为8/13、4/5和9/9。GI对PIS组(4/4)和PCS组(22/27)的检测阳性比例高于SQ(分别为2/4和18/27)和TH(分别为1/4和21/27)；SQ对MN组(5/13)的检测阳性比例高于TH或GI(均为3/13)，TH对PNS组的检测阳性比例(3/7)高于GI和SQ(分别为2/7和1/7)。结论:初步研究表明，针对有明显致痫灶的难治性癫痫患者，采用GI定位不仅检测阳性率高，还能较好地显示致痫灶的边界。而对于无明显致痫灶的难治性癫痫患者，SQ可能是更为敏感的指标。对于MRI显示致痫灶部位与其他临床检查结果不一致的难治性癫痫患者，采用TH进行辅助判断可能更为准确。

abstractsObjective:To investigate the application value of surface-based morphometry (SBM) in preoperative evaluation of refractory epilepsy.Methods:Clinical data of 51 patients with refractory epilepsy admitted to Epilepsy Center, Yuquan Hospital, Tsinghua University from November 2013 to July 2018 were retrospectively analyzed. Fifty-one patients underwent resection or radiofrequency thermo-coagulation of epileptogenic zone with the outcome of Engel Ⅰ assessed at over 6 months post operation. The three-dimensional high-resolution MRI T1-weighed imaging data were analyzed to calculate the proportion of positive results by using three characteristic values of cortical surfaces, i. e. gyrification index (GI), sqrt-transformed sulcus depth (SQ) and thickness (TH), which were obtained by SBM post-procession.Results:According to imaging characteristics and the scope of resection or radiofrequency thermocoagulation of epileptogenic zone, 51 cases were divided into four groups as follows: 13 cases as MRI negative (MN) group, 27 as MRI positive region coinciding with surgical site (PCS) group, 4 as MRI positive region including surgical site (PIS) group, and 7 as MRI positive region noncompatible with surgical site (PNS) group. PCS group was further divided into three types: 13 cases as mesial temporal lobe epilepsy (PCS-MTLE), 5 as focal cortical dysplasia (PCS-FCD), and 9 as encephalomalacia or malformations of cortical development (PCS-EMCD). With all three characteristic values, the lesion positive proportion were 3/13, 21/27, 2/4 and 1/7 for MN group, PCS group, PIS group and PNS group respectively and 8/13, 4/5 and 9/9 for the three types of PCS patients respectively. Lesion positive proportion by using GI was 4/4 in PIS group and 22/27 in PCS group, which were higher than those obtained by SQ(2/4 and 18/27)and TH(1/4 and 21/27). Lesion positive proportion obtained by SQ in MN was 5/13 and that obtained by TH in PNS group was 3/7, while those by GI in MN group and in PNS group were 3/13 and 2/7 respectively.Conclusions:The preliminary study suggests that for patients with refractory epilepsy who have distinctive lesions, GI could be used to locate the lesions with a relatively high positive proportion and good display on the margins of lesions. For MRI-negative patients, SQ may be a relatively sensitive target for lesion detection. For patients with incongruent indications from MRI and other clinical measures, TH may be the optimal target for lesion detection.