Prediction of surgical outcomes in MRI-negative refractory temporal lobe epilepsy patients using integrated PET-MRI dynamic regional homogeneity and glucose metabolism
10.3760/cma.j.cn112149-20240120-00036
- VernacularTitle:基于一体化PET-MRI动态局部一致性及葡萄糖代谢预测MRI阴性难治性颞叶癫痫患者手术预后
- Author:
Jie HU
1
;
Jingjuan WANG
1
;
Zhenming WANG
1
;
Bixiao CUI
1
;
Xiaoyin XU
1
;
Hongwei YANG
1
;
Jie LU
1
Author Information
1. 首都医科大学宣武医院放射与核医学科 磁共振成像脑信息学北京市重点实验室，北京 100053
- Publication Type:Journal Article
- Keywords:
Epilepsy, temporal lobe;
Magnetic resonance imaging;
Positron-emission tomography;
Dynamic regional homogeneity;
Engel classification
- From:
Chinese Journal of Radiology
2025;59(2):160-167
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate alterations in static regional homogeneity(ReHo) and dynamic regional homogeneity (dReHo) and glucose metabolism in MRI-negative refractory temporal lobe epilepsy (TLE) patients using resting-state PET-MRI, and to evaluate their efficacy in predicting surgical outcomes.Methods:This study was a cross-sectional design. A retrospective analysis was conducted on the clinical and imaging data of 30 patients with MRI-negative refractory TLE (patient group) treated at Xuanwu Hospital, Capital Medical University, between 2016 and 2020, and data from 30 healthy controls (control group). All MRI-negative refractory TLE patients underwent surgical treatment and were further divided into a good prognosis subgroup (Engel Class I, 16 cases) and a poor prognosis subgroup (Engel Class Ⅱ-Ⅳ, 14 cases) based on postoperative Engel classification. Analysis of variance was used to compare differences in static ReHo, dReHo, and glucose metabolism(SUVR) among the three groups. The correlation of static ReHo, dReHo, and SUVR values of differential brain regions with Engel grading was analyzed using Spearman. A support vector machine (SVM) model was constructed using the static ReHo, dReHo, and SUVR values from these differential regions to classify and predict patient prognosis. The predictive performance was evaluated using receiver operating characteristic curves and the area under the curve (AUC).Results:Differential dReHo regions among the good prognosis subgroup, poor prognosis subgroup, and control group were located in the right lateral middle temporal gyrus temporal pole, the right fusiform gyrus, the right insula subfrontal gyrus, the left cuneate lobe, the right medial and paracortical cingulate gyrus, and the right supraparietal gyrus; the differential static ReHo regions were primarily found in the bilateral inferior temporal gyrus, the supraparietal gyrus, and the right subfrontal gyrus, the left medial supraparietal gyrus, the left median frontal gyrus, and the right marginal supraparietal gyrus; SUVR differences were in the affected superior, middle and inferior temporal lobes, the internal olfactory cortex and the temporal pole region. dReHo of right middle temporal gyrus temporal pole in patients with MRI-negative TLE showed a positive correlation with Engel classification ( r=0.421, P=0.020). The SVM model based on dReHo combined with SUVR values classified patients with good and poor prognosis with an AUC of 0.825 and an accuracy of 73.3%. Conclusions:In MRI-negative refractory TLE patients, abnormal dReHo regions are predominantly located in the contralateral default mode network areas and are associated with Engel classification. Combined with glucose metabolism values, dReHo can predict postoperative outcomes in MRI-negative TLE patients.
