Objectives: To assess the clinical value of voxel-based automatic quantitative analysis using a normal brain glucose metabolism database in the preoperative localization of focal intractable temporal lobe epilepsy patients.
Methods: Patients with refractory temporal lobe epilepsy who underwent 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) imaging were retrospectively enrolled from January to June 2017. Visual analysis was performed by two nuclear medicine radiologists, and the automatic quantitative analysis was carried out using MIMneuro software based the age- and gender-stratified normal brain glucose metabolism database. Setting postoperative outcomes as reference, the consistency between visual analysis and automatic quantitative analysis was tested by Cohen's kappa coefficient, and differences in localization of epileptic foci of the two methods were compared by Chi-square test.
Results: A total of 32 patients intractable temporal lobe epilepsy were included in this study. There was a moderate agreement between the automatic quantitative analysis based on MIMneuro software and visual analysis (kappa coefficient = 0.472, p = 0.002). In terms of the efficiency of focus localization, the voxel-based automatic quantitative analysis was higher than that of visual analysis (Chi-square value = 6.969, p = 0.008).
Conclusions: The voxel-based automatic quantitative analysis combined with normal brain glucose metabolism database had a certain clinical application value for detection temporal lobe epilepsy.
Keywords: 18F-FDG PET; Epilepsy; localization; quantitative analysis.