Premium
Automated analysis of cortical volume loss predicts seizure outcomes after frontal lobectomy
Author(s) -
Whiting Alexander C.,
MoritaSherman Marcia,
Li Manshi,
Vegh Deborah,
Machado de Campos Brunno,
Cendes Fernando,
Wang Xiaofeng,
Bingaman William,
Jehi Lara E.
Publication year - 2021
Publication title -
epilepsia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.687
H-Index - 191
eISSN - 1528-1167
pISSN - 0013-9580
DOI - 10.1111/epi.16877
Subject(s) - magnetic resonance imaging , multivariate analysis , univariate analysis , epilepsy surgery , epilepsy , laterality , logistic regression , medicine , multivariate statistics , anterior temporal lobectomy , radiology , psychology , audiology , neuroscience , statistics , mathematics
Objective Patients undergoing frontal lobectomy demonstrate lower seizure‐freedom rates than patients undergoing temporal lobectomy and several other resective interventions. We attempted to utilize automated preoperative quantitative analysis of focal and global cortical volume loss to develop predictive volumetric indicators of seizure outcome after frontal lobectomy. Methods Ninety patients who underwent frontal lobectomy were stratified based on seizure freedom at a mean follow‐up time of 3.5 (standard deviation [SD] 2.5) years. Automated quantitative analysis of cortical volume loss organized by distinct brain region and laterality was performed on preoperative T1‐weighted magnetic resonance imaging (MRI) studies. Univariate statistical analysis was used to select potential predictors of seizure freedom. Backward variable selection and multivariate logistical regression were used to develop models to predict seizure freedom. Results Forty‐eight of 90 (53.3%) patients were seizure‐free at the last follow‐up. Several frontal and extrafrontal brain regions demonstrated statistically significant differences in both volumetric cortical volume loss and volumetric asymmetry between the left and right sides in the seizure‐free and non–seizure‐free cohorts. A final multivariate logistic model utilizing only preoperative quantitative MRI data to predict seizure outcome was developed with a c‐statistic of 0.846. Using both preoperative quantitative MRI data and previously validated clinical predictors of seizure outcomes, we developed a model with a c‐statistic of 0.897. Significance This study demonstrates that preoperative cortical volume loss in both frontal and extrafrontal regions can be predictive of seizure outcome after frontal lobectomy, and models can be developed with excellent predictive capabilities using preoperative MRI data. Automated quantitative MRI analysis can be quickly and reliably performed in patients with frontal lobe epilepsy, and further studies may be developed for integration into preoperative risk stratification.