Objective 3D surface evaluation of intracranial electrophysiologic correlates of cerebral glucose metabolic abnormalities in children with focal epilepsy

• Published in: Human brain mapping 2017-06, Vol.38 (6), p.3098-3112
• Main Authors: Jeong, Jeong‐Won, Asano, Eishi, Kumar Pilli, Vinod, Nakai, Yasuo, Chugani, Harry T., Juhász, Csaba
• Format: Article
• Language: English
• Subjects: Abnormalities; Adolescent; Brain; Brain - diagnostic imaging; Brain - metabolism; Brain - physiopathology; Brain Mapping; Child; Child, Preschool; Children; Clusters; Cortex; EEG; electrocorticography; Electrodes; Electroencephalography - methods; Epilepsies, Partial - diagnostic imaging; Epilepsies, Partial - physiopathology; Epilepsies, Partial - surgery; Epilepsy; epilepsy surgery; Evoked Potentials - physiology; FDG PET; Female; Firing pattern; Fluorodeoxyglucose F18 - metabolism; Glucose; Glucose - metabolism; Humans; Imaging, Three-Dimensional - methods; Infant; Localization; Magnetic resonance imaging; Male; Metabolism; pediatric epilepsy; Position (location); Positron emission; Positron emission tomography; seizure onset; Seizures; Sensitivity; Sensitivity analysis; Spiking; SPM; Statistics; Statistics, Nonparametric; Surgery; Tomography
• ISSN: 1065-9471; 1097-0193
• DOI: 10.1002/hbm.23577

To determine the spatial relationship between 2‐deoxy‐2[18F]fluoro‐D‐glucose (FDG) metabolic and intracranial electrophysiological abnormalities in children undergoing two‐stage epilepsy surgery, statistical parametric mapping (SPM) was used to correlate hypo‐ and hypermetabolic cortical regions with ictal and interictal electrocorticography (ECoG) changes mapped onto the brain surface. Preoperative FDG‐PET scans of 37 children with intractable epilepsy (31 with non‐localizing MRI) were compared with age‐matched pseudo‐normal pediatric control PET data. Hypo‐/hypermetabolic maps were transformed to 3D‐MRI brain surface to compare the locations of metabolic changes with electrode coordinates of the ECoG‐defined seizure onset zone (SOZ) and interictal spiking. While hypometabolic clusters showed a good agreement with the SOZ on the lobar level (sensitivity/specificity = 0.74/0.64), detailed surface‐distance analysis demonstrated that large portions of ECoG‐defined SOZ and interictal spiking area were located at least 3 cm beyond hypometabolic regions with the same statistical threshold (sensitivity/specificity = 0.18–0.25/0.94–0.90 for overlap 3‐cm distance); for a lower threshold, sensitivity for SOZ at 3 cm increased to 0.39 with a modest compromise of specificity. Performance of FDG‐PET SPM was slightly better in children with smaller as compared with widespread SOZ. The results demonstrate that SPM utilizing age‐matched pseudocontrols can reliably detect the lobe of seizure onset. However, the spatial mismatch between metabolic and EEG epileptiform abnormalities indicates that a more complete SOZ detection could be achieved by extending intracranial electrode coverage at least 3 cm beyond the metabolic abnormality. Considering that the extent of feasible electrode coverage is limited, localization information from other modalities is particularly important to optimize grid coverage in cases of large hypometabolic cortex. Hum Brain Mapp 38:3098–3112, 2017. © 2017 Wiley Periodicals, Inc.