Clinical Utility of Advanced Neuroimaging Techniques in the Presurgical Assessment of Patients with Epilepsy

Abstract

For the 30% of people with focal epilepsy who are medically refractory, neurosurgery is a potential treatment option. However, identifying suitable candidates and localizing the epileptogenic zone remains a significant challenge. This study evaluates the clinical utility of three neuroimaging techniques in the pre-surgical assessment at the newly designated District Epilepsy Center, Kingston Health Sciences Center. Additionally, language lateralization indices (LI) are compared when processed start-to-finish using two popular open-source software packages, SPM12 and FSL, or preprocessed with fMRIPrep. Patients in the pre-surgical pathway were discussed in multidisciplinary team meetings (n=21). Exclusions included patients who had already undergone resections (n=5) and those deemed unsuitable for surgery (n=4). Patients meeting the criteria for surgical candidacy with inconclusive clinical data, such as a negative MRI or discordant data (n=12), were recruited for comprehensive MRI evaluation. This included high-resolution structural imaging for hippocampal/amygdala volumetry and lesion detection, and fMRI for language lateralization. The benefits of this additional data were documented in a follow-up questionnaire during the re-evaluation of the patients. Patients (n=12) and healthy controls (n=12) participated in two task-based fMRI paradigms: sentence completion and word generation. The analysis involved GLM and LI calculations for three regions of interest using tools specific to each software. 144 LIs were calculated for each of the four processing pipelines. Overlap of activations were assessed visually and with Dice coefficients (DC). Following assessment, all previously discussed surgery candidates were deemed suitable or could proceed to the next assessment step (n=12). The protocol identified previously unseen imaging abnormalities (n=4) or hippocampal asymmetry (n=2) and provided useful language laterality information (n=12). Visual assessment of overlap in the fMRI processing pipelines showed similar results, but DC ranged from 0.16-0.70. However, LI remained concordant in 531/576 (92%) of combinations. These findings demonstrate that these advanced neuroimaging modalities can be effectively integrated into clinical practice at a newly established epilepsy surgery center, providing valuable information to benefit patient care. Furthermore, all fMRI processing tools performed similarly for block design language fMRI data; software choice ultimately depends on factors such as usability and alignment with the standard practices of individual surgery centers.