Abstracts

Rationale: Epilepsy is a disorder of brain networks. Focal epileptic seizures originate from a pathological brain area and later spread via normal or aberrant brain pathways.  Understanding cortical and sub-cortical connectivity is critical for the planning of stereo-EEG implantation and subsequent resective epilepsy surgery. While fMRI and DTI studies can provide resting-state functional and structural connectivity maps respectively, no connectivity maps exist that provide a functional association between brain regions during an epileptic seizure. Such maps can help understand the complex spatio-temporal patterns associated with seizure onset and propagation, and can further inform invasive evaluation and localization of epileptogenic zone. One of the challenges associated with generating a seizure-informed cerebral connectivity map is the difficulty in imaging seizures inside an MRI scanner. In this study, we developed an innovative approach to understanding the association between different cortical and subcortical brain regions during a seizure using the wealth of ictal single-photon-emission-computed-tomography (SPECT) images that were collected as part of the presurgical evaluation of patients with drug-resistant focal epilepsy.

Methods: Ictal SPECT data for this study was derived from clinical recordings obtained between 2015 and 2018 and also underwent invasive evaluation using stereo-EEG. We excluded patients with prior epilepsy surgery and those who had a postictal rather than ictal SPECT injection. First, we performed cortical reconstruction of the patient's MRI using brainsuite and then parcellated the MRI using the USC brain atlas. Second, we subtracted the interictal SPECT from ictal SPECT using techniques described in Krishnan et al to generate an ictal perfusion map (Krishnan et. al, 2021). Third, for every patient, we extracted the average ictal SPECT perfusion z-score for every parcel. Finally, we estimated the correlation of perfusion z-scores across patients to arrive at a connectivity map.

Results: Fifty-four patients satisfied the specific inclusion and exclusion criterion. Brainsuite-based parcellation resulted in 154 cortical and sub-cortical parcels. Figure 1 shows the 154x154 correlation matrix estimated from the perfusion z-score map across 54 patients. A red shade represents positive connectivity between a pair of parcels and a blue shade corresponds to negative connectivity. Figure 2A-B shows a representative perfusion-based connectivity map for the left and right hippocampus respectively. We estimated a similar connectivity map for all 154 brain parcels and visualized it using the circular graph.

Conclusions: Our study describes a novel approach to agglomerate and quantify information across ictal cerebral blood perfusion maps. Validation of this tool using a functional connectivity map generated using cortico-cortico evoked potentials and SEEG ictal propagation pattern can provide an objective tool to guide SEEG pre-implantation strategy, epileptogenic zone localization, and understand semiological progression during seizures.

Funding: None