Abstracts

Rationale: Although in the early stages of research, normative maps representing healthy brain activity have been successful in aiding the localisation of the epileptogenic zone using intracranial EEG (icEEG) [1]. Such maps were created using interictal icEEG recordings from presumed healthy brain regions in patients with drug-resistant epilepsy undergoing evaluation for resective surgery. To our knowledge, normative icEEG maps have only been created from a static viewpoint and are yet to incorporate the effect of age. As seen in other modalities, natural ageing has substantial effects on the patterns of healthy brain activity [2], so it is imperative to consider age effects in normative icEEG mapping to improve the localisation of epileptogenic tissue. It is tenable that healthy icEGG data (were it available) would not look the same for elderly individuals vs. pediatrics.

Methods: The methodology for generating the normative map remained similar to that outlined in [1]. That is, MRIs were processed for electrode localisation and resection delineation. The icEEG data was processed to obtain 60-second segments of relaxed wakefulness from each patient, with presumed abnormal tissue excluded. The chosen segments had to meet certain criteria, including being a minimum of two hours away from any seizures, to increase the chances of the data representing healthy brain activity. After applying a common average reference to all recordings in all subjects, we estimated the power spectral density with Welch’s method in each 60-second recording. The average band power within five frequency bands of interest was then calculated. Band power estimates were transformed and L1-normalised for each contact; each subject therefore has a value of relative band power assigned to each contact and each frequency band._x000D_ Bespoke statistical models were developed to estimate the normative map as a function of age. This included investigating the spatial structure of the map, in order to leverage it for a hierarchical statistical model and for data simulation in regions which lacked data at certain ages.

Results: We found age effects in the normative map that are region and frequency band dependent, supporting our initial hypothesis that it was an important variable. We successfully built a hierarchical statistical model to account for these age effects, that leverages spatial information to inform the model.

Conclusions: We expect this map to be useful for future research into improving the localisation of epileptogenic tissue from interictal icEEG. A further goal is to use the map as a clinical tool to better determine which area should be resected prior to surgery by highlighting abnormalities in patient data through direct comparison to the map.
_x000D_ References:_x000D_ 1. Taylor et al. Normative brain mapping of interictal intracranial EEG to localize epileptogenic tissue, Brain. 2022;145(3)_x000D_ 2. Gómez et al. Spectral changes in spontaneous MEG activity across the lifespan, Journal of Neural Engineering. 2013;10(6)

Funding: Industrial funding was received from AkzoNobel, Microsoft and Red Hat as part of the EPSRC Centre for Doctoral Training in Cloud Computing for Big Data EP/L015358/1.