Abstracts

This is a Late-Breaking abstract.

Rationale: The standard pre-surgical diagnostic procedure in drug-resistant epilepsy usually involves visual inspection of long intracranial EEG (iEEG) recordings to identify epileptogenic regions. This is a very time-consuming and demanding procedure that might lead to inconclusive interpretations, resulting in mistakes and incomplete diagnosis. In this context, can computational tools unravel epileptogenic significant information that is invisible to the human eye? Further, is the application of these tools equally effective across different epilepsy types, such as temporal and extratemporal epilepsies?

Methods: To address the above challenges, we have recently designed a study to validate BrainFocus, a software toolbox that combines an own developed automatic epileptogenic detection algorithm [1] and frequency-dependent visualization maps of iEEG recordings [2] and outputs a final report including a description of the seizure onset patterns and a list of regions that most likely lie or are connected to the seizure onset zone. The validation study has been performed against the standard diagnostic information and the available post-surgical outcomes provided by clinicians.

BrainFocus has been retrospectively applied to a cohort of 20 drug-resistant epileptic patients from Hospital Clínic (Barcelona, Spain), including 60% of extra-temporal lobe epilepsies. This study is mainly centered at analyzing the specificity (against the state-of-the-art diagnosis) of the epileptogenic detection algorithm as a means to assess the clinical relevance of brain sites that were not initially mapped by visual examination. Our preliminary results indicate that the mean specificity was of 0.8±0.1 over patients, and that clinicians uncovered novel sites of clinical significance in a considerable fraction of patients (~25%), especially in extratemporal epilepsy cases.

Conclusions: BrainFocus is a novel online software aimed to assist clinicians in the diagnosis of epileptic patients with EEG, extracting epileptogenic relevant information that might be hidden to the human eye in challenging cases.
_x000D_ References:_x000D_ 1. M. Vila-Vidal, C. Pérez Enríquez, A. Principe, R. Rocamora, G. Deco, and A. Tauste Campo. Low entropy map of brain oscillatory activity identifies spatially localized events: A new method for automated epilepsy focus prediction. Neuroimage. 2020;208:116410, doi.org/10.1016/j.neuroimage. 2019.116410.
2. M. Vila-Vidal, A. Principe, M. Ley, G. Deco, A. Tauste Campo, and R. Rocamora. Detection of recurrent activation patterns across focal seizures: application to seizure onset zone identification. Clin Neurophysiol. 2017;128(6):977-985. doi.org/10.1016/j.clinph.2017.03.040.
_x000D_ Funding: M.V. and ATC were supported by the Spanish Ministry of Science and Innovation, Spain (grant agreement number PID2020-119072RA-I00, MCIN/AEI/) and by “La Caixa” Foundation, CaixaImpulse Validate program (grant agreement CI20-00195).