Postdoctoral Research Fellow University College London London, England, United Kingdom
Rationale: Seizure semiology is a non-invasive instrument to localize the symptomatogenic zone, which is generally close to, or overlapping the epileptogenic zone. The diagnostic pathway would benefit from a decision support tool to aid clinicians in deducing the epileptogenic zone from seizure semiology, as surgical resection of this area may be curative for up to 50% of patients with refractory focal epilepsy. We present the first Semiology-to-Brain (SemioBrain) Atlas, based on a systematic literature review reporting patient-based semiologies with proven localizing and lateralizing value. Methods: The review was conducted by two independent clinical researchers, in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The literature search was performed in PubMed in January 2019, and produced, after filtering, a total of 1,208 papers (PRISMA Flow Chart in Figure 1). The abstract screening included only articles containing original data from epilepsy patients. The full-text screening further narrowed down the selection to those subjects presenting focal semiologies, whose localizing and/or lateralizing value was established according to one or more of the following ground-truth criteria: 1) successful epilepsy surgery, leading to seizure-freedom at a minimum follow-up of 12 months; 2) intracranial video-EEG monitoring and/or electrical brain stimulation; 3) multimodal concordance of at least two structural and functional techniques, among EEG, magnetoencephalography, MRI, fMRI, PET, single-photon emission CT. We excluded those patients whose seizures were reported as: primary generalized; non-epileptic; described without providing information about semiology (e.g., simple partial seizure); defined with a poor inter-rater agreement (Kappa < 0.4, if provided); non-localizing, due to inconclusive presurgical investigations or unsuccessful surgery (not caused by the partial resection of the epileptogenic zone); localizing to multiple distant and/or bilateral foci or to a broad unspecific brain region (e.g., extra-temporal). Results: A total of 278 papers were eligible for patient-based semiology extraction. Data collection resulted in 4,454 subjects included in the SemioBrain Database, in which each patient corresponds to one data point assigned to: 1) one or more semiologies with localizing and/or lateralizing value (early-onset ictal semiologies were selected, if chronological presentation was specified); 2) the seizure onset zone and/or epileptogenic zone, represented by one or more contiguous brain regions, as established by the ground-truth criteria. The SemioBrain Atlas provides a synthetic view of the Database, containing 2,368 lateralizing and 10,917 localizing data points, with the following gross anatomical distribution: 7,600 temporal, 1,867 frontal, 588 parietal, 489 occipital, 351 cingulate, 857 insula, 209 thalamus (samples of anatomical charts in Figure 2). Conclusions: The SemioBrain Atlas represents the first data source including thousands of patients’ semiologies with a ground-truthed localizing and/or lateralizing value. To use this clinically, we are developing a complementary interactive tool, to visualize the inferred cerebral localization of the semiology presented by a new patient. This can play a role in informing an intracranial EEG implantation strategy and supporting decision making in planning epilepsy surgery. Funding: Please list any funding that was received in support of this abstract.: This work was supported by the UCLH Trustees, the Wellcome/EPSRC Centre for Interventional and Surgical Sciences (WEISS) (203145Z/16/Z), the Wellcome Trust Health Innovation Challenge Fund (WT106882). Click here to view image/table