Abstracts

Rationale: Bottom-of-sulcus dysplasia (BOSD) is a localised subtype of focal cortical dysplasia type II (FCDII), which can be challenging to detect on MRI but rewarding to resect for seizure control.¹ Patients with BOSD often have frequent interictal epileptiform discharges (IEDs) which are known from intracranial EEG to be generated in the BOSD.¹ Electrical source imaging (ESI) incorporates temporal and spatial components of IEDs to localise the cortical generator. We aimed to test the accuracy of ESI in BOSD, BOSD being a perfect model of localised epileptogenicity.


Methods: Patients who underwent resection of BOSDs at the Royal Children’s Hospital from 2008-2023 were screened. Inclusion criteria were: MRI-positive BOSD, preoperative scalp EEG with IEDs, intraoperative ECoG with spiking, FCDII histopathology, and post-operative seizure freedom ( >12 months). Using Curry-9 (Compumedics), IEDs were manually selected from pre-operative scalp EEG (with 3-12 additional 10-10 electrodes). 15-30 IEDs were averaged to increase signal-to-noise ratio. A 3D boundary element head model and pial brain surface were constructed using the patients’ T1-weighted volumetric MRI. Source analysis was performed using moving dipole and distributed source (sLORETA) models. Solutions were examined from the onset to the peak of the averaged spikes and localisation concordance was computed from the midway point. The concordance of each ESI solution was visually compared to a manual segmentation of each BOSD, with sub-lobar localisation within the same or adjacent parcel on the Desikan-Killiany atlas considered concordant.


Results: 40 patients were included with a mean age at EEG of 9.3 (SD: 4.3) years. BOSDs were detected on the first MRI in only 10 patients (25%), highlighting their subtle nature. BOSDs were located in the frontal (18, 45%), insular (6, 15%), parietal (12, 30%) and temporal (4, 10%) lobes. 22 were on the cerebral convexity, the remainder being on the medial or basal surfaces or within the Sylvian fissure. Median BOSD size was 1.4 (0.53-2.69)cm³. Curation and computation of ESI studies took 30-45 minutes each. ESI solutions generated with moving dipole solutions were relatively restricted and stable, appearing essentially identical to rotating and fixed dipole models. The equivalent current dipole (ECD) and sLORETA solutions showed concordant hemispheric lateralisation in all but one patient with an interhemispheric BOSD. ECD and sLORETA solutions were concordant with BOSD lobar location in 36 (90%) and 34 (85%) patients, and BOSD sub-lobar localisation in 28 (70%) and 32 (80%). The confidence ellipsoid of the ECD sphere overlapped BOSD in 31 patients (78%) and the sLORETA solution overlapped in 32 (80%).


Conclusions: ESI performed on routine pre-operative scalp EEG co-localised with BOSD in ~80% patients, validating ESI’s ability to localise small cortical generators of IEDs. ESI may aid BOSD detection on MRI, particularly when performed with high-density scalp EEG and combined with ¹⁸F-FDG-PET.

1. Macdonald-Laurs E, Warren AEL, Francis P, et al. The clinical, imaging, pathological and genetic landscape of bottom-of-sulcus dysplasia. Brain 2023;147:1264-1277


Funding: EML is supported by MCRI & ANZCNS.