Planning stereoelectroencephalography using automated lesion detection: Retrospective feasibility study

Abstract

Objective This retrospective, cross‐sectional study evaluated the feasibility and potential benefits of incorporating deep‐learning on structural magnetic resonance imaging (MRI) into planning stereoelectroencephalography (sEEG) implantation in pediatric patients with diagnostically complex drug‐resistant epilepsy. This study aimed to assess the degree of colocalization between automated lesion detection and the seizure onset zone (SOZ) as assessed by sEEG. Methods A neural network classifier was applied to cortical features from MRI data from three cohorts. (1) The network was trained and cross‐validated using 34 patients with visible focal cortical dysplasias (FCDs). (2) Specificity was assessed in 20 pediatric healthy controls. (3) Feasibility of incorporation into sEEG implantation plans was evaluated in 34 sEEG patients. Coordinates of sEEG contacts were coregistered with classifier‐predicted lesions. sEEG contacts in seizure onset and irritative tissue were identified by clinical neurophysiologists. A distance of <10 mm between SOZ contacts and classifier‐predicted lesions was considered colocalization. Results In patients with radiologically defined lesions, classifier sensitivity was 74% (25/34 lesions detected). No clusters were detected in the controls (specificity = 100%). Of the total 34 sEEG patients, 21 patients had a focal cortical SOZ, of whom eight were histopathologically confirmed as having an FCD. The algorithm correctly detected seven of eight of these FCDs (86%). In patients with histopathologically heterogeneous focal cortical lesions, there was colocalization between classifier output and SOZ contacts in 62%. In three patients, the electroclinical profile was indicative of focal epilepsy, but no SOZ was localized on sEEG. In these patients, the classifier identified additional abnormalities that had not been implanted. Significance There was a high degree of colocalization between automated lesion detection and sEEG. We have created a framework for incorporation of deep‐learning–based MRI lesion detection into sEEG implantation planning. Our findings support the prospective evaluation of automated MRI analysis to plan optimal electrode trajectories.

The UK experience of stereoelectroencephalography in children: An analysis of factors predicting the identification of a seizure‐onset zone and subsequent seizure freedom

“…In addition to refining the selection of patients, implantation strategy, and subsequent surgical planning in SEEG patients using clinical data, we are likely to see increasing incorporation of quantitative methods in SEEG planning, including automated analysis of MRI 30 and computational analysis of SEEG recordings 22,31,32 33 using additional methods such as microelectrode recordings, 34 or network‐based analyses 22,35 may improve the interpretation of SEEG recordings as we move from a location‐focused to network‐based interventions.…”

The UK experience of stereoelectroencephalography in children: An analysis of factors predicting the identification of a seizure onset zone and subsequent seizure freedom

“…In addition to refining the selection of patients, implantation strategy and subsequent surgical planning in SEEG patients using clinical data, we are likely to see increasing incorporation of quantitative methods in SEEG planning, including automated analysis of MRI, 31 and computational analysis of SEEG recordings. 22,32,33 Whilst seizure (both spontaneously recorded and stimulated) have been shown to be crucial to outcomes in this present series, concepts such as identification of the SOZ from interictal recordings 34 , using additional methods such as microelectrode recordings 35 or network-based analyses 22 may improve the interpretation of SEEG recordings as we move from a location-focused to network-based interventions.…”

Quantitative MRI susceptibility mapping reveals cortical signatures of changes in iron, calcium and zinc in malformations of cortical development in children with drug-resistant epilepsy

“…The QSM and R2* surfaces were smoothed using a 10 mm vertex-wise FWHM Gaussian kernel in order to increase stability of the per‐vertex sampling at different depths ( Adler et al., 2017b ; Jin et al., 2018 ; Lorio et al., 2020 ; Wagstyl et al., 2020 ). Manual delineated lesions had a median area of 1074 mm 2 and median absolute deviation of 756 mm 2 , which is much larger than these smoothing kernels.…”