Resecting BOLD Responses Related to Spikes and High Frequency Oscillations Improves Surgical Outcome
Poster No:
2417
Submission Type:
Abstract Submission
Authors:
William Wilson1, Daniel Pittman1, Perry Dykens1, Victoria Mosher1, Laura Gill1, Joseph Peedicail1, Antis George1, Craig Beers1, Bradley Goodyear1, Paolo Federico1
Institutions:
1University of Calgary, Calgary, Alberta
First Author:
Co-Author(s):
Introduction:
Simultaneous intracranial EEG-fMRI (iEEG-fMRI) may be used to identify candidate regions for resection in patients with drug-resistant focal epilepsy. This is achieved by mapping the blood oxygen-level dependent (BOLD) activation associated with events of interest, typically epileptiform spikes, captured in the EEG. However, the activation maps associated with spikes have shown only moderate sensitivity and specificity for epileptogenic tissue essential for resection (Koupparis et al., 2021). High frequency oscillations (HFOs) have been proposed as a novel EEG marker that may better localize epileptogenic tissue than spikes in the pre-operative brain (Thomschewski et al., 2019). The hemodynamic response to HFOs measured using simultaneous iEEG-fMRI and its relation to epileptogenic tissue is not known. In this study, we evaluated the relationship between resection of areas of HFO-related BOLD activation and post-operative outcome. Further, we compared the post-operative predictive capabilities of both HFO-related BOLD activation and the BOLD activation associated with the more conventional marker, spikes.
Methods:
Thirty-eight adult patients with focal epilepsy undergoing intracranial video-EEG monitoring were enrolled in this study. Each patient underwent a 60-minute iEEG-fMRI scan. Standard preprocessing steps were applied to the fMRI data (motion correction, slice-timing correction, non-brain removal, spatial smoothing, high-pass temporal filtering, and noise removal using independent component analysis). iEEG data were preprocessed by removing gradient switching artifact using an in-house algorithm and downsampling to 2500Hz. The timing of spikes and HFOs in the EEG recording were identified for each patient and used to generate statistical maps of related activity. The Maximum positive BOLD cluster was identified in each map. The resulting maps were then stratified into low, medium, and high confidence groups based on the magnitude of the z-score of the Maximum cluster relative to the second maximum cluster, a method shown to identify more localizing maps (Koupparis et al., 2021; Khoo et al., 2017). Patients then underwent epilepsy surgery guided by established clinical investigations, and not the iEEG-fMRI data. Spatial concordance of the Maximum cluster with the area of resection was determined using post-operative imaging. Concordance was related to post-operative outcome >12 months after surgery.
Results:
At the time of last clinical follow up (mean interval = 4.5 years; range = 1.0 – 14.9 years), 21/38 (55%) subjects had a good post-operative outcome, and 17/38 (45%) subjects had a poor post-operative outcome. Resection of the peak voxel of high confidence Maximum clusters was associated with good post-operative outcomes for both spikes and HFOs (p = 0.040 and 0.035, Fisher exact test). HFOs were found to additionally predict outcome in cases where the peak voxel was not removed, but some other part of the Maximum cluster overlapped with the area of resection. More specifically, we found that resection of >30% of the Maximum HFO-related cluster was associated with good outcome (sensitivity = 80% and specificity = 80%). In all significant associations, a negative predictive value of 100% was observed accompanied by poor positive predictive values.
Conclusions:
Investigating the hemodynamic response to HFOs using iEEG-fMRI is feasible and generates maps of related activity that may further enhance localization of epileptogenic tissue critical to surgical planning. Furthermore, confidence level, which takes into account how distinct the Maximum cluster is relative to other areas of activation, should be considered when interpreting EEG-fMRI results of spikes and HFOs. Finally, removal of the peak voxel associated with both spikes and HFOs is necessary but not sufficient to achieve a good post-operative outcome. However, HFOs may provide useful spatial information in a broader range of patients undergoing intracranial monitoring.
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI) 2
EEG/MEG Modeling and Analysis
Novel Imaging Acquisition Methods:
Multi-Modal Imaging 1
Keywords:
ADULTS
Electroencephaolography (EEG)
Epilepsy
FUNCTIONAL MRI
Source Localization
Other - Surgery
1|2Indicates the priority used for review
Provide references using author date format
Koupparis, A. (2021), 'Association of EEG-fMRI Responses and Outcome After Epilepsy Surgery', Neurology, vol. 97, no. 15, pp. e1523–e1536
Thomschewski, A. (2019), 'Localization of the Epileptogenic Zone Using High Frequency Oscillations', Frontiers in Neurology, vol. 10, no. 94