Spatial and Temporal Volatility of High-frequency Oscillations
Abstract number :
1.197
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2022
Submission ID :
2204028
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:23 AM
Authors :
Naoto Kuroda, MD – Wayne State University; Keiki Inoue, Medical student – International University Health and welfare; Aimee Luat, MD – Associate Professor, Department of Pediatrics, Central Michigan University; Neena Marupudi, MD, MS, FAANS – Assistant Professor, Department of Neurosurgery, Wayne State University; Sandeep Sood, MD – Professor, Department of Neurosurgery, Wayne State University; Eishi Asano, MD, PhD – Professor, Department of Pediatrics & Neurology, Wayne State University
Rationale: To determine the diagnostic utility of spatial and temporal volatility of high-frequency oscillations (HFOs) in epilepsy presurgical evaluation.
Methods: The study included 135 patients who underwent chronic intracranial EEG monitoring followed by cortical resection. We computed the HFO’s occurrence rate (HFOor), spatial volatility (HFOsv), and temporal volatility (HFOtv) at all intracranial electrode sites during interictal slow-wave sleep. We used the automatic HFO detector based on the Montreal Neurological Institute algorithm and the verification function incorporated in a toolbox, RIPPLELAB. HFOsv was defined as the average of HFOor difference between adjacent electrode sites, whereas HFOtv was defined as the average of HFOor difference between adjacent 30-second epochs. We determined the diagnostic utility of each of the three HFO-related biomarkers using a summary measure referred to as subtraction HFObiomarker. This summary measure was defined as a subtraction of a given biomarker value averaged across all preserved sites from that averaged across all resected sites. Multivariate logistic regression analysis tested the hypothesis that a larger subtraction HFObiomarker would be associated with a better seizure outcome, independently of the standard-care variables based on clinical, seizure onset zone, and MRI data. The receiver operating characteristic (ROC) analysis determined which subtraction HFObiomarker would best improve the standard-care model in classifying those achieving ILAE Class-1 outcome.
Results: Multivariate logistic regression analysis revealed that higher subtraction HFOor (p = 0.037), subtraction HFOsv (p = 0.010), and subtraction HFOtv (p = 0.010) was associated with a better chance of ILAE Class-1 outcome, independently of the standard-care variables. The standard-care model classified patients achieving ILAE Class-1 outcome with an area under the ROC curve (AUC) of 0.756. Adding subtraction HFOor, subtraction HFOsv, and subtraction HFOtv to the standard-care model improved the AUC to 0.815, 0.830, and 0.814. The DeLong test indicated that the improved AUC by subtraction HFOsv was statistically significant (difference: 0.07; Bonferroni-corrected p = 0.045; 95% confidence interval: 0.014 to 0.134).
Conclusions: HFOsv best contributed to the improved classification of postoperative seizure outcomes in this cohort. Our observations may be attributed to a substantial spatial instability of HFO occurrence rates in the epileptogenic zone.
Funding: NIH grant NS64033 (to E. Asano), JSPS KAKENHI Grant Number JP 22J23281 (to N. Kuroda)
Neurophysiology