Beyond Brainstorms: Predicting Epilepsy Surgery Outcome with Functional Connectivity Centrality
Citation
Zhu, Siye. 2022. Beyond Brainstorms: Predicting Epilepsy Surgery Outcome with Functional Connectivity Centrality. Bachelor's thesis, Harvard College.Abstract
For patients with medically refractory epilepsy (MRE) who do not respond to anti-seizure medicine, invasive intracranial EEG (icEEG) and resective surgery are used to respectively identify and remove epileptogenic zones (EZ).Advances in signal processing and network analysis have sparked interest in the processing and interpretation of icEEG data for more accurate identification of the EZ. The aim of this study is to use functional connectivity centrality measures extracted from interictal icEEG recording to localize seizure onset zones (SOZ) and predict post-surgical outcomes. The SOZ is the most reliable estimator for EZ with observable ground truth, and the current standard practice relies on clinicians to visually identify individual electrodes as SOZ during every ictal recording, which is an inaccurate and inefficient method. This paper analyzed five minutes of interictal icEEG data from 40 MRE patients who underwent surgery at Boston Children’s Hospital. First, classification models using centrality measures were able to robustly identify SOZ regions (AUC 0.70). Then, using likelihood-based diagnostic models and proportion of resection, we were able to predict post-surgical outcome with AUC 0.74. Our results suggest that functional connectivity centrality measures are efficient and useful features for clinicians in the treatment of MRE.
Terms of Use
This article is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAACitable link to this page
https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37371766
Collections
- FAS Theses and Dissertations [6138]
Contact administrator regarding this item (to report mistakes or request changes)