Avoiding Bias in Intracranial EEG Functional Connectivity in Epilepsy
Abstract number :
1.202
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2022
Submission ID :
2204152
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:23 AM
Authors :
Erin Conrad, MD – University of Pennsylvania; John Bernabei, PhD – University of Pennsylvania; Nishant Sinha, PhD – University of Pennsylvania; Nina Ghosn, BS – University of Pennsylvania; Joel Stein, MD, PhD – University of Pennsylvania; Russell Shinohara, PhD – University of Pennsylvania; Brian Litt, MD – University of Pennsylvania
Rationale: To determine the effect of epilepsy on intracranial EEG functional connectivity and the ability of functional connectivity to localize the seizure onset zone, controlling for important biases (Figure 1).
Methods: We analyzed intracranial EEG data from sequential patients with drug-resistant epilepsy admitted to the Hospital of the University of Pennsylvania for pre-surgical planning. We calculated intracranial EEG functional networks. First, we determined whether changes in functional connectivity lateralized epilepsy using a spatial subsampling method to control for spatial bias. Next, we measured the effect interictal spikes have on functional connectivity. Finally, we developed a “spatial null model” to classify electrodes as belonging to the seizure onset zone or not using only spatial sampling information. We compared the performance of this spatial null model against models incorporating EEG functional connectivity and spike rate data.
Results:: A total of 109 patients were included in the study, though the number of patients analyzed varied across individual analyses. Controlling for spatial sampling, average connectivity was lower in the seizure onset zone relative to the contralateral region. Electrodes with more spikes tended to have lower average connectivity, with lower connectivity surrounding individual spikes than in non-spike time periods. A spatial null model incorporating spatial sampling information alone (prior to examining EEG data) outperformed a chance model (AUC 0.70). A model incorporating EEG functional connectivity and spike rate data further outperformed this spatial null model (AUC 0.78, p = 0.002 via bootstrapping compared to the spatial null model) (Figure 2).
Conclusions: Intracranial EEG functional connectivity is reduced in the seizure onset zone. Interictal intracranial EEG data can be used to localize the seizure onset zone. Here we highlight important sources of bias when studying intracranial EEG functional connectivity in epilepsy. We provide methods to appropriately control for spatial sampling to avoid biasing results of intracranial EEG functional connectivity analyses and to avoid overestimating model performance.
Funding: Erin Conrad received funding from NIH-NINDS (1K23NS121401-01A1) and the Burroughs Wellcome Fund Career Award for Medical Scientists.
Neurophysiology