A Bayesian-based Probabilistic Framework Using Azimuth Angles to Predict Cortico-cortical Evoked Potentials with Minimal Single-pulse Electrical Stimulation
Abstract number :
1.314
Submission category :
4. Clinical Epilepsy / 4B. Clinical Diagnosis
Year :
2024
Submission ID :
916
Source :
www.aesnet.org
Presentation date :
12/7/2024 12:00:00 AM
Published date :
Authors :
Presenting Author: Sahaj Patel, PhD – University of Alabama at Birmingham
Helen Brinyark, BS – University of Alabama at Birmingham
Kyle Evans-Lee, PhD – Science and Technology Research
Joshua LaRocque, MD, PhD – University of Pennsylvania
Erin Conrad, MD – University of Pennsylvania
Arie Nakhmani, PhD – University of Alabama at Birmingham
Benjamin Cox, MD – University of Alabama at Birmingham
Rachel Smith, PhD – University of Alabama at Birmingham
Rationale: In recent years, the use of single-pulse electrical stimulation (SPES) has expanded in neurology, offering to uncover the underlying human brain connectivity networks across different cortices. However, these techniques come with a significant drawback-they require extensive clinician time to understand individual brain connectivity fully. On the other hand, existing Bayesian-based probabilistic models, while capable of predicting CCEP responses, rely on virtual brain models that fail to fully mimic individual neural plasticity.
Methods: To address these challenges, we developed a novel Bayesian-based probabilistic framework, the Azimuth-Based Cortico-Response Prediction Model (ACRPM), to predict Root Mean Square (RMS) values for CCEPs with fewer SPES trials that does not rely on individual virtual brain models. Instead, it utilizes the No-U-Turn Sampler (NUTS) Markov Chain Monte Carlo (MCMC) sampling method, and estimates RMS values using the azimuth angles of individual electrode locations in a spherical coordinate system.
Results: To evaluate the performance of the ACRPM, we conducted a study on a single patient undergoing SPES at the UAB Epilepsy Center. The patient received stimulation with a 5mA current at 1 Hz for thirty trials, with 31 total electrode pairs stimulated across the cortex. Three stimulation sites and their respective CCEP responses were used in the ACRPM to estimate the posterior distribution for other stimulation locations. Based on this estimated distribution, the ACRPM can predict the anticipated CCEP RMS responses for one SPES electrode pair with an average Pearson coefficient of 0.5998.
Conclusions: The proposed ACRPM can identify the unique cortico-cortical effective network of individual epilepsy patients while minimizing the time required to perform SPES. We will generalize this framework by adding patients from a cohort we have gathered at UAB and Penn, which includes over 30 total patients.
Funding: This project is funded by the American Epilepsy Society Junior Investigator Award 1042632 and CURE Taking Flight Award 1061181.
Clinical Epilepsy