Abstracts

Implementation of Neural Fragility in R for Visualization and Validation of Epileptogenic Zone

Abstract number : 3.313
Submission category : 9. Surgery / 9A. Adult
Year : 2022
Submission ID : 2204426
Source : www.aesnet.org
Presentation date : 12/5/2022 12:00:00 PM
Published date : Nov 22, 2022, 05:25 AM

Authors :
Oliver Zhou, BA – University of Texas Medical Branch; John Magnotti, PhD – Neurosurgery – University of Pennsylvania; Zhengjia Wang, PhD – Neurosurgery – University of Pennsylvania; Michael Beauchamp, PhD – Neurosurgery – University of Pennsylvania; Sean O'Leary, BSA, BA – Neurosurgery – University of Texas Medical Branch; Adam Husain, BS – Neurosurgery – University of Texas Medical Branch; Anthony Price, BA – Neurosurgery – University of Texas Medical Branch; Ron Gadot, BS – Neurosurgery – Baylor College of Medicine; Sameer Sheth, MD, PhD – Neurosurgery – Baylor College of Medicine; Liliana Camarillo Rodriguez, MD, PhD – Neurosurgery – University of Texas Medical Branch; Patrick Karas, MD – Neurosurgery – University of Texas Medical Branch

Rationale: Localization of the epileptogenic zone (EZ) in patients with drug-resistant epilepsy is a key step in determining a surgical plan. Novel techniques have been developed to detect intracranial EEG (iEEG) markers of the EZ as supplements to electro-anatomo-clinical localization. Neural fragility is one such measure of epileptogenicity, based on a time-varying linear model of iEEG signals. As described by Li et al (Nat Neurosci. 2021;24:1465-1474), it was able to outperform 20 other iEEG biomarkers in predictive power and interpretability.
_x000D_ Many of these novel techniques, including neural fragility, require extensive processing of iEEG data. Independently replicating these analyses as part of a clinical or investigational pipeline is time-consuming and can be error prone. Furthermore, the coding languages that these programs are written in vary and may require costly licensing. Lack of a standardized, easy-to-use application to perform these analyses means that accessibility poses a significant challenge for any surgical epilepsy program that may be considering them. R Analysis and Visualization of Intracranial EEG, or “RAVE,” is a powerful, free, open-source, NIH-funded research tool designed to analyze and visualize iEEG data. RAVE also allows users to code modules for custom analyses within the RAVE environment. RAVE is written in R, a free software program, and runs from any web browser. Here we present an accessible implementation of neural fragility built as a graphical user interface within a RAVE module.

Methods: RAVE is written using the R Shiny package to run interactive web browser applications. Source code and documentation are available for RAVE at https://openwetware.org/wiki/RAVE and the Fragility module at https://github.com/ozmosis17/Fragility. RAVE can be run from a computer, and users may interact via web browser from any internet-connected device (Figure 1). The Fragility module approximates the analysis outlined by Li et al. After pre-processing patient data, a time-varying linear model of the iEEG data is generated. Using this model, each electrode’s fragility can be calculated across time. Fragility is then visualized in various ways, including by electrode over time (Figure 2A) or averaged and projected onto electrodes in the brain (Figure 2B). The original iEEG traces can also be displayed for reference. Using these visual tools, the most fragile electrodes and their locations can be identified.

Results: The RAVE Fragility module allows users to easily analyze the neural fragility of an epilepsy patient within a web browser environment. The calculations required to obtain this biomarker are built-in to the module, allowing users with no coding or statistical experience to access a similar analysis to that of Li et al.

Conclusions: The implementation of neural fragility into RAVE is a first step in making complex iEEG biomarkers more accessible for surgical epilepsy programs. As a free, open-source tool that can be deployed from a server and run on a web browser anywhere, this RAVE module will allow for broader validation of neural fragility and streamlined sharing of these analyses.

Funding: None
Surgery