# Prediction of Epilepsy Development in Traumatic Brain Injury Patients from Diffusion Weighted MRI

**Authors:** Md Navid Akbar, Marianna La Rocca, Rachael Garner, Dominique Duncan, Deniz Erdoğmuş  
**arXiv:** [2004.14580](https://arxiv.org/abs/2004.14580)  
**Date:** 2020-04-30  
**Summary page:** https://www.summarizepaper.com/en/paper/2004.14580v2/

## Key points

- Traumatic brain injury (TBI) can lead to post-traumatic epilepsy (PTE)
- Predicting which TBI patients will develop epilepsy is a challenge
- Researchers used diffusion-weighted imaging (DWI) and fractional anisotropy (FA) analysis obtained from tract-based spatial statistic (TBSS) to measure and analyze brain changes in 14 TBI patients recruited in the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx)
- The measurements were used to train two support vector machine (SVM) models to predict which TBI patients have developed epilepsy
- With a leave-two-out cross-validation, the approach achieved an accuracy of 0.857 $\pm$ 0.18 with a 95% level of confidence
- This study provides insights into the pathomechanism underlying epileptogenesis in TBI patients
- The approach offers a promising tool for predicting PTE development

## Summary

Traumatic brain injury (TBI) is a major public health problem that can lead to post-traumatic epilepsy (PTE), a lifelong complication with an estimated incidence ranging from 2%-50%. Predicting which TBI patients will develop epilepsy remains a challenge in the epilepsy community. To address this issue, researchers used diffusion-weighted imaging (DWI) and fractional anisotropy (FA) analysis obtained from tract-based spatial statistic (TBSS) to measure and analyze brain changes in 14 TBI patients recruited in the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx). The measurements were then used to train two support vector machine (SVM) models to predict which TBI patients have developed epilepsy. With a leave-two-out cross-validation, the approach achieved an accuracy of 0.857 $\pm$ 0.18 with a 95% level of confidence, demonstrating its potential for early characterization of PTE. This study provides insights into the pathomechanism underlying epileptogenesis in TBI patients and offers a promising tool for predicting PTE development.