Related Experiment Video
Updated: Nov 12, 2025
12:09
Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy
Published on: August 5, 2014
18.3K
Identification of focal epilepsy by diffusion tensor imaging using machine learning.
Dong Ah Lee1, Ho-Joon Lee2, Byung Joon Kim1
1Department of Neurology, Haeundae Paik Hospital, Inje University College of Medicine, Busan, Korea.
Acta Neurologica Scandinavica
|March 18, 2021
Summary
Machine learning with diffusion tensor imaging (DTI) effectively distinguishes focal epilepsy patients from controls. However, this method struggles to predict antiseizure medication (ASM) response, though combining DTI with connectomic profiles improves accuracy.
Area of Science:
- Neuroimaging
- Machine Learning
- Epilepsy Research
Background:
- Focal epilepsy diagnosis and treatment response prediction remain clinical challenges.
- Diffusion Tensor Imaging (DTI) offers insights into white matter integrity.
- Machine learning (ML) can analyze complex neuroimaging data.
Purpose of the Study:
- To assess the feasibility of ML models using DTI measures for differentiating focal epilepsy patients from healthy controls.
- To evaluate the capability of ML-DTI models in predicting antiseizure medication (ASM) responsiveness.
- To compare the performance of conventional DTI measures versus combined DTI and structural connectomic profiles.
Main Methods:
- Retrospective analysis of 456 focal epilepsy patients and 100 healthy controls.
- Acquisition and analysis of conventional DTI measures and structural connectomic profiles.
- Application of Support Vector Machine (SVM) classifiers for patient classification.
Main Results:
- ML-DTI models achieved 76.5% accuracy distinguishing epilepsy patients from controls.
- Combining structural connectomic profiles with DTI improved classification accuracy to 82.8%.
- ML-DTI models showed limited success in predicting ASM responsiveness (accuracy ~55-59%).
Conclusions:
- DTI combined with ML is a promising tool for identifying focal epilepsy.
- Current ML-DTI approaches are insufficient for predicting ASM responsiveness.
- Integrating structural connectomic profiles enhances the diagnostic performance of DTI-based ML models.
