Abstracts

Asymmetry in the Diffusion Tensor Image Analysis Along the Perivascular Space Index Predicts Seizure Outcome in MRI-negative Focal Epilepsy

Abstract number : 2.294
Submission category : 5. Neuro Imaging / 5A. Structural Imaging
Year : 2025
Submission ID : 797
Source : www.aesnet.org
Presentation date : 12/7/2025 12:00:00 AM
Published date :

Authors :
Presenting Author: Jungyon Yum, MD – Yongin Severance Hospital, Yonsei University College of Medicine

Wonwoo Lee, MD – Yongin Severance Hospital, Yonsei University College of Medicine
Kyung Min Kim, MD, PhD – Severance Hospital, Yonsei University College of Medicine
Min Kyung Chu, MD, PhD – Severance Hospital, Yonsei University College of Medicine
Won-Joo Kim, MD PhD – Gangnam Severance Hospital, Yonsei University College of Medicine
Soomi Cho, MD – Severance Hospital, Yonsei University College of Medicine

Rationale:

We aimed to investigate glymphatic system function using the diffusion tensor image analysis along the perivascular space (DTI-ALPS) index and evaluate the prognostic value of its asymmetry in patients with MRI-negative focal epilepsy.



Methods:

We retrospectively included 134 patients with MRI-negative focal epilepsy who underwent DTI with the same 3-T MRI scanner and had > 2-year follow-up data. The DTI-ALPS asymmetry index (AI) was defined as the absolute difference between hemispheres divided by their mean. Multivariable logistic regression identified predictors of seizure freedom (SF; no seizures for > 1 year) at last follow-up. Variables that significantly correlated with the DTI-ALPS AI were tested as mediators. Kaplan–Meier analysis assessed associations between the DTI-ALPS AI and time to (1) first SF after MRI or (2) SF at last follow-up.



Results:

Of the 134 patients (median age, 31 years; 72 female), 76 (57%) achieved SF at last follow-up. Compared to those without SF, patients with SF had significantly shorter epilepsy duration (median, 8 years vs. 15 years; p = 0.029), lower seizure frequency (median, 1 vs. 5; p < 0.001), and lower DTI-ALPS AI (median, 0.064 vs. 0.119; p < 0.001). In multivariable logistic regression, DTI-ALPS AI (odds ratio [OR] per 0.1 increase, 0.271; 95% confidence interval [CI],

Neuro Imaging