Normative Modeling of Diffusion MRI-derived Parameters for Temporal Lobe Epilepsy Lateralization
Abstract number :
1.525
Submission category :
5. Neuro Imaging / 5A. Structural Imaging
Year :
2025
Submission ID :
1278
Source :
www.aesnet.org
Presentation date :
12/6/2025 12:00:00 AM
Published date :
Rationale: Diffusion MRI-derived parameters are sensitive to microstructural changes in brain tissue. Diffusion abnormalities in epilepsy patients relate to lesion locations1 and surgical outcomes,2 demonstrating potential for tissue anomaly mapping and surgical planning. However, given the many reconstruction models and parameters available, it remains unclear which are most abnormal in focal epilepsy. Further, most studies on diffusion parameter abnormalities in epilepsy have analyzed grey and white matter in isolation and have not appropriately accounted for age- and sex-related variability, despite the interrelated nature of grey and white matter microstructure in typical development and disease.
Methods: We leveraged multi-shell diffusion MRI data from 60 epilepsy patients with unilateral temporal seizure onset (n=36 left, n=24 right) and 1,349 healthy controls aggregated from Penn (n=55), HCP-YA (n=1064), and HCP-Aging (n=224). Data were preprocessed with QSIPrep and diffusion models fit with QSIRecon, generating 38 diffusion parameter maps across five models (DTI, DKI, GQI, NODDI, MAP-MRI). Diffusion parameters are averaged within grey and white matter regions. Differences explained by MRI scanners are accounted for using CovBat. Normative parameter values are modeled as a function of age (non-linear fit) and sex using generalized additive models, with patient-specific deviations quantified as z-scores. We summarized abnormalities as mean |z| across regions and parameters.
Results: In grey matter, abnormalities were greatest in the ipsilateral hippocampus for both left (|z| = 1.73) and right (|z| = 2.30) TLE patients. In white matter, left TLE patients showed most extensive abnormalities in the ipsilateral inferior longitudinal fasciculus (|z| = 1.34) and parahippocampal cingulum (|z| = 1.32), whereas abnormalities were greatest for right TLE patients in the ipsilateral fornix (|z| = 1.94) and uncinate fasciculus (|z| = 1.84). DTI and DKI parameters were among the most abnormal parameters for both left and right TLE, while MAP-MRI parameter abnormalities were greatest for left TLE patients.
Conclusions: Demographically-informed normative models of diffusion MRI-derived parameters can be used to localize tissue anomalies in TLE patients. Distinct regional patterns of diffusion parameter abnormalities in left and right TLE underscore unique microstructural changes to each TLE subtype. Future work will apply more spatially resolved approaches (e.g., white matter tract and cortical depth profiling).
Funding: MJ received support from Penn Biomedical Graduate Studies and NINDS (F31NS145529-01). KAD received support from NINDS (R01NS116504).
Neuro Imaging