Improved pre-surgical localization of epileptogenic tubers in pediatric tuberous sclerosis complex using mean apparent propagator MRI.

Abstract

Surgery is the most effective treatment for controlling refractory epilepsy in Tuberous Sclerosis Complex (TSC) patients, and accurate pre-surgical localization of epileptogenic tubers is crucial for improving outcomes. However, identifying epileptogenic tubers using conventional MRI remains challenging. This study aimed to evaluate the potential of combining Diffusion Tensor Imaging (DTI) and Mean Apparent Propagator (MAP) MRI for non-invasive tuber identification, enhancing pre-surgical assessment. This prospective study included 42 children with TSC who underwent preoperative MRI, including DTI and MAP sequences. A total of 342 cortical tubers were segmented and split into 70% training and 30% validation sets. LASSO regression was used for feature selection, and a multi-parametric logistic regression model was developed. The combined DTI-MAP model achieved an AUC of 0.83 (95% CI: 0.75-0.91) in the validation cohort, outperforming DTI alone (AUC = 0.769) and MAP alone (AUC = 0.804). Key predictive features included tuber volume (OR = 1.573, p = 0.03), Axial Diffusivity (AD, OR = 31.35, p = 0.011), Fractional Anisotropy (FA, OR = 0.26, p = 0.005), and Mean Squared Displacement (MSD, OR = 0.045, p = 0.023). A nomogram was constructed from these features, providing a visual tool for risk estimation and showing good calibration. Combining DTI and MAP MRI parameters significantly improves non-invasive identification of epileptogenic tubers, providing better guidance for surgical planning and improving long-term seizure control and clinical outcomes in children with TSC.