Purpose: Tuberous sclerosis complex (TSC) is a genetic disorder affecting two million people worldwide. Epilepsy is a common symptom, affecting over 85% of patients, which is linked to benign lesions called tubers. The objective of this project was to apply quantitative radiomics and atlas-based approaches to distinguish these brain abnormalities from normal tissue in TSC for improved diagnosis.
Methods: This study enrolled 20 TSC and 15 controls patients who received 3D FLAIR MRI. Preprocessing included skull stripping, coregistration, and intensity normalization. Using the Brainnetome and Harvard-Oxford atlases, brain regions were parcellated into 248 discrete areas. Expert neuroradiologists spatially labeled tubers in TSC patients using ITK-SNAP. Using pyradiomics, 88 radiomic features were extracted to characterize normal brain, tuber-affected, and non-tuber-affected parenchyma, following IBSI guidelines. LASSO (least absolute shrinkage and selection operator) was used to perform variable selection and regularization. Relevant radiomic features selected by LASSO were combined to produce a summary score ω, defined as the sum of squared differences from average control group values. Region-specific ω scores were converted to heatmaps and coregistered with brain MRI to reflect overall radiomic deviation from normal. The Wilcoxon paired test was used to compare heat maps between TSC and control patients.
Results: Feature selection by LASSO identified five of 88 radiomic features to distinguish normal brain from brain abnormalities in TSC patients. Average scores were significantly higher for TSC patients versus controls, with an average ± SD of 0.48 ± 1.56 vs. 0.08 ± 0.05 (p<0.0001). Heat maps provided a visual assessment of the radiomic summary score ω coregistered with lesions in MRI space.
Conclusion: We have established a novel pipeline for quantitative radiomic feature extraction and selection in TSC, with automated calculation of a summary score and heat maps coregistered to MRI. This novel approach could become a clinical decision support tool.
Funding Support, Disclosures, and Conflict of Interest: This research was supported by a grant from the DoD Tuberous Sclerosis Complex Research Program (TSCRP) to Mark Hester (W81XWH-21-1-0278)