Multi-parametric diffusion spectrum imaging in tuberous sclerosis complex: Identifying cortical tubers and predicting genotypes.

Objectives: This study employed advanced MRI diffusion imaging techniques to identify cortical tubers in Tuberous Sclerosis Complex (TSC) patients and compared the diagnostic efficacy of various diffusion model parameters in predicting TSC genotypes.

Methods: From July 2019 to April 2024, a prospective study was conducted at our Hospital. Participants meeting specific criteria underwent genetic testing and Diffusion Spectrum Imaging (DSI) data collection. The Dipy toolbox calculated parameters for Diffusion Tensor Imaging (DTI), Diffusion Kurtosis Imaging (DKI), Neurite Orientation Dispersion and Density Imaging (NODDI), and Mean Apparent Propagator (MAP) models. Lesion visibility and contrast were scored by two neuroradiologists. Significant parameters were identified through univariate logistic regression, and predictive models were developed using multivariate logistic regression and backward stepwise regression, resulting in a nomogram.

Results: Eighty-three TSC patients were included (49 females, median age 5 years, IQR 3-9 years). Significant differences were found in lesion visibility and contrast among different diffusion model parameter maps (p < 0.001), with NODDI-ICVF and MAP-QIV showing clear advantages. The DTI, DKI, and MAP models struggled to distinguish small lesions near cerebral sulci from cerebrospinal fluid, while NODDI-ICVF performed well. The combined model using ICVF, QIV, and RTOP parameters demonstrated potentially better diagnostic performance compared to single diffusion models, with the nomogram indicating strong discrimination (AUC of 0.89, 95 % CI: 0.86-0.92). Clinical decision curves indicated significant net benefits at probability thresholds of 15 %-95 %.

Conclusion: NODDI and MAP models reveal cortical tubers more clearly. The combined model based on advanced diffusion parameters offers the best predictive efficiency for TSC genotypes.