arXiv subjects

Giovanna Rizzo

Publications and source records attributed to Giovanna Rizzo.

2 recordsLinked to original sources

arXiv2026

Combined Diffusion-Relaxation MRI to Assess Muscle Microstructure and Composition

Quantifying muscle tissue properties is crucial for understanding pathophysiological changes occurring in skeletal muscle (SM). In particular, T2 relaxation and diffusion MRI (dMRI) are promising techniques. However, typical methods measure T2 and diffusion separately, making them less specific to microstructure than emerging combined diffusion-relaxation techniques. Here we demonstrate a combined diffusion-relaxation MRI approach for disentangling T2 and diffusivity properties in SM. A diffusion-relaxation acquisition was implemented on a 3 T scanner, combining six b-values and four echo times within a 12-min single-slice protocol. Five healthy participants were enrolled. Data were analysed with six microstructural diffusion and diffusion-relaxation models. Mean parameter values were extracted from manually segmented calf muscles. Models neglecting T2 relaxation showed strong TE dependence: mean diffusivity (MD) decreased by up to 47\%, fractional anisotropy (FA) increased by up to 75\%, and vascular fraction fv increased by up to 297\% when TE increased from 50 to 90 ms. Diffusion-relaxation models produced TE-independent estimates. Tissue and vascular relaxation times ranged 31-36 ms T2t and 66-86 ms T2v, respectively. Simulations confirmed improved accuracy for fv estimation (r=0.95; RMSE=0.03) and reduced TE-related bias. Combined diffusion-relaxation MRI provides robust, TE-independent estimates of muscle microstructural and perfusion-related biomarkers. The quantitative improvements observed - particularly in the estimation of fv - show its potential to provide non-invasive biomarkers for the assessment of muscle physiology, exercise adaptation, rehabilitation, and neuromuscular pathology.

physics.med-ph↗

arXiv2025

A Comprehensive Framework for Uncertainty Quantification of Voxel-wise Supervised Models in IVIM MRI

Accurate estimation of intravoxel incoherent motion (IVIM) parameters from diffusion-weighted MRI remains challenging due to the ill-posed nature of the inverse problem and high sensitivity to noise, particularly in the perfusion compartment. In this work, we propose a probabilistic deep learning framework based on Deep Ensembles (DE) of Mixture Density Networks (MDNs), enabling estimation of total predictive uncertainty and decomposition into aleatoric (AU) and epistemic (EU) components. The method was benchmarked against non probabilistic neural networks, a Bayesian fitting approach and a probabilistic network with single Gaussian parametrization. Supervised training was performed on synthetic data, and evaluation was conducted on both simulated and an in vivo dataset. The reliability of the quantified uncertainties was assessed using calibration curves, output distribution sharpness, and the Continuous Ranked Probability Score (CRPS). MDNs produced more calibrated and sharper predictive distributions for the diffusion coefficient D and fraction f parameters, although slight overconfidence was observed in pseudo-diffusion coefficient D*. The Robust Coefficient of Variation (RCV) indicated smoother in vivo estimates for D* with MDNs compared to Gaussian model. Despite the training data covering the expected physiological range, elevated EU in vivo suggests a mismatch with real acquisition conditions, highlighting the importance of incorporating EU, which was allowed by DE. Overall, we present a comprehensive framework for IVIM fitting with uncertainty quantification, which enables the identification and interpretation of unreliable estimates. The proposed approach can also be adopted for fitting other physical models through appropriate architectural and simulation adjustments.

eess.IV↗