Deep Learning-Driven Inversion Framework for Shear Modulus Estimation in Magnetic Resonance Elastography (DIME)
Preprint
Abstract
The Multimodal Direct Inversion (MMDI) algorithm is widely used in Magnetic Resonance Elastography (MRE) to estimate tissue shear stiffness. However, MMDI relies on the Helmholtz equation, which assumes wave propagation in a uniform, homogeneous, and infinite medium. Furthermore, the use of the Laplacian operator makes MMDI highly sensitive to noise, which compromises the accuracy and reliability of stiffness estimates. In this study, we propose the Deep-Learning driven Inversion Framework for Shear Modulus Estimation in MRE (DIME), aimed at enhancing the robustness of inversion. DIME is trained on the displacement fields-stiffness maps pair generated through Finite Element Modelling (FEM) simulations. To capture local wave behavior and improve robustness to global image variations, DIME is trained on small image patches. We first validated DIME using homogeneous and heterogeneous datasets simulated with FEM, where DIME produced stiffness maps with low inter-pixel variability, accurate boundary delineation, and higher correlation with ground truth (GT) compared to MMDI. Next, DIME was evaluated in a realistic anatomy-informed simulated liver dataset with known GT and compared directly to MMDI. DIME reproduced ground-truth stiffness patterns with high fidelity (r = 0.99, R^2 = 0.98), while MMDI showed greater underestimation. After validating DIME on synthetic data, we tested the model in in vivo liver MRE data from eight healthy and seven fibrotic subjects. DIME preserved physiologically consistent stiffness patterns and closely matched MMDI, which showed directional bias. Overall, DIME showed higher correlation with ground truth and visually similar stiffness patterns, whereas MMDI displayed a larger bias that can potentially be attributed to directional filtering. These preliminary results highlight the feasibility of DIME for clinical applications in MRE.Keywords
All Related Versions
This publication has 31 references indexed in Scilit:
- Deep learning for accelerated and robust MRI reconstructionMagnetic Resonance Materials in Physics, Biology and Medicine, 2024
- Deep learning for medical image segmentation: State-of-the-art advancements and challengesInformatics in Medicine Unlocked, 2024
- Physics-Informed Neural Networks for Tissue Elasticity Reconstruction in Magnetic Resonance Elastography2023
- Direct Comparison of US and MR Elastography for Staging Liver Fibrosis in Patients With Nonalcoholic Fatty Liver DiseaseClinical Gastroenterology and Hepatology, 2022
- Deep Image PriorInternational Journal of Computer Vision, 2020
- Artificial neural networks for stiffness estimation in magnetic resonance elastographyMagnetic Resonance in Medicine, 2017
- Comparison of gradient-recalled echo and spin-echo echo-planar imaging MR elastography in staging liver fibrosis: a meta-analysisEuropean Radiology, 2017
- Magnetic resonance elastography: evaluation of new inversion algorithm and quantitative analysis methodAbdominal Radiology, 2015
- Mechanical Properties of Tissue1990
- A flow velocity zeugmatographic interlace for NMR imaging in humansMagnetic Resonance Imaging, 1982