Rapid mono and biexponential 3D-T1ρ mapping of knee cartilage using variational networks

Abstract

In this study we use undersampled MRI acquisition methods to obtain accelerated 3D mono and biexponential spin–lattice relaxation time in the rotating frame (T1ρ) mapping of knee cartilage, reducing the usual long scan time. We compare the accelerated T1ρ maps obtained by deep learning-based variational network (VN) and compressed sensing (CS). Both methods were compared with spatial (S) and spatio-temporal (ST) filters. Complex-valued fitting was used for T1ρ parameters estimation. We tested with seven in vivo and six synthetic datasets, with acceleration factors (AF) from 2 to 10. Median normalized absolute deviation (MNAD), analysis of variance (ANOVA), and coefficient of variation (CV) were used for analysis. The methods CS-ST, VN-S, and VN-ST performed well for accelerating monoexponential T1ρ mapping, with MNAD around 5% for AF = 2, which increases almost linearly with the AF to an MNAD of 13% for AF = 8, with all methods. For biexponential mapping, the VN-ST was the best method starting with MNAD of 7.4% for AF = 2 and reaching MNAD of 13.1% for AF = 8. The VN was able to produce 3D-T1ρ mapping of knee cartilage with lower error than CS. The best results were obtained by VN-ST, improving CS-ST method by nearly 7.5%.

Optimization of spin-lock times for T1ρ mapping of human knee cartilage with bi- and stretched-exponential models

“…Future studies include the use of variational networks 40 , 41 proved to be capable of reconstructing images from undersampled k-space data. The use of this kind of network along with another network responsible for data fitting, such as Recurrent Inference Machines (RIM) 42 , can be used to form an end-to-end network for quantitative parametric mapping.…”

CoRRECT: A Deep Unfolding Framework for Motion-Corrected Quantitative R2* Mapping

“…Recent work has also focused on DL-based image reconstruction methods for improving qMRI estimation. The qMRI maps can be computed from these reconstructions either using standard fitting method [26,27] or DL-based mapping [28]. The idea of combining the DL-based MRI reconstruction and qMRI estimation into a single imaging pipeline trainable end-to-end was also considered in [61].…”

“…Despite the rich literature on qMRI, the majority of work in the area has considered separately artifacts due to accelerated data acquisition, involuntary physical motion, and magnetic-field inhomogeneities. In particular, it is common to view qMRI parameter estimation as a post-processing step decoupled from the MRI reconstruction [26][27][28]. We address this issue by presenting a new unified qMRI framework-called Co-design of MRI Reconstruction and R * 2 Estimation with Correction for Motion (CoRRECT)-for recovering high-quality quantitative R * 2 maps directly from noisy, subsampled, and artifact-corrupted MRI measurements.…”

A Fast CS-Based Reconstruction Model with Total Variation Constraint for MRI Enhancement in K-Space Domain

“…As we all know, accelerating the speed of imaging has always been the focus of Magnetic Resonance Imaging research, which can not only improve the efficiency of MRI system, but also weaken or eliminate motion artifacts and noise, so that MRI can be better applied to medical diagnosis [6,[11][12][13]]. In MRI system, there are generally two methods to shorten the time of data acquisition: one is to use multichannel parallel imaging technology; the other is to use non-Cartesian sampling trajectory to fill K-space, such as radial subsampling and sparse sampling [14][15][16][17].…”