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%.
Highlights
Quantitative mapping using the spin–lattice relaxation time in the rotating frame (T1ρ) has shown to be useful for early detection of osteoarthritis (OA)[1], since T1ρ mapping is sensitive to the proteoglycan content of the cartilage[2]
The accelerated methods are compared against the reference, since no ground truth (GT) is known for all knee cartilage images
The use of variational network (VN) is advantageous over compressive sensing (CS), even though both approaches provide very satisfying quality for most acceleration factors (AF)
Summary
Quantitative mapping using the spin–lattice relaxation time in the rotating frame (T1ρ) has shown to be useful for early detection of osteoarthritis (OA)[1], since T1ρ mapping is sensitive to the proteoglycan content of the cartilage[2]. The VN uses relatively fast algorithmic implementation based on convolutional layers, enabling faster reconstructions than typical iterative algorithms used in CS This comes at the cost of formulating the image reconstruction problem into a highly non-linear and non-convex optimization problem. This opens the question if the found local minimum generalizes to different types of data. We compare the VN, trained with real and synthetically generated knee cartilage images, against CS approaches for mono and biexponential T1ρ mapping[11,12]. After reconstruction, complexvalued fitting is used to find the T1ρ mapping parameters for mono and biexponential models
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
