Purpose
To introduce a blind compressed sensing (BCS) framework to accelerate multi-parameter MR mapping, and demonstrate its feasibility in high-resolution, whole-brain T1ρ and T2 mapping.
Methods
BCS models the evolution of magnetization at every pixel as a sparse linear combination of bases in a dictionary. Unlike compressed sensing (CS), the dictionary and the sparse coefficients are jointly estimated from under-sampled data. Large number of non-orthogonal bases in BCS accounts for more complex signals than low rank representations. The low degree of freedom of BCS, attributed to sparse coefficients, translates to fewer artifacts at high acceleration factors(R).
Results
From 2D retrospective under-sampling experiments, the mean square errors in T1ρ and T2 maps were observed to be within 0.1% up to R=10. BCS was observed to be more robust to patient-specific motion as compared to other CS schemes and resulted in minimal degradation of parameter maps in the presence of motion. Our results suggested that BCS can provide an acceleration factor of 8 in prospective 3D imaging with reasonable reconstructions.
Conclusion
BCS considerably reduces scan time for multi-parameter mapping of the whole brain with minimal artifacts, and is more robust to motion-induced signal changes compared to current CS and PCA based techniques.
The proposed sparsity constrained spectral deconvolution scheme is effective in minimizing the line-shape distortions. The dual resolution reconstruction scheme is capable of minimizing spectral leakage artifacts.
Simultaneous multi-slice acquisitions are essential for modern neuroimaging research, enabling high temporal resolution functional and high-resolution q-space sampling diffusion acquisitions. Recently, deep learning reconstruction techniques have been introduced for unaliasing these accelerated acquisitions, and robust artificial-neural-networks for k-space interpolation (RAKI) have shown promising capabilities. This study systematically examines the impacts of hyperparameter selections for RAKI networks, and introduces a novel technique for training data generation which is analogous to the split-slice formalism used in slice-GRAPPA. Methods: RAKI networks were developed with variable hyperparameters and with and without split-slice training data generation. Each network was trained and applied to five different datasets including acquisitions harmonized with Human Connectome Project lifespan protocol. Unaliasing performance was assessed through L1 errors computed between unaliased and calibration frequency-space data. Results: Split-slice training significantly improved network performance in nearly all hyperparameter configurations. Best unaliasing results were achieved with three layer RAKI networks using at least 64 convolutional filters with receptive fields of 7 voxels, 128 single-voxel filters in the penultimate RAKI layer, batch normalization, and no training dropout with the split-slice augmented training dataset. Networks trained without the split-slice technique showed symptoms of network over-fitting. Conclusions: Split-slice training for simultaneous multi-slice RAKI networks positively impacts network performance. Hyperparameter tuning of such reconstruction networks can lead to further improvements in unaliasing performance.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.