Constrained projection reconstruction for reduced encoding MR diffusion tensor imaging

MR diffusion tensor imaging (DTI) is a promising tool for characterizing microstructures of ordered tissues. However, its practical applications have been hampered by the low signal-to-noise-ratio and spatial and temporal resolutions. K-space sharing with constrained reconstruction has been effective for accelerating DTI, though the implementation was based on rectilinear k-space sampling. Due to its intrinsic oversampling and averaging of central k-space and isotropic trajectories, projection reconstruction, or radial sampling, imaging may be more suited for k-space sharing reduced encoding schemes. In this study, filtered reduced encoding projection reconstruction (FREPR) is implemented. The key features include the use of radial mask filters to select for spatial frequency information and constrained reconstruction in the form of phase and magnitude baseline correction. The performance of FREPR is evaluated in terms of image differences and deviations in measured fiber orientation with respect to the full-encoded dataset, both in simulation and experiment, against rectilinear k-space sharing method. Results indicate FREPR provides a significant improvement in the acquisition-time efficiency of DTI.