Purpose To increase the effectiveness of respiratory gating in radial stack‐of‐stars MRI, particularly when imaging at high spatial resolutions or with multiple echoes. Methods Free induction decay (FID) navigators were integrated into a three‐dimensional gradient echo radial stack‐of‐stars pulse sequence. These navigators provided a motion signal with a high temporal resolution, which allowed single‐spoke binning (SSB): each spoke at each phase encode step was sorted individually to the corresponding motion state of the respiratory signal. SSB was compared with spoke‐angle binning (SAB), in which all phase encode steps of one projection angle were sorted without the use of additional navigator data. To illustrate the benefit of SSB over SAB, images of a motion phantom and of six free‐breathing volunteers were reconstructed after motion‐gating using either method. Image sharpness was quantitatively compared using image gradient entropies. Results The proposed method resulted in sharper images of the motion phantom and free‐breathing volunteers. Differences in gradient entropy were statistically significant (p = 0.03) in favor of SSB. The increased accuracy of motion‐gating led to a decrease of streaking artifacts in motion‐gated four‐dimensional reconstructions. To consistently estimate respiratory signals from the FID‐navigator data, specific types of gradient spoiler waveforms were required. Conclusion SSB allowed high‐resolution motion‐corrected MR imaging, even when acquiring multiple gradient echo signals or large acquisition matrices, without sacrificing accuracy of motion‐gating. SSB thus relieves restrictions on the choice of pulse sequence parameters, enabling the use of motion‐gated radial stack‐of‐stars MRI in a broader domain of clinical applications.

In in vivo 1H-MRSI of the prostate, small matrix sizes can cause voxel bleeding extending to regions far from a voxel, dispersing a signal of interest outside that voxel and mixing extra-prostatic residual lipid signals into the prostate. To resolve this problem, we developed a three-dimensional overdiscretized reconstruction method. Without increasing the acquisition time from current 3D MRSI acquisition methods, this method is aimed to improve the localization of metabolite signals in the prostate without compromising on SNR. The proposed method consists of a 3D spatial overdiscretization of the MRSI grid, followed by noise decorrelation with small random spectral shifts and weighted spatial averaging to reach a final target spatial resolution. We successfully applied the three-dimensional overdiscretized reconstruction method to 3D prostate 1H-MRSI data at 3T. Both in phantom and in vivo, the method proved to be superior to conventional weighted sampling with Hamming filtering of k-space. Compared with the latter, the overdiscretized reconstructed data with smaller voxel size showed up to 10% less voxel bleed while maintaining higher SNR by a factor of 1.87 and 1.45 in phantom measurements. For in vivo measurements, within the same acquisition time and without loss of SNR compared with weighted k-space sampling and Hamming filtering, we achieved increased spatial resolution and improved localization in metabolite maps.

Motion-compensated images can be created from motion-binned undersampled radial stack-of-stars data through compressed sensing and image registration. However, for long repetition times or for many partitions, the acquisition time for one radial projection with all phase-encode steps becomes too long to sample the motion via self-gating, which leads to motion artifacts. Therefore, we estimate motion from FID-navigators and perform binning on a single-readout level to gain higher spatiotemporal resolutions. Our methods are tested on a motion phantom and volunteer with gridding and motion-compensated reconstructions. Our results show accurate detection of the motion signal and reduced motion blur in reconstructions.

Undersampled k-space data reconstruction results in aliasing artifacts. Compressed sensing theory enables image reconstruction by using a priori knowledge in the form of regularization. Increasingly, Machine Learning methods are used to learn the regularization from data itself, but these methods can result in unstable reconstructions. We propose a translation equivariant single-layer neural network for reconstruction of radially measured k-space data. By exploiting translation symmetry, it can learn from randomly simulated data while still being applicable to in-vivo measurements. We tested robustness to small perturbations and reliability of the reconstruction of unexpected objects.