This study aims to improve upon Self-consistent Robust Artificial-neural-networks for k-space Interpolation (sRAKI), which is a deep learning-based parallel imaging technique for accelerated MRI reconstruction. The proposed technique, called sRAKI-RNN, combines the calibration and reconstruction phases of sRAKI into a single step that jointly learns the self-consistency rule and performs iterative reconstruction using recurrent neural networks (RNN). Similar to sRAKI, sRAKI-RNN supports arbitrary undersampling patterns and is a databasefree technique that is trained on autocalibrating signal (ACS) data from the same scan. Densely connected blocks are used in each iteration of the RNN to improve the convergence during the learning phase. sRAKI-RNN was evaluated on targeted right coronary artery (RCA) MRI. The results indicate that sRAKI-RNN further improves the noise resilience of sRAKI in a shorter running time and also considerably outperforms its linear counterpart, SPIRiT, in suppressing reconstruction noise.
Proc. SPIE. 6142, Medical Imaging 2006: Physics of Medical Imaging
KEYWORDS: Signal to noise ratio, Magnetic resonance imaging, Signal detection, Calibration, Computer programming, Functional magnetic resonance imaging, Brain, Image resolution, Image segmentation, Data acquisition
Multi-channel acquisition is employed in MRI to decrease total imaging time. In this paper, artifact free images are calculated by utilizing the difference in spatial encoding of the MR signal from neighboring channels. The encoding functions are estimated in the presence of noise and motion. For fMRI studies, the temporal stability of the signal is essential, since neuronal activity in the brain is detected by probing subtle BOLD (blood oxygen level dependent) signal changes. To ensure artifact free noise representation a new type of weight is used. By effectively selecting and eliminating low SNR pixels, increased temporal stability is achieved. Using the parallel imaging method SENSE the proposed method is tested with in-vivo data to ensure noise suppression and demonstrate correct assignment of fMRI activation.