Improved signal‐to‐noise performance of MultiNet GRAPPA ¹H FID MRSI reconstruction with semi‐synthetic calibration data

Abstract

To further develop MultiNet GRAPPA, a neural-network-based reconstruction, for lower SNR proton MRSI ( 1 H MRSI) data using adapted undersampling schemes and improved training sets. Methods:1 H FID-MRSI data and an anatomical image for GRAPPA reconstruction were acquired in two slices in the human brain (n = 6) at 7T. MRSI data were retrospectively undersampled for a 4×, 6×, and 7× acceleration rate. Signal-to-noise, relative error (RE) between accelerated and fully sampled metabolic maps, RMS of the lipid artifacts, and fitting reliability were compared across acceleration rates, to the fully sampled data, and with different kinds and amounts of training images. Results: Training with semi-synthetic images resulted in higher SNR and lower lipid RMS relative to training with acquired images from one or several subjects. SNR increased with the number of semi-synthetic training images and the 4× accelerated data retains ∼30% more SNR than other accelerated data. Spectra reconstructed with 20 semi-synthetic averages retained ∼100% more SNR and had ∼5% lower lipid RMS than those reconstructed with the center k-space points of one image as was originally proposed for very high SNR MRSI data and had higher fitting reliability. The metabolite RE was lowest when training with 20-semi-synthetic training images and highest when training with the center k-space points of one image. Conclusion: MultiNet GRAPPA is feasible with lower SNR 1 H MRSI data if 20-semi-synthetic training images are used at a 4× acceleration rate. This acceleration rate provided the best trade-off between scan time and spectral SNR.

Glutathione in the Pons Is Associated With Clinical Status Improvements in Subacute Spinal Cord Injury

“…Future studies might be conducted on ultra high field scanners that improve both imaging of the spinal cord [44][45][46] and provide access to high-resolution MRSI 47,48 enabling molecular insights.…”

Magnetic Resonance Spectroscopy for Cervical Cancer: Review and Potential Prognostic Applications

“…MRSI has been severely under-utilized and was only used in two studies; furthermore, metabolic maps of the different biochemicals were not present in the literature. There have been significant advances in the field of MRSI, mostly facilitated by various acceleration methods including the shortening of repetition times [ 66 ], under-sampling of k-space [ 67 , 68 , 69 ], and spatial-spectral encoding [ 70 ]. Metabolic maps may allow for better tumor delineation, may serve a prognostic role, or may help guide therapy.…”

A review of machine learning applications for the proton MR spectroscopy workflow

“…Additionally, the authors propose variable density under‐sampling schemes to achieve even higher acceleration factors and alter their ML framework by first using 2‐voxel cross‐neighbor and adjacent‐neighbor NNs before using the previously mentioned 1‐voxel neighbor NNs. Although in this work the networks are trained in a subject‐specific manner, various strategies are investigated in Reference 30 to improve this approach with more samples. The results suggest the use of NNs for GRAPPA reconstruction reduces aliasing artifacts thereby positively impacting metabolite concentration maps and significantly boosting the performance compared to regular GRAPPA.…”