23 Na MRI in ischemic stroke: Acquisition time reduction using postprocessing with convolutional neural networks.

Quantitative 23 Na magnetic resonance imaging (MRI) provides tissue sodium concentration (TSC), which is connected to cell viability and vitality. Long acquisition times are one of the most challenging aspects for its clinical establishment. K-space undersampling is an approach for acquisition time reduction, but generates noise and artifacts. The use of convolutional neural networks (CNNs) is increasing in medical imaging and they are a useful tool for MRI postprocessing. The aim of this study is 23 Na MRI acquisition time reduction by k-space undersampling. CNNs were applied to reduce the resulting noise and artifacts. A retrospective analysis from a prospective study was conducted including image datasets from 46 patients (aged 72 ± 13 years; 25 women, 21 men) with ischemic stroke; the 23 Na MRI acquisition time was 10 min. The reconstructions were performed with full dataset (FI) and with a simulated dataset an image that was acquired in 2.5 min (RI). Eight different CNNs with either U-Net-based or ResNet-based architectures were implemented with RI as input and FI as label, using batch normalization and the number of filters as varying parameters. Training was performed with 9500 samples and testing included 400 samples. CNN outputs were evaluated based on signal-to-noise ratio (SNR) and structural similarity (SSIM). After quantification, TSC error was calculated. The image quality was subjectively rated by three neuroradiologists. Statistical significance was evaluated by Student's t-test. The average SNR was 21.72 ± 2.75 (FI) and 10.16 ± 0.96 (RI). U-Nets increased the SNR of RI to 43.99 and therefore performed better than ResNet. SSIM of RI to FI was improved by three CNNs to 0.91 ± 0.03. CNNs reduced TSC error by up to 15%. The subjective rating of CNN-generated images showed significantly better results than the subjective image rating of RI. The acquisition time of 23 Na MRI can be reduced by 75% due to postprocessing with a CNN on highly undersampled data.

X-nuclei imaging: Current state, technical challenges, and future directions.

1 H imaging is concerned with contrast generation among anatomically distinct soft tissues. X-nuclei imaging, on the other hand, aims to reveal the underlying changes in the physiological processes on a cellular level. Advanced clinical MR hardware systems improved 1 H image quality and simultaneously enabled X-nuclei imaging. Adaptation of 1 H methods and optimization of both sequence design and postprocessing protocols launched X-nuclei imaging past feasibility studies and into clinical studies. This review outlines the current state of X-nuclei MRI, with the focus on 23 Na, 35 Cl, 39 K, and 17 O. Currently, various aspects of technical challenges limit the possibilities of clinical X-nuclei MRI applications. To address these challenges, quintessential physical and technical concepts behind different applications are presented, and the advantages and drawbacks are delineated. The working process for methods such as quantification and multiquantum imaging is shown step-by-step. Clinical examples are provided to underline the potential value of X-nuclei imaging in multifaceted areas of application. In conclusion, the scope of the latest technical advance is outlined, and suggestions to overcome the most fundamental hurdles on the way into clinical routine by leveraging the full potential of X-nuclei imaging are presented. Level of Evidence: 1 Technical Efficacy Stage: 3 J. Magn. Reson. Imaging 2020;51:355-376.

Comparison of image-based and reconstruction-based respiratory motion correction for golden radial phase encoding coronary MR angiography.

PURPOSE: To evaluate two commonly used respiratory motion correction techniques for coronary magnetic resonance angiography (MRA) regarding their dependency on motion estimation accuracy and final image quality and to compare both methods to the respiratory gating approach used in clinical practice. MATERIALS AND METHODS: Ten healthy volunteers were scanned using a non-Cartesian radial phase encoding acquisition. Respiratory motion was corrected for coronary MRA according to two motion correction techniques, image-based (IMC) and reconstruction-based (RMC) respiratory motion correction. Both motion correction approaches were compared quantitatively and qualitatively against a reference standard navigator-based respiratory gating (RG) approach. Quantitative comparisons were performed regarding visible vessel length, vessel sharpness, and total acquisition time. Two experts carried out a visual scoring of image quality. Additionally, numerical simulations were performed to evaluate the effect of motion estimation inaccuracy on RMC and IMC. RESULTS: RMC led to significantly better image quality than IMC (P's paired Student's t-test were smaller than 0.001 for vessel sharpness and visual scoring). RMC did not show a statistically significant difference compared to reference standard RG (vessel length [99% confidence interval]: 86.913 [83.097-95.015], P = 0.107; vessel sharpness: 0.640 [0.605-0.802], P = 0.012; visual scoring: 2.583 [2.410-3.424], P = 0.018) in terms of vessel visualization and image quality while reducing scan times by 56%. Simulations showed higher dependencies for RMC than for IMC on motion estimation inaccuracies. CONCLUSION: RMC provides a similar image quality as the clinically used RG approach but almost halves the scan time and is independent of subjects' breathing patterns. Clinical validation of RMC is now desirable.