Purpose: To demonstrate robust myelin water fraction (MWF) mapping using an artificial neural network (ANN) with multi-echo gradient-echo (GRE) signal. Methods: Multi-echo gradient-echo signals simulated with a three-pool exponential model were used to generate the training data set for the ANN, which was designed to yield the MWF. We investigated the performance of our proposed ANN for various conditions using both numerical simulations and in vivo data. Simulations were conducted with various SNRs to investigate the performance of the ANN. In vivo data with high spatial resolutions were applied in the analyses, and results were compared with MWFs derived by the nonlinear least-squares algorithm using a complex three-pool exponential model. Results: The network results for the simulations show high accuracies against noise compared with nonlinear least-squares MWFs: RMS-error value of 5.46 for the nonlinear least-squares MWF and 3.56 for the ANN MWF at an SNR of 150 (relative gain = 34.80%). These effects were also found in the in vivo data, with reduced SDs in the region-of-interest analyses. These effects of the ANN demonstrate the feasibility of acquiring high-resolution myelin water images. Conclusion: The simulation results and in vivo data suggest that the ANN facilitates more robust MWF mapping in multi-echo gradient-echo sequences compared with the conventional nonlinear least-squares method. K E Y W O R D S artificial neural network, multi-echo gradient echo, myelin water imaging, T * 2 distribution [Correction added after online publication 18 August, 2020. The authors have corrected the spelling of author name Won-Jin Moon.] | 381 JUNG et al. 1 | INTRODUCTION Myelin water fraction (MWF) as a method for measuring quantitative myelin signals has demonstrated potential to diagnose various demyelinating diseases such as multiple sclerosis, schizophrenia, and stroke. 1,2 Conventional myelin water imaging (MWI) uses multi-echo spin-echo acquisition and nonnegative least-squares estimation, 3,4 whereas more recently multi-echo gradient echo (mGRE) has been suggested. 5-9 These methods provide benefits such as faster acquisition time and lower specific absorption rates. Several studies have proposed methods to acquire high-quality MWI data using mGRE. Such studies suggest applying the nonlinear least-squares (NLLS) algorithm to the acquired signal using a defined model, such as the three-pool exponential model. 8,10 These methods are based on the assumption that the white-matter (WM) water can be reliably modeled by three-pool exponential components with individual frequency shifts. 5,6,8 These methods can be further improved by physiological noise compensation 11 and B 0 field inhomogeneity correction. 10,12,13 Despite these developments, there are still challenges in improving the accuracy and robustness of the MWF. The NLLS (used for estimating MWFs) has been reported to be inaccurate and unstable, especially at low-to-moderate SNRs. 14-16 This requires high SNR data acquisition for MWFs, limiting the scan...

With the development of deep-learning techniques, the application of deep learning in MR imaging processing seems to be growing. Accordingly, deep learning has also been introduced in motion correction and seemed to work as well as do conventional motion-compensation methods. In this article, we review the motion-correction methods based on deep learning, focusing especially on the motion-simulation methods adopted. We then propose a new motion-simulation tool, which we call view2Dmotion.

To denoise B + 1 phase using a deep learning method for phase-based in vivo electrical conductivity reconstruction in a 3T MR system. Methods: For B + 1 phase deep-learning denoising, a convolutional neural network (U-net) was chosen. Training was performed on data sets from 10 healthy volunteers. Input data were the real and imaginary components of single averaged spin-echo data (SNR = 45), which was used to approximate the B + 1 phase. For label data, multiple signal-averaged spin-echo data (SNR = 128) were used. Testing was performed on in silico and in vivo data. Reconstructed conductivity maps were derived using phase-based conductivity reconstructions. Additionally, we investigated the usability of the network to various SNR levels, imaging contrasts, and anatomical sites (ie, T 1 , T 2 , and proton density-weighted brain images and proton density-weighted breast images. In addition, conductivity reconstructions from deep learning-based denoised data were compared with conventional image filters, which were used for data denoising in electrical properties tomography (ie, the Gaussian filtering and the Savitzky-Golay filtering). Results: The proposed deep learning-based denoising approach showed improvement for B + 1 phase for both in silico and in vivo experiments with reduced quantitative error measures compared with other methods. Subsequently, this resulted in an improvement of reconstructed conductivity maps from the denoised B + 1 phase with deep learning. Conclusion:The results suggest that the proposed approach can be used as an alternative preprocessing method to denoise B + 1 maps for phase-based conductivity reconstruction without relying on image filters or signal averaging. K E Y W O R D SB + 1 phase, deep learning, denoising, electrical properties tomography, phase-based conductivity reconstruction | 2085 JUNG et al.

Purpose: Previously, an artificial neural network method was introduced to estimate quantitative myelin water fraction (MWF) using multi-echo gradient-echo data. However, the fiber orientation of white matter with respect to B 0 could bias the quantification of MWF. Here, we developed an advanced workflow for MWF estimation that could improve the quantification of MWF. Methods:To adopt fiber orientation effects, a complex-valued neural network with complex-valued operation was used. In addition, to compensate for the bias from different scan parameters, a signal model incorporating the T 1 value was devised for training data generation. At the testing stage, a voxel-spread function approach was utilized for spatial B 0 artifact correction. Finally, dropout-based variational inference was implemented for uncertainty estimates on the network model to provide a confidence interpretation of the output. Results: According to simulation and in vivo analysis, the proposed method suggests improved quality of MWF estimation by correcting the bias and artifacts.The proposed complex-valued neural network approach can alleviate the dependency of fiber orientation effects compared to previous artificial neural network method. Uncertainty estimates provides information different from fitting error that can be used as a confidence level of the resulting MWF values. Conclusion: An improved MWF mapping using complex-valued neural network analysis has been proposed.

No abstract