SepUnet: Depthwise Separable Convolution Integrated U-Net For MRI Reconstruction

Super-resolution reconstruction of single anisotropic 3D MR images using residual convolutional neural network

Radial encoding is often performed for fast data acquisition in magnetic resonance imaging (MRI). Therefore, the gridding method, which is a kind of algorithm for Fast Fourier Transform, is widely used in fast MRI reconstruction. In this paper, an improved gridding method has been utilized to 2D MRI reconstruction, which precomputes an 1D finely sampled table of a high-accuracy interpolator and then uses separable bilinear interpolation by look-up table. Experimental comparisons are made by assessing the performance of the direct reconstruction, the conventional KB gridding and the proposed method. The experimental results show that the proposed method provides high accuracy, fast computation, and dramatic memory saving.

In this paper, we present a fast iterative magnetic resonance imaging (MRI) reconstruction algorithm taking advantage of the prevailing GPGPU programming paradigm. In clinical environment, MRI reconstruction is usually performed via fast Fourier transform (FFT). However, imaging artifacts (i.e. signal loss) resulting from susceptibility-induced magnetic field inhomogeneities degrade the quality of reconstructed images. These artifacts must be addressed using accurate modeling of the physics of the system coupled with iterative reconstruction. We have developed a reconstruction algorithm with improved image quality at the expense of computation time and hence an implementation on GPUs achieving significant speedup. In this work, we extend our previous work on GPU implementation by adding several new features. First, we enable Sensitivity Encoding for Fast MRI (SENSE) reconstruction (from data acquired using a multi-receiver coil array) which can reduce the acquisition time. Besides, we have implemented a GPU-based total variation regularization in our SENSE reconstruction framework. In this paper, we describe the different optimizations employed from levels of algorithm, program code structures, and specific architecture performance tuning, featuring both our MRI reconstruction algorithm and GPU hardware specifics. Results show that the current GPU implementation produces accurate image estimates while significantly accelerating the reconstruction.

Magnetic resonance imaging (MRI) acquisition is an inherently slow process whose acceleration has been the subject of much investigation. In recent years, the explosive advance of deep learning techniques for computer vision and image reconstruction has led to the investigation of deep neural networks for the reconstruction of MRI with under-sampled k-space. In this work, we propose a new image domain architecture that directly produces a sum-of-squares image from under-sampled multi-coil MRI acquisition. This model, called BarbellNet, is a fully convolutional neural network architecture that utilizes the channel attention mechanism using the residual channel attention block (RCAB). Through extensive experiments with the fastMRI data set, we confirm the efficacy of BarbellNet.

Deep learning-based magnetic resonance imaging (MRI) reconstruction obtains extremely high reconstruction performance by training with large amounts of data. However, the acquisition of fully sampled MRI data is expensive and time-consuming, making it difficult to obtain large amounts of fully sampled data. This not only poses a challenge for models that require more fully sampled data for training but also limits the generalization ability of network models. To address the above problems, we propose a semi-supervised MRI reconstruction algorithm based on migration learning by analyzing the characteristics of under-sampled MRI reconstruction. To address the lack of data scarcity and generalization in the reconstruction process. The model fuses shallow features extracted from the convolutional layer and deep features obtained from the Swin Transformer Block (STB) to greatly improve the reconstruction capability of the network. However, the process of distillation cannot predict the goodness of the knowledge transferred, which hinders the performance of the semi-supervised algorithm. We further propose the use of privilege loss as a way to improve the distillation of useful knowledge. Experimental results show that student networks trained by this algorithm are able to rival supervised teacher networks and outperform unsupervised network models in brain data reconstruction.KeywordsMagnetic resonance imagingSwin transformerKnowledge distillationPrivileged information

This paper introduces an efficient algorithm for magnetic resonance (MR) image reconstruction. The proposed method minimizes a linear combination of nonlocal total variation and least-square data-fitting term to reconstruct the MR images from undersampled k-space data. The nonlocal total variation is taken as the L 1-regularization functional and solved using Split Bregman iteration. The proposed algorithm is compared with previous methods in terms of the reconstruction accuracy and computational complexity. The comparison results demonstrate the superiority of the proposed algorithm for compressed MR image reconstruction.

In many in vivo magnetic resonance imaging (MRI) acquisitions, the image resolution is low and induces considerable segmentation errors with a poor and limited visual interpretation. Acquiring high-resolution images reduces the patient comfort and results in considerable cost because of an increase of the acquisition time. We developed a progressive technique based on multidimensional dual kriging to interpolate 2D and 3D multi-parametric MRI data. The objectives were to (1) optimise kriging parameters for 2D images and 3D volumes interpolation, (2) validate the technique in increasing the image resolution and (3) provide a direct application to 3D non-parametric volumes interpolation and image filtering. We carried out two multi-parametric MRI acquisitions on a bovine tail segment including intervertebral discs from the second to the fourth caudal vertebrae with two different spatial resolutions. We adopted the dual kriging formulation and introduced the technique of progressive kriging, which consists in kriging the given Cartesian data region-by-region. Kriging parameters were optimised to minimise the relative errors. For each MRI sequence, kriging was compared with other interpolation techniques in terms of signal mean within the region of interest, including the zero-padding, the nearest-neighbour, the bilinear and the bicubic techniques. Optimal kriging parameters for transverse relaxation time maps interpolation were a cubic covariance function and a distance of influence of only 1–8 pixels. Our technique was found to be efficient to interpolate the longitudinal and transverse relaxation times, the magnetisation transfer ratio and the apparent diffusion coefficient maps, compared with zero-padding, nearest-neighbour, bilinear and bicubic techniques. Kriging provided relatively high performance and low error, whereas image edges tend to be corrected and better defined compared with other techniques. Progressive dual kriging is a flexible technique that can be optimised to interpolate multidimensional data based on the signal distribution and should open doors to many future applications in medical imaging such as diffusion MRI, image filtering and non-parametric 3D volume reconstruction.

Compressive sensing magnetic resonance (MR) imaging is aimed at achieving high-quality MR image reconstruction by undersampling K-space data. It is crucial to explore prior information since compressive sensing MR imaging relies heavily on some prior assumptions, such as signal's sparse property. In this study, in order to explore the prior information fully, an improved MR image reconstruction model based on compressive sensing theory is proposed, named reference image MR imaging with adaptive tight frame. In the proposed model, an adaptive tight frame is involved to explore the sparse prior information adapt to MR images and the similarity prior information to the target image. Meanwhile, improved adaptive weighting parameters are used to trade off the sparsity between the regions with much similarity and that of little similarity. In addition, the smoothing-based fast iterative shrinkage-threshold algorithm is utilised to tackle the optimisation problem so as to speed up imaging. The experimental results demonstrate that the proposed MR image reconstruction method outperforms some state-of-the-art methods in terms of quantitative results.

An approach to Magnetic Resonance (MR) image reconstruction from undersampled data is proposed. Undersampling artifacts are removed using an iterative thresholding algorithm applied to nonlinearly transformed image block arrays. Each block array is transformed using kernel principal component analysis where the contribution of each image block to the transform depends in a nonlinear fashion on the distance to other image blocks. Elimination of undersampling artifacts is achieved by conventional principal component analysis in the nonlinear transform domain, projection onto the main components and back-mapping into the image domain. Iterative image reconstruction is performed by interleaving the proposed undersampling artifact removal step and gradient updates enforcing consistency with acquired k-space data. The algorithm is evaluated using retrospectively undersampled MR cardiac cine data and compared to k-t SPARSE-SENSE, block matching with spatial Fourier filtering and k-t ℓ1-SPIRiT reconstruction. Evaluation of image quality and root-mean-squared-error (RMSE) reveal improved image reconstruction for up to 8-fold undersampled data with the proposed approach relative to k-t SPARSE-SENSE, block matching with spatial Fourier filtering and k-t ℓ1-SPIRiT. In conclusion, block matching and kernel methods can be used for effective removal of undersampling artifacts in MR image reconstruction and outperform methods using standard compressed sensing and ℓ1-regularized parallel imaging methods.

Fast magnetic resonance imaging (MRI) is crucial for clinical applications that can alleviate motion artefacts and increase patient throughput. K-space undersampling is an obvious approach to accelerate MR acquisition. However, undersampling of k-space data can result in blurring and aliasing artefacts for the reconstructed images. Recently, several studies have been proposed to use deep learning-based data-driven models for MRI reconstruction and have obtained promising results. However, the comparison of these methods remains limited because the models have not been trained on the same datasets and the validation strategies may be different. The purpose of this work is to conduct a comparative study to investigate the generative adversarial network (GAN)-based models for MRI reconstruction. We reimplemented and benchmarked four widely used GAN-based architectures including DAGAN, ReconGAN, RefineGAN and KIGAN. These four frameworks were trained and tested on brain, knee and liver MRI images using twofold, fourfold and sixfold accelerations, respectively, with a random undersampling mask. Both quantitative evaluations and qualitative visualization have shown that the RefineGAN method has achieved superior performance in reconstruction with better accuracy and perceptual quality compared to other GAN-based methods. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

The deep learning method is an efficient solution for improving the quality of undersampled magnetic resonance (MR) image reconstruction while reducing lengthy data acquisition. Most deep learning methods neglect the mutual constraints between the real and imaginary components of complex-valued k-space data. In this paper, a new complex-valued convolutional neural network, namely, Dense-U-Dense Net (DUD-Net), is proposed to interpolate the undersampled k-space data and reconstruct MR images. The proposed network comprises dense layers, U-Net, and other dense layers in sequence. The dense layers are used to simulate the mutual constraints between real and imaginary components, and U-Net performs feature sparsity and interpolation estimation for k-space data. Two MRI datasets were used to evaluate the proposed method: brain magnitude-only MR images and knee complex-valued k-space data. Several operations were conducted for data preprocessing. First, the complex-valued MR images were synthesized by phase modulation on magnitude-only images. Second, a radial trajectory based on the golden angle was used for k-space undersampling, whereby a reversible normalization method was proposed to balance the distribution of positive and negative values in k-space data. The optimal performance of DUD-Net was demonstrated based on a quantitative evaluation of inter-method and intra-method comparisons. When compared with other methods, significant improvements were achieved, PSNRs were increased by 10.78 and 5.74dB, whereas RMSEs were decreased by 71.53% and 30.31% for magnitude and phase image, respectively. It is concluded that DUD-Net significantly improves the performance of MR image reconstruction.

Regularized iterative reconstruction algorithms for imaging inverse problems require selection of appropriate regularization parameter values. We focus on the challenging problem of tuning regularization parameters for nonlinear algorithms for the case of additive (possibly complex) Gaussian noise. Generalized cross-validation (GCV) and (weighted) mean-squared error (MSE) approaches (based on Steinfs Unbiased Risk Estimate. SURE) need the Jacobian matrix of the nonlinear reconstruction operator (representative of the iterative algorithm) with respect to the data. We derive the desired Jacobian matrix for two types of nonlinear iterative algorithms: a fast variant of the standard iterative reweighted least-squares method and the contemporary split-Bregman algorithm, both of which can accommodate a wide variety of analysis- and synthesis-type regularizers. The proposed approach iteratively computes two weighted SURE-type measures: Predicted-SURE and Projected-SURE (that require knowledge of noise variance Ð2), and GCV (that does not need Ð2) for these algorithms. We apply the methods to image restoration and to magnetic resonance image (MRI) reconstruction using total variation (TV) and an analysis-type .1-regularization. We demonstrate through simulations and experiments with real data that minimizing Predicted-SURE and Projected-SURE consistently lead to near-MSE-optimal reconstructions. We also observed that minimizing GCV yields reconstruction results that are near-MSE-optimal for image restoration and slightly suboptimal for MRI. Theoretical derivations in this work related to Jacobian matrix evaluations can be extended, in principle, to other types of regularizers and reconstruction algorithms.

The aim of this study was to investigate the impact of a deep learning-based superresolution reconstruction technique for T1-weighted volume-interpolated breath-hold examination (VIBESR) on image quality in comparison with standard VIBE images (VIBESD). Between May and August 2020, a total of 46 patients with various abdominal pathologies underwent contrast-enhanced upper abdominal VIBE magnetic resonance imaging (MRI) at 1.5 T. After data acquisition, the precontrast and postcontrast T1-weighted VIBE raw data were processed by a deep learning-based prototype algorithm for deblurring and denoising the images as well as for enhancing their sharpness (VIBESR). In a randomized and blinded manner, 2 radiologists independently analyzed the image data sets using the unprocessed images VIBESD as a standard reference. Outcome measures were as follows: overall image quality, anatomic clarity of organ borders, sharpness of vessels, artifacts, noise, and diagnostic confidence. All ratings were performed on an ordinal 4-point Likert scale. If the MRI examination encompassed a hepatic lesion, the maximum diameter of the largest hepatic lesion was quantified, and lesion sharpness and conspicuity were evaluated on an ordinal 4-point Likert scale. In addition, a post hoc regression analysis for lesion evaluation was computed. Finally, interrater/intrarater agreement was analyzed. The overall image quality, anatomic clarity of organ borders, and sharpness of vessels in both precontrast and postcontrast images were rated significantly higher in VIBESR than in VIBESD (P < 0.001). Similarly, diagnostic confidence was higher in VIBESR than in VIBESD (P < 0.001). Furthermore, VIBESR images were rated to have significantly less noise and fewer artifacts in comparison with VIBESD (P < 0.001). The interreader agreement was substantial with a Cohen κ of 0.72 for the precontrast analysis and a κ of 0.74 for the postcontrast analysis. A total of 28 hepatic lesions were analyzed. For both readers, lesion sharpness and conspicuity were rated significantly better in VIBESR than in VIBESD in both the precontrast and postcontrast data sets (P < 0.01), which was consistent with the post hoc regression analysis (for every 1-point increase in sharpness/conspicuity, the odds ratio revealed a positive relation with VIBESR of 13-fold to 17-fold in comparison with VIBESD; P < 0.001). In terms of lesion size, there was no significant difference between the precontrast VIBESD and VIBESR or between the postcontrast VIBESD and VIBESR for both readers. Similarly, there was an excellent interreader agreement regarding lesion size (intraclass correlation coefficient, >0.9). The data-driven superresolution reconstruction (VIBESR) is clinically feasible for precontrast and postcontrast upper abdominal VIBE MRI, providing improved image quality, diagnostic confidence, and lesion conspicuity compared with standard VIBESD images.

Mild traumatic brain injury (mTBI) sustained in early childhood affects the brain at a peak developmental period and may disrupt sensitive stages of skill acquisition, thereby compromising child functioning. However, due to the challenges of collecting non-sedated neuroimaging data in young children, the consequences of mTBI on young children's brains have not been systematically studied. In typically developing preschool children (of age 3-5years), a brief behavioral-play familiarization provides an effective alternative to sedation for acquiring awake magnetic resonance imaging (MRI) in a time- and resource-efficient manner. To date, no study has applied such an approach for acquiring non-sedated MRI in preschool children with mTBI who may present with additional MRI acquisition challenges such as agitation or anxiety. The present study aimed to compare the effectiveness of a brief behavioral-play familiarization for acquiring non-sedated MRI for research purposes between young children with and without mTBI, and to identify factors associated with successful MRI acquisition. Preschool children with mTBI (n=13) and typically developing children (n=24) underwent a 15-minutes behavioral-play MRI familiarization followed by a 35-minutes non-sedated MRI protocol. Success rate was compared between groups, MRI quality was assessed quantitatively, and factors predicting success were documented. Among the 37 participants, 15 typically developing children (63%) and 10 mTBI (77%) reached the MRI acquisition success criteria (i.e., completing the two first sequences). The success rate was not significantly different between groups (p=.48; 95% CI [-0.36 14.08]; Cramer's V=.15). The images acquired were of high-quality in 100% (for both groups) of the structural images, and 60% (for both groups) of the diffusion images. Factors associated with success included older child age (Β=0.73, p=.007, exp(B)=3.11, 95% CI [1.36 7.08]) and fewer parental concerns (Β=-1.56, p=.02, exp(Β)=0.21, 95% CI [0.05 0.82]) about the MRI procedure. Using brief behavioral-play familiarization allows acquisition of high-quality non-sedated MRI in young children with mTBI with success rates comparable to those of non-injured peers.

Sparsity-based and deep learning-based image reconstruction algorithms are two promising approaches to accelerate the image acquisition process for localization-based super-resolution microscopy, by allowing a higher density of fluorescing emitters to be imaged in a single frame. Despite the surging popularity, a comprehensive parametric study guiding the practical applications of sparsity-based and deep learning-based image reconstruction algorithms is yet to be conducted. In this study, we examined the performance of sparsity- and deep learning-based algorithms in reconstructing super-resolution images using simulated fluorescent microscopy images. The simulated images were synthesized with varying levels of sparsity and connectivity. We found the deep learning-based VDSR recovers image faster, with a higher recall rate and localization accuracy. The sparsity-based SPIDER recovers more zero pixels truthfully. We also compared the two algorithms using images acquired from a real super-resolution experiment, yielding results agreeing with the results from the evaluation using simulated images. We concluded that VDSR is preferable when accurate emitter localization is needed while SPIDER is more suitable when evaluation of the number of emitters is critical.