Distortion correction of single-shot EPI enabled by deep-learning

Deep learning methods for 3D magnetic resonance image denoising, bias field and motion artifact correction: a comprehensive review

Magnetic resonance imaging (MRI) provides detailed structural information of the internal body organs and soft tissue regions of a patient in clinical diagnosis for disease detection, localization, and progress monitoring. MRI scanner hardware manufacturers incorporate various post-acquisition image-processing techniques into the scanner's computer software tools for different post-processing tasks. These tools provide a final image of adequate quality and essential features for accurate clinical reporting and predictive interpretation for better treatment planning. Different post-acquisition image-processing tasks for MRI quality enhancement include noise removal, motion artifact reduction, magnetic bias field correction, and eddy electric current effect removal. Recently, deep learning (DL) methods have shown great success in many research fields, including image and video applications. DL-based data-driven feature-learning approaches have great potential for MR image denoising and image-quality-degrading artifact correction. Recent studies have demonstrated significant improvements in image-analysis tasks using DL-based convolutional neural network techniques. The promising capabilities and performance of DL techniques in various problem-solving domains have motivated researchers to adapt DL methods to medical image analysis and quality enhancement tasks. This paper presents a comprehensive review of DL-based state-of-the-art MRI quality enhancement and artifact removal methods for regenerating high-quality images while preserving essential anatomical and physiological feature maps without destroying important image information. Existing research gaps and future directions have also been provided by highlighting potential research areas for future developments, along with their importance and advantages in medical imaging.

Artificial intelligence for neuro MRI acquisition: a review

OBJECT

To review recent advances of artificial intelligence (AI) in enhancing the efficiency and throughput of the MRI acquisition workflow in neuroimaging, including planning, sequence design, and correction of acquisition artifacts.

MATERIALS AND METHODS

A comprehensive analysis was conducted on recent AI-based methods in neuro MRI acquisition. The study focused on key technological advances, their impact on clinical practice, and potential risks associated with these methods.

RESULTS

The findings indicate that AI-based algorithms have a substantial positive impact on the MRI acquisition process, improving both efficiency and throughput. Specific algorithms were identified as particularly effective in optimizing acquisition steps, with reported improvements in workflow efficiency.

DISCUSSION

The review highlights the transformative potential of AI in neuro MRI acquisition, emphasizing the technological advances and clinical benefits. However, it also discusses potential risks and challenges, suggesting areas for future research to mitigate these concerns and further enhance AI integration in MRI acquisition.

PyHySCO: GPU-enabled susceptibility artifact distortion correction in seconds

Over the past decade, reversed gradient polarity (RGP) methods have become a popular approach for correcting susceptibility artifacts in echo-planar imaging (EPI). Although several post-processing tools for RGP are available, their implementations do not fully leverage recent hardware, algorithmic, and computational advances, leading to correction times of several minutes per image volume. To enable 3D RGP correction in seconds, we introduce PyTorch Hyperelastic Susceptibility Correction (PyHySCO), a user-friendly EPI distortion correction tool implemented in PyTorch that enables multi-threading and efficient use of graphics processing units (GPUs). PyHySCO uses a time-tested physical distortion model and mathematical formulation and is, therefore, reliable without training. An algorithmic improvement in PyHySCO is its use of the one-dimensional distortion correction method by Chang and Fitzpatrick to initialize the non-linear optimization. PyHySCO is published under the GNU public license and can be used from the command line or its Python interface. Our extensive numerical validation using 3T and 7T data from the Human Connectome Project suggests that PyHySCO can achieve accuracy comparable to that of leading RGP tools at a fraction of the cost. We also validate the new initialization scheme, compare different optimization algorithms, and test the algorithm on different hardware and arithmetic precisions.

Predicting dynamic, motion-related changes in B0 field in the brain at a 7T MRI using a subject-specific fine-trained U-net

PURPOSE

Subject movement during the MR examination is inevitable and causes not only image artifacts but also deteriorates the homogeneity of the main magnetic field (B0 ), which is a prerequisite for high quality data. Thus, characterization of changes to B0 , for example induced by patient movement, is important for MR applications that are prone to B0 inhomogeneities.

METHODS

We propose a deep learning based method to predict such changes within the brain from the change of the head position to facilitate retrospective or even real-time correction. A 3D U-net was trained on in vivo gradient-echo brain 7T MRI data. The input consisted of B0 maps and anatomical images at an initial position, and anatomical images at a different head position (obtained by applying a rigid-body transformation on the initial anatomical image). The output consisted of B0 maps at the new head positions. We further fine-trained the network weights to each subject by measuring a limited number of head positions of the given subject, and trained the U-net with these data.

RESULTS

Our approach was compared to established dynamic B0 field mapping via interleaved navigators, which suffer from limited spatial resolution and the need for undesirable sequence modifications. Qualitative and quantitative comparison showed similar performance between an interleaved navigator-equivalent method and proposed method.

CONCLUSION

It is feasible to predict B0 maps from rigid subject movement and, when combined with external tracking hardware, this information could be used to improve the quality of MR acquisitions without the use of navigators.

Feature-guided deep learning reduces signal loss and increases lesion CNR in diffusion-weighted imaging of the liver

PURPOSE

This research aims to develop a feature-guided deep learning approach and compare it with an optimized conventional post-processing algorithm in order to enhance the image quality of diffusion-weighted liver images and, in particular, to reduce the pulsation-induced signal loss occurring predominantly in the left liver lobe.

METHODS

Data from 40 patients with liver lesions were used. For the conventional approach, the best-suited out of five examined algorithms was chosen. For the deep learning approach, a U-Net was trained. Instead of learning "gold-standard" target images, the network was trained to optimize four image features (lesion CNR, vessel darkness, data consistency, and pulsation artifact reduction), which could be assessed quantitatively using manually drawn ROIs. A quality score was calculated from these four features. As an additional quality assessment, three radiologists rated different features of the resulting images.

RESULTS

The conventional approach could substantially increase the lesion CNR and reduce the pulsation-induced signal loss. However, the vessel darkness was reduced. The deep learning approach increased the lesion CNR and reduced the signal loss to a slightly lower extent, but it could additionally increase the vessel darkness. According to the image quality score, the quality of the deep-learning images was higher than that of the images obtained using the conventional approach. The radiologist ratings were mostly consistent with the quantitative scores, but the overall quality ratings differed among the readers.

CONCLUSION

Unlike the conventional algorithm, the deep-learning algorithm increased the vessel darkness. Therefore, it may be a viable alternative to conventional algorithms.

Thin-slice reverse encoding distortion correction DWI facilitates visualization of non-functioning pituitary neuroendocrine tumor (PitNET)/pituitary adenoma and surrounding normal structures

BACKGROUND

To evaluate the clinical usefulness of thin-slice echo-planar imaging (EPI)-based diffusion-weighted imaging (DWI) with an on-console distortion correction technique, termed reverse encoding distortion correction DWI (RDC-DWI), in patients with non-functioning pituitary neuroendocrine tumor (PitNET)/pituitary adenoma.

METHODS

Patients with non-functioning PitNET/pituitary adenoma who underwent 3-T RDC-DWI between December 2021 and September 2022 were retrospectively enrolled. Image quality was compared among RDC-DWI, DWI with correction for distortion induced by B0 inhomogeneity alone (B0-corrected-DWI), and original EPI-based DWI with anterior-posterior phase-encoding direction (AP-DWI). Susceptibility artifact, anatomical visualization of cranial nerves, overall tumor visualization, and visualization of cavernous sinus invasion were assessed qualitatively. Quantitative assessment of geometric distortion was performed by evaluation of anterior and posterior displacement between each DWI and the corresponding three-dimensional T2-weighted imaging. Signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and apparent diffusion coefficient values were measured.

RESULTS

Sixty-four patients (age 70.8 ± 9.9 years [mean ± standard deviation]; 33 females) with non-functioning PitNET/pituitary adenoma were evaluated. In terms of susceptibility artifacts in the frontal and temporal lobes, visualization of left trigeminal nerve, overall tumor visualization, and anterior displacement, RDC-DWI performed the best and B0-corrected-DWI performed better than AP-DWI. The right oculomotor and right trigeminal nerves were better visualized by RDC-DWI than by B0-corrected-DWI and AP-DWI. Visualization of cavernous sinus invasion and posterior displacement were better by RDC-DWI and B0-corrected-DWI than by AP-DWI. SNR and CNR were the highest for RDC-DWI.

CONCLUSIONS

RDC-DWI achieved excellent image quality regarding susceptibility artifact, geometric distortion, and tumor visualization in patients with non-functioning PitNET/pituitary adenoma.

RELEVANCE STATEMENT

RDC-DWI facilitates excellent visualization of the pituitary region and surrounding normal structures, and its on-console distortion correction technique is convenient. RDC-DWI can clearly depict cavernous sinus invasion of PitNET/pituitary adenoma even without contrast medium.

KEY POINTS

• RDC-DWI is an EPI-based DWI technique with a novel on-console distortion correction technique. • RDC-DWI corrects distortion due to B0 field inhomogeneity and eddy current. • We evaluated the usefulness of thin-slice RDC-DWI in non-functioning PitNET/pituitary adenoma. • RDC-DWI exhibited excellent visualization in the pituitary region and surrounding structures. • In addition, the on-console distortion correction of RDC-DWI is clinically convenient.

FD-Net: An unsupervised deep forward-distortion model for susceptibility artifact correction in EPI

PURPOSE

To introduce an unsupervised deep-learning method for fast and effective correction of susceptibility artifacts in reversed phase-encode (PE) image pairs acquired with echo planar imaging (EPI).

METHODS

Recent learning-based correction approaches in EPI estimate a displacement field, unwarp the reversed-PE image pair with the estimated field, and average the unwarped pair to yield a corrected image. Unsupervised learning in these unwarping-based methods is commonly attained via a similarity constraint between the unwarped images in reversed-PE directions, neglecting consistency to the acquired EPI images. This work introduces a novel unsupervised deep Forward-Distortion Network (FD-Net) that predicts both the susceptibility-induced displacement field and the underlying anatomically correct image. Unlike previous methods, FD-Net enforces the forward-distortions of the correct image in both PE directions to be consistent with the acquired reversed-PE image pair. FD-Net further leverages a multiresolution architecture to maintain high local and global performance.

RESULTS

FD-Net performs competitively with a gold-standard reference method (TOPUP) in image quality, while enabling a leap in computational efficiency. Furthermore, FD-Net outperforms recent unwarping-based methods for unsupervised correction in terms of both image and field quality.

CONCLUSION

The unsupervised FD-Net method introduces a deep forward-distortion approach to enable fast, high-fidelity correction of susceptibility artifacts in EPI by maintaining consistency to measured data. Therefore, it holds great promise for improving the anatomical accuracy of EPI imaging.

Distortion correction using topup algorithm by single k-space (TASK) for echo planar imaging

Distortion of echo planar imaging (EPI) can be corrected using B0 field maps, which can be estimated with the topup algorithm that requires two EPI images with opposite distortions. In this study, we propose a new algorithm, termed topup algorithm by single K-space (TASK), to generate two input images from a single k-space for the topup algorithm to correct EPI distortions. The centric EPI contains the opposite phase-encoding polarities in one k-space, which can be divided into two halves with opposite distortions. Therefore, two inputs could be extracted by dividing the k-space into halves and processing them using the proposed procedure including an iterative procedure of automatic brain masking and uniformity correction. The efficiency of TASK was evaluated using 3D EPI. Quantitative evaluations showed that TASK corrected EPI distortion at a similar level to the traditional methods. The estimated field maps from the conventional topup and TASK showed a high correlation ([Formula: see text]). An ablation study showed the validity of every suggested step. Furthermore, it was confirmed that TASK was effective for distortion correction of two-shot centric EPI as well, demonstrating its wider applicability. In conclusion, TASK can correct EPI distortions by its own single k-space information with no additional scan.

Improved dynamic distortion correction for fMRI using single-echo EPI and a readout-reversed first image (REFILL)

The boundaries between tissues with different magnetic susceptibilities generate inhomogeneities in the main magnetic field which change over time due to motion, respiration and system instabilities. The dynamically changing field can be measured from the phase of the fMRI data and corrected. However, methods for doing so need multi-echo data, time-consuming reference scans and/or involve error-prone processing steps, such as phase unwrapping, which are difficult to implement robustly on the MRI host. The improved dynamic distortion correction method we propose is based on the phase of the single-echo EPI data acquired for fMRI, phase offsets calculated from a triple-echo, bipolar reference scan of circa 3-10 s duration using a method which avoids the need for phase unwrapping and an additional correction derived from one EPI volume in which the readout direction is reversed. This Reverse-Encoded First Image and Low resoLution reference scan (REFILL) approach is shown to accurately measure B0 as it changes due to shim, motion and respiration, even with large dynamic changes to the field at 7 T, where it led to a > 20% increase in time-series signal to noise ratio compared to data corrected with the classic static approach. fMRI results from REFILL-corrected data were free of stimulus-correlated distortion artefacts seen when data were corrected with static field mapping. The method is insensitive to shim changes and eddy current differences between the reference scan and the fMRI time series, and employs calculation steps that are simple and robust, allowing most data processing to be performed in real time on the scanner image reconstruction computer. These improvements make it feasible to routinely perform dynamic distortion correction in fMRI.

Repeatability of neurite orientation dispersion and density imaging in patients with traumatic brain injury

BACKGROUND AND PURPOSE

The aim of this study was to assess the repeatability of neurite orientation dispersion and density imaging in healthy controls (HCs) and traumatic brain injury (TBI).

METHODS

Seventeen HCs and 48 TBI patients were scanned twice over 18 weeks with diffusion imaging. Orientation dispersion (ODI), neurite density (NDI), and the fraction of isotropic diffusion (F-ISO) were quantified in regions of interest (ROIs) from a gray matter, subcortical, and white matter atlas and compared using the coefficient of variation for repeated measures (CVrep ), which quantifies the expected percent change on repeated measurement. We used a modified signed likelihood ratio test (M-SLRT) to compare the CVrep between groups in each ROI while correcting for multiple comparisons.

RESULTS

NDI exhibited excellent repeatability in both groups; the only group difference was found in the fusiform gyrus, where HCs exhibited better repeatability (M-SLRT = 9.463, p = .0021). ODI also had excellent repeatability in both groups, although repeatability was significantly better in HCs in 16 cortical ROIs (p < .0022) and in the bilateral white matter and bilateral cortex (p < .0027). F-ISO exhibited relatively poor repeatability in both groups, with few group differences.

CONCLUSION

Overall, the repeatability of the NDI, ODI, and F-ISO metrics over an 18-week period is acceptable for assessing the effects of behavioral or pharmacological interventions, though caution is advised when assessing F-ISO changes over time.

Regional microstructural differences in ADHD youth with and without a family history of bipolar I disorder

BACKGROUND

Having a first-degree relative with bipolar I disorder (BD) in conjunction with prodromal attention deficit/hyperactivity disorder (ADHD) may represent a unique phenotype that confers greater risk for developing BD than ADHD alone. However, underlying neuropathoetiological mechanisms remain poorly understood. This cross-sectional study compared regional microstructure in psychostimulant-free ADHD youth with ('high-risk', HR) and without ('low-risk', LR) a first-degree relative with BD, and healthy controls (HC).

METHODS

A total of 140 (high-risk, n = 44; low-risk, n = 49; and HC, n = 47) youth (mean age: 14.1 ± 2.5 years, 65 % male) were included in the analysis. Diffusion tensor images were collected and fractional anisotropy (FA) and mean diffusivity (MD) maps were calculated. Both tract-based and voxel-based analyses were performed. Correlations between clinical ratings and microstructural metrics that differed among groups were examined.

RESULTS

No significant group differences in major long-distance fiber tracts were observed. The high-risk ADHD group exhibited predominantly higher FA and lower MD in frontal, limbic, and striatal subregions compared with the low-risk ADHD group. Both low-risk and high-risk ADHD groups exhibited higher FA in unique and overlapping regions compared with HC subjects. Significant correlations between regional microstructural metrics and clinical ratings were observed in ADHD groups.

LIMITATIONS

Prospective longitudinal studies will be required to determine the relevance of these findings to BD risk progression.

CONCLUSIONS

Psychostimulant-free ADHD youth with a BD family history exhibit different microstructure alterations in frontal, limbic, and striatal regions compared with ADHD youth without a BD family history, and may therefore represent a unique phenotype relevant to BD risk progression.

Impact of susceptibility-induced distortion correction on perfusion imaging by pCASL with a segmented 3D GRASE readout

PURPOSE

The consensus for the clinical implementation of arterial spin labeling (ASL) perfusion imaging recommends a segmented 3D Gradient and Spin-Echo (GRASE) readout for optimal signal-to-noise-ratio(SNR). The correction of the associated susceptibility-induced geometric distortions has been shown to improve diagnostic precision, but its impact on ASL data has not been systematically assessed and it is not consistently part of pre-processing pipelines. Here, we investigate the effects of susceptibility-induced distortion correction on perfusion imaging by pseudo-continuous ASL (pCASL) with a segmented 3D GRASE readout.

METHODS

Data acquired from 28 women using pCASL with 3D GRASE at 3T was analyzed using three pre-processing options: without distortion correction, with distortion correction, and with spatial smoothing (without distortion correction) matched to control for blurring effects induced by distortion correction. Maps of temporal SNR (tSNR) and relative perfusion were analyzed in eight regions-of-interest (ROIs) across the brain.

RESULTS

Distortion correction significantly affected tSNR and relative perfusion across the brain. Increases in tSNR were like those produced by matched spatial smoothing in most ROIs, indicating that they were likely due to blurring effects. However, that was not the case in the frontal and temporal lobes, where we also found increased relative perfusion with distortion correction even compared with matched spatial smoothing. These effects were found in both controls and patients, with no interactions with the participant group.

CONCLUSION

Correction of Susceptibility-induced distortions significantly impacts ASL perfusion imaging using a segmented 3D GRASE readout, and this step should therefore be considered in ASL pre-processing pipelines. This is of special importance in clinical studies, reporting perfusion across ROIs defined on relatively undistorted images and when conducting group analyses requiring the alignment of images across different subjects.

Rapid estimation of 2D relative B 1 + -maps from localizers in the human heart at 7T using deep learning

PURPOSE

Subject-tailored parallel transmission pulses for ultra-high fields body applications are typically calculated based on subject-specific B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps of all transmit channels, which require lengthy adjustment times. This study investigates the feasibility of using deep learning to estimate complex, channel-wise, relative 2D B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps from a single gradient echo localizer to overcome long calibration times.

METHODS

126 channel-wise, complex, relative 2D B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps of the human heart from 44 subjects were acquired at 7T using a Cartesian, cardiac gradient-echo sequence obtained under breath-hold to create a library for network training and cross-validation. The deep learning predicted maps were qualitatively compared to the ground truth. Phase-only B 1 + $$ {\mathrm{B}}_1^{+} $$ -shimming was subsequently performed on the estimated B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps for a region of interest covering the heart. The proposed network was applied at 7T to 3 unseen test subjects.

RESULTS

The deep learning-based B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps, derived in approximately 0.2 seconds, match the ground truth for the magnitude and phase. The static, phase-only pulse design performs best when maximizing the mean transmission efficiency. In-vivo application of the proposed network to unseen subjects demonstrates the feasibility of this approach: the network yields predicted B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps comparable to the acquired ground truth and anatomical scans reflect the resulting B 1 + $$ {\mathrm{B}}_1^{+} $$ -pattern using the deep learning-based maps.

CONCLUSION

The feasibility of estimating 2D relative B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps from initial localizer scans of the human heart at 7T using deep learning is successfully demonstrated. Because the technique requires only sub-seconds to derive channel-wise B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps, it offers high potential for advancing clinical body imaging at ultra-high fields.

Technical note: A method to synthesize magnetic resonance images in different patient rotation angles with deep learning for gantry-free radiotherapy

BACKGROUND

Recently, patient rotating devices for gantry-free radiotherapy, a new approach to implement external beam radiotherapy, have been introduced. When a patient is rotated in the horizontal position, gravity causes anatomic deformation. For treatment planning, one feasible method is to acquire simulation images at different horizontal rotation angles.

PURPOSE

This study aimed to investigate the feasibility of synthesizing magnetic resonance (MR) images at patient rotation angles of 180° (prone position) and 90° (lateral position) from those at a rotation angle of 0° (supine position) using deep learning.

METHODS

This study included 23 healthy male volunteers. They underwent MR imaging (MRI) in the supine position and then in the prone (23 volunteers) and lateral (16 volunteers) positions. T1-weighted fast spin echo was performed for all positions with the same parameters. Two two-dimensional deep learning networks, pix2pix generative adversarial network (pix2pix GAN) and CycleGAN, were developed for synthesizing MR images in the prone and lateral positions from those in the supine position, respectively. For the evaluation of the models, leave-one-out cross-validation was performed. The mean absolute error (MAE), Dice similarity coefficient (DSC), and Hausdorff distance (HD) were used to determine the agreement between the prediction and ground truth for the entire body and four specific organs.

RESULTS

For pix2pix GAN, the synthesized images were visually bad, and no quantitative evaluation was performed. The quantitative evaluation metrics of the body outlines calculated for the synthesized prone and lateral images using CycleGAN were as follows: MAE, 35.63 ± 3.98 and 40.45 ± 5.83, respectively; DSC, 0.97 ± 0.01 and 0.94 ± 0.01, respectively; and HD (in pixels), 16.74 ± 3.55 and 31.69 ± 12.03, respectively. The quantitative metrics of the bladder and prostate performed were also promising for both the prone and lateral images, with mean values >0.90 in DSC (p > 0.05). The mean DSC and HD values of the bilateral femur for the prone images were 0.96 and 3.63 (in pixels), respectively, and 0.78 and 12.65 (in pixels) for the lateral images, respectively (p < 0.05).

CONCLUSIONS

The CycleGAN could synthesize the MRI at lateral and prone positions using images at supine position, and it could benefit gantry-free radiation therapy.

Accelerated Diffusion-Weighted MR Image Reconstruction Using Deep Neural Networks

Under-sampling in diffusion-weighted imaging (DWI) decreases the scan time that helps to reduce off-resonance effects, geometric distortions, and susceptibility artifacts; however, it leads to under-sampling artifacts. In this paper, diffusion-weighted MR image (DWI-MR) reconstruction using deep learning (DWI U-Net) is proposed to recover artifact-free DW images from variable density highly under-sampled k-space data. Additionally, different optimizers, i.e., RMSProp, Adam, Adagrad, and Adadelta, have been investigated to choose the best optimizers for DWI U-Net. The reconstruction results are compared with the conventional Compressed Sensing (CS) reconstruction. The quality of the recovered images is assessed using mean artifact power (AP), mean root mean square error (RMSE), mean structural similarity index measure (SSIM), and mean apparent diffusion coefficient (ADC). The proposed method provides up to 61.1%, 60.0%, 30.4%, and 28.7% improvements in the mean AP value of the reconstructed images in our experiments with different optimizers, i.e., RMSProp, Adam, Adagrad, and Adadelta, respectively, as compared to the conventional CS at an acceleration factor of 6 (i.e., AF = 6). The results of DWI U-Net with the RMSProp, Adam, Adagrad, and Adadelta optimizers show 13.6%, 10.0%, 8.7%, and 8.74% improvements, respectively, in terms of mean SSIM with respect to the conventional CS at AF = 6. Also, the proposed technique shows 51.4%, 29.5%, 24.04%, and 18.0% improvements in terms of mean RMSE using the RMSProp, Adam, Adagrad, and Adadelta optimizers, respectively, with reference to the conventional CS at AF = 6. The results confirm that DWI U-Net performs better than the conventional CS reconstruction. Also, when comparing the different optimizers in DWI U-Net, RMSProp provides better results than the other optimizers.

Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging-State-of-the-Art and Challenges

Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast for clinical diagnoses and research which underpin many recent breakthroughs in medicine and biology. The post-processing of reconstructed MR images is often automated for incorporation into MRI scanners by the manufacturers and increasingly plays a critical role in the final image quality for clinical reporting and interpretation. For image enhancement and correction, the post-processing steps include noise reduction, image artefact correction, and image resolution improvements. With the recent success of deep learning in many research fields, there is great potential to apply deep learning for MR image enhancement, and recent publications have demonstrated promising results. Motivated by the rapidly growing literature in this area, in this review paper, we provide a comprehensive overview of deep learning-based methods for post-processing MR images to enhance image quality and correct image artefacts. We aim to provide researchers in MRI or other research fields, including computer vision and image processing, a literature survey of deep learning approaches for MR image enhancement. We discuss the current limitations of the application of artificial intelligence in MRI and highlight possible directions for future developments. In the era of deep learning, we highlight the importance of a critical appraisal of the explanatory information provided and the generalizability of deep learning algorithms in medical imaging.

Simultaneous superresolution reconstruction and distortion correction for single-shot EPI DWI using deep learning

PURPOSE

Single-shot (SS) EPI is widely used for clinical DWI. This study aims to develop an end-to-end deep learning-based method with a novel loss function in an improved network structure to simultaneously increase the resolution and correct distortions for SS-EPI DWI.

THEORY AND METHODS

Point-spread-function (PSF)-encoded EPI can provide high-resolution, distortion-free DWI images. A distorted image from SS-EPI can be described as the convolution between a PSF function with a distortion-free image. The deconvolution process to recover the distortion-free image can be achieved with a convolution neural network, which also learns the mapping function between low-resolution SS-EPI and high-resolution reference PSF-EPI to achieve superresolution. To suppress the oversmoothing effect, we proposed a modified generative adversarial network structure, in which a dense net with gradient map guidance and a multilevel fusion block was used as the generator. A fractional anisotropy loss was proposed to utilize the diffusion anisotropy information among diffusion directions. In vivo brain DWI data were used to test the proposed method.

RESULTS

The results show that distortion-corrected high-resolution DWI images with restored structural details can be obtained from low-resolution SS-EPI images by taking advantage of the high-resolution anatomical images. Additionally, the proposed network can improve the quantitative accuracy of diffusion metrics compared with previously reported networks.

CONCLUSION

Using high-resolution, distortion-free EPI-DWI images as references, a deep learning-based method to simultaneously increase the perceived resolution and correct distortions for low-resolution SS-EPI was proposed. The results show that DWI image quality and diffusion metrics can be improved.

IMPULSED model based cytological feature estimation with U-Net: Application to human brain tumor at 3T

PURPOSE

This work introduces and validates a deep-learning-based fitting method, which can rapidly provide accurate and robust estimation of cytological features of brain tumor based on the IMPULSED (imaging microstructural parameters using limited spectrally edited diffusion) model fitting with diffusion-weighted MRI data.

METHODS

The U-Net was applied to rapidly quantify extracellular diffusion coefficient (D_ex ), cell size (d), and intracellular volume fraction (v_in ) of brain tumor. At the training stage, the image-based training data, synthesized by randomizing quantifiable microstructural parameters within specific ranges, was used to train U-Net. At the test stage, the pre-trained U-Net was applied to estimate the microstructural parameters from simulated data and the in vivo data acquired on patients at 3T. The U-Net was compared with conventional non-linear least-squares (NLLS) fitting in simulations in terms of estimation accuracy and precision.

RESULTS

Our results confirm that the proposed method yields better fidelity in simulations and is more robust to noise than the NLLS fitting. For in vivo data, the U-Net yields obvious quality improvement in parameter maps, and the estimations of all parameters are in good agreement with the NLLS fitting. Moreover, our method is several orders of magnitude faster than the NLLS fitting (from about 5 min to <1 s).

CONCLUSION

The image-based training scheme proposed herein helps to improve the quality of the estimated parameters. Our deep-learning-based fitting method can estimate the cell microstructural parameters fast and accurately.

Mitigating transmit‐B 1 artifacts by predicting parallel transmission images with deep learning: A feasibility study using high‐resolution whole‐brain diffusion at 7 Tesla

Purpose

To propose a novel deep learning (DL) approach to transmit‐B₁ (B₁⁺)‐artifact mitigation without direct use of parallel transmission (pTx), by predicting pTx images from single‐channel transmission (sTx) images.

Methods

A deep encoder–decoder convolutional neural network was constructed and trained to learn the mapping from sTx to pTx images. The feasibility was demonstrated using 7 T Human‐Connectome Project (HCP)‐style diffusion MRI. The training dataset comprised images acquired on 5 healthy subjects using commercial Nova RF coils. Relevant hyperparameters were tuned with a nested cross‐validation, and the generalization performance evaluated using a regular cross‐validation.

Results

Our DL method effectively improved the image quality for sTx images by restoring the signal dropout, with quality measures (including normalized root‐mean‐square error, peak SNR, and structural similarity index measure) improved in most brain regions. The improved image quality was translated into improved performances for diffusion tensor imaging analysis; our method improved accuracy for fractional anisotropy and mean diffusivity estimations, reduced the angular errors of principal eigenvectors, and improved the fiber orientation delineation relative to sTx images. Moreover, the final DL model trained on data of all 5 subjects was successfully used to predict pTx images for unseen new subjects (randomly selected from the 7 T HCP database), effectively recovering the signal dropout and improving color‐coded fractional anisotropy maps with largely reduced noise levels.

Conclusion

The proposed DL method has potential to provide images with reduced B1⁺ artifacts in healthy subjects even when pTx resources are inaccessible on the user side.

What's new and what's next in diffusion MRI preprocessing

Diffusion MRI (dMRI) provides invaluable information for the study of tissue microstructure and brain connectivity, but suffers from a range of imaging artifacts that greatly challenge the analysis of results and their interpretability if not appropriately accounted for. This review will cover dMRI artifacts and preprocessing steps, some of which have not typically been considered in existing pipelines or reviews, or have only gained attention in recent years: brain/skull extraction, B-matrix incompatibilities w.r.t the imaging data, signal drift, Gibbs ringing, noise distribution bias, denoising, between- and within-volumes motion, eddy currents, outliers, susceptibility distortions, EPI Nyquist ghosts, gradient deviations, B1 bias fields, and spatial normalization. The focus will be on "what's new" since the notable advances prior to and brought by the Human Connectome Project (HCP), as presented in the predecessing issue on "Mapping the Connectome" in 2013. In addition to the development of novel strategies for dMRI preprocessing, exciting progress has been made in the availability of open source tools and reproducible pipelines, databases and simulation tools for the evaluation of preprocessing steps, and automated quality control frameworks, amongst others. Finally, this review will consider practical considerations and our view on "what's next" in dMRI preprocessing.

Current Landscape and Future Perspectives of Abbreviated MRI for Hepatocellular Carcinoma Surveillance

While ultrasound (US) is considered an important tool for hepatocellular carcinoma (HCC) surveillance, it has limited sensitivity for detecting early-stage HCC. Abbreviated MRI (AMRI) has recently gained popularity owing to better sensitivity in its detection of early-stage HCC than US, while also minimizing the time and cost in comparison to complete contrast-enhanced MRI, as AMRI includes only a few essential sequences tailored for detecting HCC. Currently, three AMRI protocols exist, namely gadoxetic acid-enhanced hepatobiliary-phase AMRI, dynamic contrast-enhanced AMRI, and non-enhanced AMRI. In this study, we discussed the rationale and technical details of AMRI techniques for achieving optimal surveillance performance. The strengths, weaknesses, and current issues of each AMRI protocol were also elucidated. Moreover, we scrutinized previously performed AMRI studies regarding clinical and technical factors. Reporting and recall strategies were discussed while considering the differences in AMRI protocols. A risk-stratified approach for the target population should be taken to maximize the benefits of AMRI and the cost-effectiveness should be considered. In the era of multiple HCC surveillance tools, patients need to be fully informed about their choices for better adherence to a surveillance program.

Deep learning based multiplexed sensitivity-encoding (DL-MUSE) for high-resolution multi-shot DWI

PURPOSE

A phase correction method for high-resolution multi-shot (MSH) diffusion weighted imaging (DWI) is proposed. The efficacy and generalization capability of the method were validated on both healthy volunteers and patients.

THEORY AND METHODS

Conventionally, inter-shot phase variations for MSH echo-planar imaging (EPI) DWI are corrected by model-based algorithms. However, many acquisition imperfections are hard to measure accurately for conventional model-based methods, making the phase estimation and artifacts suppression unreliable. We propose a deep learning multiplexed sensitivity-encoding (DL-MUSE) framework to improve the phase estimations based on convolutional neural network (CNN) reconstruction. Aliasing-free single-shot (SSH) DW images, which have been used routinely in clinical settings, were used for training before the aliasing correction of MSH-DWI images. A dual-channel U-net comprising multiple convolutional layers was used for the phase estimation of MSH-EPI. The network was trained on a dataset containing 30 healthy volunteers and tested on another dataset of 52 healthy subjects and 15 patients with lesions or tumors with different shot numbers (4, 6 and 8). To further validate the generalization capability of our network, we acquired a dataset with different numbers of shots, TEs, partial Fourier factors, resolutions, ETLs, FOVs, coil numbers, and image orientations from two sites. We also compared the reconstruction performance of our proposed method with that of the conventional MUSE and SSH-EPI qualitatively and quantitatively.

RESULTS

Our results show that DL-MUSE is capable of correcting inter-shot phase errors with high and robust performance. Compared to conventional model-based MUSE, our method, by applying deep learning-based phase corrections, showed reduced distortion, noise level, and signal loss in high b-value DWIs. The improvements of image quality become more evident as the shot number increases from 4 to 8, especially in those central regions of the images, where g-factor artifacts are severe. Furthermore, the proposed method could provide the information about the orientation of the white matter with better consistency and achieve finer fibers delineation compared to the SSH-EPI method. Besides, the experiments on volunteers and patients from two different sites demonstrated the generalizability of our proposed method preliminarily.

CONCLUSION

A deep learning-based reconstruction algorithm for MSH-EPI images, which helps improve image quality greatly, was proposed. Results from healthy volunteers and tumor patients demonstrated the feasibility and generalization performances of our method for high-resolution MSH-EPI DWI, which can be used for routine clinical applications as well as neuroimaging research.

Disruption of the white matter structural network and its correlation with baseline progression rate in patients with sporadic amyotrophic lateral sclerosis

OBJECTIVE

There is increasing evidence that amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease impacting large-scale brain networks. However, it is still unclear which structural networks are associated with the disease and whether the network connectomics are associated with disease progression. This study was aimed to characterize the network abnormalities in ALS and to identify the network-based biomarkers that predict the ALS baseline progression rate.

METHODS

Magnetic resonance imaging was performed on 73 patients with sporadic ALS and 100 healthy participants to acquire diffusion-weighted magnetic resonance images and construct white matter (WM) networks using tractography methods. The global and regional network properties were compared between ALS and healthy subjects. The single-subject WM network matrices of patients were used to predict the ALS baseline progression rate using machine learning algorithms.

RESULTS

Compared with the healthy participants, the patients with ALS showed significantly decreased clustering coefficient Cp (P = 0.0034, t = 2.98), normalized clustering coefficient γ (P = 0.039, t = 2.08), and small-worldness σ (P = 0.038, t = 2.10) at the global network level. The patients also showed decreased regional centralities in motor and non-motor systems including the frontal, temporal and subcortical regions. Using the single-subject structural connection matrix, our classification model could distinguish patients with fast versus slow progression rate with an average accuracy of 85%.

CONCLUSION

Disruption of the WM structural networks in ALS is indicated by weaker small-worldness and disturbances in regions outside of the motor systems, extending the classical pathophysiological understanding of ALS as a motor disorder. The individual WM structural network matrices of ALS patients are potential neuroimaging biomarkers for the baseline disease progression in clinical practice.

Modified Diffusion Tensor Image Processing Pipeline for Archived Studies of Patients With Leukoencephalopathy

BACKGROUND

In archived diffusion tensor imaging (DTI) studies, a reversed-phase encoding (PE) scan required to correct the distortion in single-shot echo-planar imaging (EPI) may not have been acquired. Furthermore, DTI tractography is adversely affected by incorrect white matter segmentation due to leukoencephalopathy (LE). All these issues need to be addressed.

PURPOSE

To propose and evaluate a modified DTI processing pipeline with DIstortion COrrection using pseudo T₂ -weighted images (DICOT) to overcome limitations in existing acquisition protocols.

STUDY TYPE

Retrospective feasibility.

SUBJECTS

DICOT was assessed in simulated data and 84 acute lymphoblastic leukemia (ALL) patients with reversed PE acquired. The pipeline was then tested in 522 scans from 261 ALL patients without a reversed PE acquired.

FIELD STRENGTH/SEQUENCE

A 3 T; diffusion-weighted EPI; 3D magnetization prepared rapid acquisition gradient echo (MPRAGE).

STATISTICAL TESTS

Repeated measures analysis of variance and Tukey post hoc tests were performed to compare fractional anisotropy (FA) values obtained by different methods.

ASSESSMENT

FA and corresponding absolute error maps were obtained using TOPUP, DICOT, INVERSION (Inverse contrast Normalization for VERy Simple registratION) and NO CORR (no correction). Each method was assessed by comparing to TOPUP. The pipeline in the ALL patients was evaluated based on the failure rate of the distortion correction using the global correlation values.

RESULTS

Using DICOT reduced the mean absolute errors by an average of 32% in FA in simulation datasets. In 84 patients, the error reductions were approximately 15% in FA with DICOT, while it was 5% with INVERSION. No significant differences between the TOPUP and DICOT were observed in FA with P = 0.090/0.894(AP/PA). Only 15 of 516 examinations requiring any additional manual intervention.

CONCLUSION

This modified pipeline produced better results than the INVERSION. Furthermore, robust performance was demonstrated in archived patient scans acquired without an inverse PE necessary for TOPUP correction.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 2.