Low-dimensional-structure self-learning and thresholding: regularization beyond compressed sensing for MRI reconstruction

Exploring the limits of scan time reduction for ferumoxytol-enhanced whole-heart angiography in congenital heart disease patients

BACKGROUND

One major challenge in cardiovascular magnetic resonance is reducing scan times to be more compatible with clinical workflows. In 3D magnetic resonance imaging (MRI), strategies to shorten scan times mostly rely on ECG-triggering or self-navigation for motion management, but are affected by heart rate variabilities or respiratory drifts. A similarity-driven multi-dimensional binning algorithm (SIMBA) was introduced for 3D whole-heart angiography from ferumoxytol-enhanced free-running MRI. This study explores acceleration limits using SIMBA, and its compressed-sensing extension extra-dimensional motion-compensation (XD-MC)-SIMBA, while preserving image quality.

METHODS

Data from 6-min free-running acquisitions of 30 congenital heart disease (CHD) patients were retrospectively undersampled to simulate 5-, 4-, 3-, 2-, and 1-min datasets. SIMBA and XD-MC-SIMBA reconstructions were applied. and the consistency of the data selection together with sharpness metrics were computed as a function of undersampling. Image quality was rated on a 5-point Likert scale. Shorter 3-minute acquisitions were prospectively acquired in nine CHD patients.

RESULTS

SIMBA's motion state selection was consistent across undersampling levels, with only 2 of 30 cases showing completely different selections. Image quality metrics decreased with increased undersampling, with SIMBA scoring lower compared to XD-MC-SIMBA. The diagnostic quality was good, with lower scores for 2- and 1-min datasets. Using XD-MC-SIMBA, 43% (31/72) of cases showed improved scores compared to SIMBA and 58% (7/12) of 1-min datasets improved to good or excellent quality.

CONCLUSIONS

This study demonstrates that ferumoxytol-enhanced free-running MRI can be highly accelerated for 3D angiography in CHD.With the aid of compressed sensing, XD-MC-SIMBA supports the acceleration down to 3 minutes or less.

A motion-corrected deep-learning reconstruction framework for accelerating whole-heart magnetic resonance imaging in patients with congenital heart disease

BACKGROUND

Cardiovascular magnetic resonance (CMR) is an important imaging modality for the assessment and management of adult patients with congenital heart disease (CHD). However, conventional techniques for three-dimensional (3D) whole-heart acquisition involve long and unpredictable scan times and methods that accelerate scans via k-space undersampling often rely on long iterative reconstructions. Deep-learning-based reconstruction methods have recently attracted much interest due to their capacity to provide fast reconstructions while often outperforming existing state-of-the-art methods. In this study, we sought to adapt and validate a non-rigid motion-corrected model-based deep learning (MoCo-MoDL) reconstruction framework for 3D whole-heart MRI in a CHD patient cohort.

METHODS

The previously proposed deep-learning reconstruction framework MoCo-MoDL, which incorporates a non-rigid motion-estimation network and a denoising regularization network within an unrolled iterative reconstruction, was trained in an end-to-end manner using 39 CHD patient datasets. Once trained, the framework was evaluated in eight CHD patient datasets acquired with seven-fold prospective undersampling. Reconstruction quality was compared with the state-of-the-art non-rigid motion-corrected patch-based low-rank reconstruction method (NR-PROST) and against reference images (acquired with three-or-four-fold undersampling and reconstructed with NR-PROST).

RESULTS

Seven-fold undersampled scan times were 2.1 ± 0.3 minutes and reconstruction times were ∼30 seconds, approximately 240 times faster than an NR-PROST reconstruction. Image quality comparable to the reference images was achieved using the proposed MoCo-MoDL framework, with no statistically significant differences found in any of the assessed quantitative or qualitative image quality measures. Additionally, expert image quality scores indicated the MoCo-MoDL reconstructions were consistently of a higher quality than the NR-PROST reconstructions of the same data, with the differences in 12 of the 22 scores measured for individual vascular structures found to be statistically significant.

CONCLUSION

The MoCo-MoDL framework was applied to an adult CHD patient cohort, achieving good quality 3D whole-heart images from ∼2-minute scans with reconstruction times of ∼30 seconds.

Optimal MRI undersampling patterns for ultimate benefit of medical vision tasks

Compressed sensing is commonly concerned with optimizing the image quality after a partial undersampling of the measurable k-space to accelerate MRI. In this article, we propose to change the focus from the quality of the reconstructed image to the quality of the downstream image analysis outcome. Specifically, we propose to optimize the patterns according to how well a sought-after pathology could be detected or localized in the reconstructed images. We find the optimal undersampling patterns in k-space that maximize target value functions of interest in commonplace medical vision problems (reconstruction, segmentation, and classification) and propose a new iterative gradient sampling routine universally suitable for these tasks. We validate the proposed MRI acceleration paradigm on three classical medical datasets, demonstrating a noticeable improvement of the target metrics at the high acceleration factors (for the segmentation problem at ×16 acceleration, we report up to 12% improvement in Dice score over the other undersampling patterns).

Adapting model-based deep learning to multiple acquisition conditions: Ada-MoDL

PURPOSE

The aim of this work is to introduce a single model-based deep network that can provide high-quality reconstructions from undersampled parallel MRI data acquired with multiple sequences, acquisition settings, and field strengths.

METHODS

A single unrolled architecture, which offers good reconstructions for multiple acquisition settings, is introduced. The proposed scheme adapts the model to each setting by scaling the convolutional neural network (CNN) features and the regularization parameter with appropriate weights. The scaling weights and regularization parameter are derived using a multilayer perceptron model from conditional vectors, which represents the specific acquisition setting. The perceptron parameters and the CNN weights are jointly trained using data from multiple acquisition settings, including differences in field strengths, acceleration, and contrasts. The conditional network is validated using datasets acquired with different acquisition settings.

RESULTS

The comparison of the adaptive framework, which trains a single model using the data from all the settings, shows that it can offer consistently improved performance for each acquisition condition. The comparison of the proposed scheme with networks that are trained independently for each acquisition setting shows that it requires less training data per acquisition setting to offer good performance.

CONCLUSION

The Ada-MoDL framework enables the use of a single model-based unrolled network for multiple acquisition settings. In addition to eliminating the need to train and store multiple networks for different acquisition settings, this approach reduces the training data needed for each acquisition setting.

Automated detection of cardiac rest period for trigger delay calculation for image-based navigator coronary magnetic resonance angiography

BACKGROUND

Coronary magnetic resonance angiography (coronary MRA) is increasingly being considered as a clinically viable method to investigate coronary artery disease (CAD). Accurate determination of the trigger delay to place the acquisition window within the quiescent part of the cardiac cycle is critical for coronary MRA in order to reduce cardiac motion. This is currently reliant on operator-led decision making, which can negatively affect consistency of scan acquisition. Recently developed deep learning (DL) derived software may overcome these issues by automation of cardiac rest period detection.

METHODS

Thirty individuals (female, n = 10) were investigated using a 0.9 mm isotropic image-navigator (iNAV)-based motion-corrected coronary MRA sequence. Each individual was scanned three times utilising different strategies for determination of the optimal trigger delay: (1) the DL software, (2) an experienced operator decision, and (3) a previously utilised formula for determining the trigger delay. Methodologies were compared using custom-made analysis software to assess visible coronary vessel length and coronary vessel sharpness for the entire vessel length and the first 4 cm of each vessel.

RESULTS

There was no difference in image quality between any of the methodologies for determination of the optimal trigger delay, as assessed by visible coronary vessel length, coronary vessel sharpness for each entire vessel and vessel sharpness for the first 4 cm of the left mainstem, left anterior descending or right coronary arteries. However, vessel length of the left circumflex was slightly greater using the formula method. The time taken to calculate the trigger delay was significantly lower for the DL-method as compared to the operator-led approach (106 ± 38.0 s vs 168 ± 39.2 s, p < 0.01, 95% CI of difference 25.5-98.1 s).

CONCLUSIONS

Deep learning-derived automated software can effectively and efficiently determine the optimal trigger delay for acquisition of coronary MRA and thus may simplify workflow and improve reproducibility.

Latest Advances in Image Acceleration: All Dimensions are Fair Game

Magnetic resonance imaging (MRI) is a versatile modality that can generate high-resolution images with a variety of tissue contrasts. However, MRI is a slow technique and requires long acquisition times, which increase with higher temporal and spatial resolution and/or when multiple contrasts and large volumetric coverage is required. In order to speedup MR data acquisition, several approaches have been introduced in the literature. Most of these techniques acquire less data than required and exploit intrinsic redundancies in the MR images to recover the information that was not sampled. This article presents a review of MR acquisition and reconstruction methods that have exploited redundancies in the temporal, spatial, and contrast/parametric dimensions to accelerate image data acquisition, focusing on cardiac and abdominal MR imaging applications. The review describes how each of these dimensions has been separately exploited for speeding up MR acquisition to then discuss more advanced techniques where multiple dimensions are exploited together for further reducing scan times. Finally, future directions for multidimensional image acceleration and remaining technical challenges are discussed. EVIDENCE LEVEL: 5 TECHNICAL EFFICACY: 1.

Multi-mask self-supervised learning for physics-guided neural networks in highly accelerated magnetic resonance imaging

Self-supervised learning has shown great promise because of its ability to train deep learning (DL) magnetic resonance imaging (MRI) reconstruction methods without fully sampled data. Current self-supervised learning methods for physics-guided reconstruction networks split acquired undersampled data into two disjoint sets, where one is used for data consistency (DC) in the unrolled network, while the other is used to define the training loss. In this study, we propose an improved self-supervised learning strategy that more efficiently uses the acquired data to train a physics-guided reconstruction network without a database of fully sampled data. The proposed multi-mask self-supervised learning via data undersampling (SSDU) applies a holdout masking operation on the acquired measurements to split them into multiple pairs of disjoint sets for each training sample, while using one of these pairs for DC units and the other for defining loss, thereby more efficiently using the undersampled data. Multi-mask SSDU is applied on fully sampled 3D knee and prospectively undersampled 3D brain MRI datasets, for various acceleration rates and patterns, and compared with the parallel imaging method, CG-SENSE, and single-mask SSDU DL-MRI, as well as supervised DL-MRI when fully sampled data are available. The results on knee MRI show that the proposed multi-mask SSDU outperforms SSDU and performs as well as supervised DL-MRI. A clinical reader study further ranks the multi-mask SSDU higher than supervised DL-MRI in terms of signal-to-noise ratio and aliasing artifacts. Results on brain MRI show that multi-mask SSDU achieves better reconstruction quality compared with SSDU. The reader study demonstrates that multi-mask SSDU at R = 8 significantly improves reconstruction compared with single-mask SSDU at R = 8, as well as CG-SENSE at R = 2.

Simultaneous multi-parametric acquisition and reconstruction techniques in cardiac magnetic resonance imaging: Basic concepts and status of clinical development

Simultaneous multi-parametric acquisition and reconstruction techniques (SMART) are gaining attention for their potential to overcome some of cardiovascular magnetic resonance imaging's (CMR) clinical limitations. The major advantages of SMART lie within their ability to simultaneously capture multiple "features" such as cardiac motion, respiratory motion, T1/T2 relaxation. This review aims to summarize the overarching theory of SMART, describing key concepts that many of these techniques share to produce co-registered, high quality CMR images in less time and with less requirements for specialized personnel. Further, this review provides an overview of the recent developments in the field of SMART by describing how they work, the parameters they can acquire, their status of clinical testing and validation, and by providing examples for how their use can improve the current state of clinical CMR workflows. Many of the SMART are in early phases of development and testing, thus larger scale, controlled trials are needed to evaluate their use in clinical setting and with different cardiac pathologies.

Improved myocardial scar visualization with fast free-breathing motion-compensated black-blood T1-rho-prepared late gadolinium enhancement MRI

PURPOSE

Clinical guidelines recommend the use of bright-blood late gadolinium enhancement (BR-LGE) for the detection and quantification of regional myocardial fibrosis and scar. This technique, however, may suffer from poor contrast at the blood-scar interface, particularly in patients with subendocardial myocardial infarction. The purpose of this study was to assess the clinical performance of a two-dimensional black-blood LGE (BL-LGE) sequence, which combines free-breathing T₁-rho-prepared single-shot acquisitions with an advanced non-rigid motion-compensated patch-based reconstruction.

MATERIALS AND METHODS

Extended phase graph simulations and phantom experiments were performed to investigate the performance of the motion-correction algorithm and to assess the black-blood properties of the proposed sequence. Fifty-one patients (37 men, 14 women; mean age, 55 ± 15 [SD] years; age range: 19-81 years) with known or suspected cardiac disease prospectively underwent free-breathing T₁-rho-prepared BL-LGE imaging with inline non-rigid motion-compensated patch-based reconstruction at 1.5T. Conventional breath-held BR-LGE images were acquired for comparison purposes. Acquisition times were recorded. Two readers graded the image quality and relative contrasts were calculated. Presence, location, and extent of LGE were evaluated.

RESULTS

BL-LGE images were acquired with full ventricular coverage in 115 ± 25 (SD) sec (range: 64-160 sec). Image quality was significantly higher on free-breathing BL-LGE imaging than on its breath-held BR-LGE counterpart (3.6 ± 0.7 [SD] [range: 2-4] vs. 3.9 ± 0.2 [SD] [range: 3-4]) (P <0.01) and was graded as diagnostic for 44/51 (86%) patients. The mean scar-to-myocardium and scar-to-blood relative contrasts were significantly higher on BL-LGE images (P < 0.01 for both). The extent of LGE was larger on BL-LGE (median, 5 segments [IQR: 2, 7 segments] vs. median, 4 segments [IQR: 1, 6 segments]) (P < 0.01), the method being particularly sensitive in segments with LGE involving the subendocardium or papillary muscles. In eight patients (16%), BL-LGE could ascertain or rule out a diagnosis otherwise inconclusive on BR-LGE.

CONCLUSION

Free-breathing T₁-rho-prepared BL-LGE imaging with inline motion compensated reconstruction offers a promising diagnostic technology for the non-invasive assessment of myocardial injuries.

Deep-learning based super-resolution for 3D isotropic coronary MR angiography in less than a minute

PURPOSE

To develop and evaluate a novel and generalizable super-resolution (SR) deep-learning framework for motion-compensated isotropic 3D coronary MR angiography (CMRA), which allows free-breathing acquisitions in less than a minute.

METHODS

Undersampled motion-corrected reconstructions have enabled free-breathing isotropic 3D CMRA in ~5-10 min acquisition times. In this work, we propose a deep-learning-based SR framework, combined with non-rigid respiratory motion compensation, to shorten the acquisition time to less than 1 min. A generative adversarial network (GAN) is proposed consisting of two cascaded Enhanced Deep Residual Network generator, a trainable discriminator, and a perceptual loss network. A 16-fold increase in spatial resolution is achieved by reconstructing a high-resolution (HR) isotropic CMRA (0.9 mm3 or 1.2 mm3 ) from a low-resolution (LR) anisotropic CMRA (0.9 × 3.6 × 3.6 mm3 or 1.2 × 4.8 × 4.8 mm3 ). The impact and generalization of the proposed SRGAN approach to different input resolutions and operation on image and patch-level is investigated. SRGAN was evaluated on a retrospective downsampled cohort of 50 patients and on 16 prospective patients that were scanned with LR-CMRA in ~50 s under free-breathing. Vessel sharpness and length of the coronary arteries from the SR-CMRA is compared against the HR-CMRA.

RESULTS

SR-CMRA showed statistically significant (P < .001) improved vessel sharpness 34.1% ± 12.3% and length 41.5% ± 8.1% compared with LR-CMRA. Good generalization to input resolution and image/patch-level processing was found. SR-CMRA enabled recovery of coronary stenosis similar to HR-CMRA with comparable qualitative performance.

CONCLUSION

The proposed SR-CMRA provides a 16-fold increase in spatial resolution with comparable image quality to HR-CMRA while reducing the predictable scan time to <1 min.

Coronary Magnetic Resonance Angiography in Chronic Coronary Syndromes

Cardiovascular disease is the leading cause of mortality worldwide, with atherosclerotic coronary artery disease (CAD) accounting for the majority of cases. X-ray coronary angiography and computed tomography coronary angiography (CCTA) are the imaging modalities of choice for the assessment of CAD. However, the use of ionising radiation and iodinated contrast agents remain drawbacks. There is therefore a clinical need for an alternative modality for the early identification and longitudinal monitoring of CAD without these associated drawbacks. Coronary magnetic resonance angiography (CMRA) could be a potential alternative for the detection and monitoring of coronary arterial stenosis, without exposing patients to ionising radiation or iodinated contrast agents. Further advantages include its versatility, excellent soft tissue characterisation and suitability for repeat imaging. Despite the early promise of CMRA, widespread clinical utilisation remains limited due to long and unpredictable scan times, onerous scan planning, lower spatial resolution, as well as motion related image quality degradation. The past decade has brought about a resurgence in CMRA technology, with significant leaps in image acceleration, respiratory and cardiac motion estimation and advanced motion corrected or motion-resolved image reconstruction. With the advent of artificial intelligence, great advances are also seen in deep learning-based motion estimation, undersampled and super-resolution reconstruction promising further improvements of CMRA. This has enabled high spatial resolution (1 mm isotropic), 3D whole heart CMRA in a clinically feasible and reliable acquisition time of under 10 min. Furthermore, latest super-resolution image reconstruction approaches which are currently under evaluation promise acquisitions as short as 1 min. In this review, we will explore the recent technological advances that are designed to bring CMRA closer to clinical reality.

Motion compensation combining with local low rank regularization for low dose dynamic CT myocardial perfusion reconstruction

Dynamic CT myocardial perfusion imaging (DCT-MPI) is a reliable examination tool for the assessment of myocardium and vascular, while its special scan protocol may result in excessive radiation exposure to patients and inevitable inter-frame motion. Lowering the tube current is a simple way to reduce radiation exposure. However, low mAs will certainly cause severe image noise, thus may further impact the accuracy of functional hemodynamic parameters, which are used for the assessment of blood supply. In this work, we present a novel scheme applying motion compensation and local low rank regularization (MC-LLR) for obtaining high quality motion compensated DCT-MPI images. Specifically, motion compensation by using robust data decomposition registration (RDDR) was introduced. Robust principal component analysis coupled with optical flow-based registration algorithm were used in RDDR. Then, the local low rank constraint on the motion compensated time series images was applied for the DCT-MPI reconstruction. One healthy mini pig and two patient datasets were used to evaluate the proposed MC-LLR algorithm. Results show that the present method achieved satisfactory image quality with higher CNRs, smaller rRMSEs, and more accurate hemodynamic parameter maps.

Accelerated free-breathing 3D whole-heart magnetic resonance angiography with a radial phyllotaxis trajectory, compressed sensing, and curvelet transform

PURPOSE

To develop and validate an accelerated free-breathing 3D whole-heart magnetic resonance angiography (MRA) technique using a radial k-space trajectory with compressed sensing and curvelet transform.

METHOD

A 3D radial phyllotaxis trajectory was implemented to traverse the centerline of k-space immediately before the segmented whole-heart MRA data acquisition at each cardiac cycle. The k-space centerlines were used to correct the respiratory-induced heart motion in the acquired MRA data. The corrected MRA data were then reconstructed by a novel compressed sensing algorithm using curvelets as the sparsifying domain. The proposed 3D whole-heart MRA technique (radial CS curvelet) was then prospectively validated against compressed sensing with a conventional wavelet transform (radial CS wavelet) and a standard Cartesian acquisition in terms of scan time and border sharpness.

RESULTS

Fifteen patients (females 10, median age 34-year-old) underwent 3D whole-heart MRA imaging using a standard Cartesian trajectory and our proposed radial phyllotaxis trajectory. Scan time for radial phyllotaxis was significantly shorter than Cartesian (4.88 ± 0.86 min. vs. 6.84 ± 1.79 min., P-value = 0.004). Radial CS curvelet border sharpness was slightly lower than Cartesian and, for the majority of vessels, was significantly better than radial CS wavelet (P-value < 0.050).

CONCLUSION

The proposed technique of 3D whole-heart MRA acquisition with a radial CS curvelet has a shorter scan time and slightly lower vessel sharpness compared to the Cartesian acquisition with radial profile ordering, and has slightly better sharpness than radial CS wavelet. Future work on this technique includes additional clinical trials and extending this technique to 3D cine imaging.

Improving subspace constrained radial fast spin echo MRI using block matching driven non-local low rank regularization

Subspace-constrained reconstruction methods restrict the relaxation signals (of size M) in the scene to a pre-determined subspace (of size K≪M) and allow multi-contrast imaging and parameter mapping from accelerated acquisitions. However, these constraints yield poor image quality at some imaging contrasts, which can impact the parameter mapping performance. Additional regularization such as the use of joint-sparse (JS) or locally-low-rank (LLR) constraints can help improve the recovery of these images but are not sufficient when operating at high acceleration rates. We propose a method, non-local rank 3D (NLR3D), that is built on block matching and transform domain low rank constraints to allow high quality recovery of subspace-coefficient images (SCI) and subsequent multi-contrast imaging and parameter mapping. The performance of NLR3D was evaluated using Monte-Carlo (MC) simulations and compared against the JS and LLR methods. In vivo T 2 mapping results are presented on brain and knee datasets. MC results demonstrate improved bias, variance, and MSE behavior in both the multi-contrast images and parameter maps when compared to the JS and LLR methods. In vivo brain and knee results at moderate and high acceleration rates demonstrate improved recovery of high SNR early TE images as well as parameter maps. No significant difference was found in the T2 values measured in ROIs between the NLR3D reconstructions and the reference images (Wilcoxon signed rank test). The proposed method, NLR3D, enables recovery of high-quality SCI and, consequently, the associated multi-contrast images and parameter maps.

PIC-GAN: A Parallel Imaging Coupled Generative Adversarial Network for Accelerated Multi-Channel MRI Reconstruction

In this study, we proposed a model combing parallel imaging (PI) with generative adversarial network (GAN) architecture (PIC-GAN) for accelerated multi-channel magnetic resonance imaging (MRI) reconstruction. This model integrated data fidelity and regularization terms into the generator to benefit from multi-coils information and provide an "end-to-end" reconstruction. Besides, to better preserve image details during reconstruction, we combined the adversarial loss with pixel-wise loss in both image and frequency domains. The proposed PIC-GAN framework was evaluated on abdominal and knee MRI images using 2, 4 and 6-fold accelerations with different undersampling patterns. The performance of the PIC-GAN was compared to the sparsity-based parallel imaging (L1-ESPIRiT), the variational network (VN), and conventional GAN with single-channel images as input (zero-filled (ZF)-GAN). Experimental results show that our PIC-GAN can effectively reconstruct multi-channel MR images at a low noise level and improved structure similarity of the reconstructed images. PIC-GAN has yielded the lowest Normalized Mean Square Error (in ×10-5) (PIC-GAN: 0.58 ± 0.37, ZF-GAN: 1.93 ± 1.41, VN: 1.87 ± 1.28, L1-ESPIRiT: 2.49 ± 1.04 for abdominal MRI data and PIC-GAN: 0.80 ± 0.26, ZF-GAN: 0.93 ± 0.29, VN:1.18 ± 0.31, L1-ESPIRiT: 1.28 ± 0.24 for knee MRI data) and the highest Peak Signal to Noise Ratio (PIC-GAN: 34.43 ± 1.92, ZF-GAN: 31.45 ± 4.0, VN: 29.26 ± 2.98, L1-ESPIRiT: 25.40 ± 1.88 for abdominal MRI data and PIC-GAN: 34.10 ± 1.09, ZF-GAN: 31.47 ± 1.05, VN: 30.01 ± 1.01, L1-ESPIRiT: 28.01 ± 0.98 for knee MRI data) compared to ZF-GAN, VN and L1-ESPIRiT with an under-sampling factor of 6. The proposed PIC-GAN framework has shown superior reconstruction performance in terms of reducing aliasing artifacts and restoring tissue structures as compared to other conventional and state-of-the-art reconstruction methods.

Coronary Magnetic Resonance Angiography: Technical Innovations Leading Us to the Promised Land?

Coronary artery disease remains the leading cause of cardiovascular morbidity and mortality. Invasive X-ray angiography and coronary computed tomography angiography are established gold standards for coronary luminography. However, they expose patients to invasive complications, ionizing radiation, and iodinated contrast agents. Among a number of imaging modalities, coronary cardiovascular magnetic resonance (CMR) angiography may be used in some cases as an alternative for the detection and monitoring of coronary arterial stenosis, with advantages including its versatility, excellent soft tissue characterization, and avoidance of ionizing radiation and iodinated contrast agents. In this review, we explore the recent advances in motion correction, image acceleration, and reconstruction technologies that are bringing coronary CMR angiography closer to widespread clinical implementation.

Self-supervised learning of physics-guided reconstruction neural networks without fully sampled reference data

PURPOSE

To develop a strategy for training a physics-guided MRI reconstruction neural network without a database of fully sampled data sets.

METHODS

Self-supervised learning via data undersampling (SSDU) for physics-guided deep learning reconstruction partitions available measurements into two disjoint sets, one of which is used in the data consistency (DC) units in the unrolled network and the other is used to define the loss for training. The proposed training without fully sampled data is compared with fully supervised training with ground-truth data, as well as conventional compressed-sensing and parallel imaging methods using the publicly available fastMRI knee database. The same physics-guided neural network is used for both proposed SSDU and supervised training. The SSDU training is also applied to prospectively two-fold accelerated high-resolution brain data sets at different acceleration rates, and compared with parallel imaging.

RESULTS

Results on five different knee sequences at an acceleration rate of 4 shows that the proposed self-supervised approach performs closely with supervised learning, while significantly outperforming conventional compressed-sensing and parallel imaging, as characterized by quantitative metrics and a clinical reader study. The results on prospectively subsampled brain data sets, in which supervised learning cannot be used due to lack of ground-truth reference, show that the proposed self-supervised approach successfully performs reconstruction at high acceleration rates (4, 6, and 8). Image readings indicate improved visual reconstruction quality with the proposed approach compared with parallel imaging at acquisition acceleration.

CONCLUSION

The proposed SSDU approach allows training of physics-guided deep learning MRI reconstruction without fully sampled data, while achieving comparable results with supervised deep learning MRI trained on fully sampled data.

Non-contrast coronary magnetic resonance angiography: current frontiers and future horizons

Coronary magnetic resonance angiography (coronary MRA) is advantageous in its ability to assess coronary artery morphology and function without ionizing radiation or contrast media. However, technical limitations including reduced spatial resolution, long acquisition times, and low signal-to-noise ratios prevent it from clinical routine utilization. Nonetheless, each of these limitations can be specifically addressed by a combination of novel technologies including super-resolution imaging, compressed sensing, and deep-learning reconstruction. In this paper, we first review the current clinical use and motivations for non-contrast coronary MRA, discuss currently available coronary MRA techniques, and highlight current technical developments that hold unique potential to optimize coronary MRA image acquisition and post-processing. In the final section, we examine the various research-based coronary MRA methods and metrics that can be leveraged to assess coronary stenosis severity, physiological function, and atherosclerotic plaque characterization. We specifically discuss how such technologies may contribute to the clinical translation of coronary MRA into a robust modality for routine clinical use.

Dense Recurrent Neural Networks for Accelerated MRI: History-Cognizant Unrolling of Optimization Algorithms

Inverse problems for accelerated MRI typically incorporate domain-specific knowledge about the forward encoding operator in a regularized reconstruction framework. Recently physics-driven deep learning (DL) methods have been proposed to use neural networks for data-driven regularization. These methods unroll iterative optimization algorithms to solve the inverse problem objective function, by alternating between domain-specific data consistency and data-driven regularization via neural networks. The whole unrolled network is then trained end-to-end to learn the parameters of the network. Due to simplicity of data consistency updates with gradient descent steps, proximal gradient descent (PGD) is a common approach to unroll physics-driven DL reconstruction methods. However, PGD methods have slow convergence rates, necessitating a higher number of unrolled iterations, leading to memory issues in training and slower reconstruction times in testing. Inspired by efficient variants of PGD methods that use a history of the previous iterates, we propose a history-cognizant unrolling of the optimization algorithm with dense connections across iterations for improved performance. In our approach, the gradient descent steps are calculated at a trainable combination of the outputs of all the previous regularization units. We also apply this idea to unrolling variable splitting methods with quadratic relaxation. Our results in reconstruction of the fastMRI knee dataset show that the proposed history-cognizant approach reduces residual aliasing artifacts compared to its conventional unrolled counterpart without requiring extra computational power or increasing reconstruction time.

Towards accelerated quantitative sodium MRI at 7 T in the skeletal muscle: Comparison of anisotropic acquisition- and compressed sensing techniques

PURPOSE

To compare three anisotropic acquisition schemes and three compressed sensing (CS) approaches for accelerated tissue sodium concentration (TSC) quantification using ²³Na MRI at 7 T.

MATERIALS AND METHODS

Three anisotropic 3D-radial acquisition sequences were evaluated using simulations, phantom- and in vivo TSC measurements: An anisotropic density-adapted 3D-radial sequence (3DPR-C), a 3D acquisition-weighted density-adapted stack-of-stars sampling scheme (SOS) and a SOS approach with golden-ratio rotation (SOS-GR). Eight healthy volunteers were examined at a 7 Tesla MRI system. TSC measurements of the calf were conducted with a nominal spatial resolution of Δx = (3.0 × 3.0 × 15.0) mm³ and a field of view of (156.0 × 156.0 × 240.0) mm³ for multiple undersampling factors (USF). Three CS reconstructions were evaluated: Total variation CS (TV-CS), 3D dictionary-learning compressed sensing (3D-DLCS) and TV-CS with a block matching prior (TV-BL-CS). Results of the simulations and measurements were compared to a simulated ground truth (GT) or a fully sampled reference measurement (FS), respectively. The deviation of the mean TSC evaluated in multiple ROI (mE_GT/FS) and the normalized root-mean-squared error (NRMSE) for simulations were evaluated for CS and NUFFT reconstructions.

RESULTS

In simulations, the SOS-GR yielded the lowest NRMSE and mE_GT (< 4%) with NUFFT for an acquisition time (TA) of less than 2 min. CS further improved the results. In simulations and measurements, the best TSC quantification results were obtained with 3D-DLCS and SOS-GR (lowest NRMSE, mE_GT < 2.6% in simulations, mE_GT < 10.7% for phantom measurements and mE_FS < 6% in vivo) with an USF = 4.1 (TA < 2 min). TV-CS showed no or only slight improvements to NUFFT. The results of TV-BL-CS were similar to 3D-DLCS.

DISCUSSION

The TA for TSC measurements could be reduced to less than 2 min by using adapted sequences such as SOS-GR and CS reconstruction approaches such as 3D-DLCS or TV-BL-CS, while the quantitative accuracy stays comparable to a fully sampled NUFFT reconstruction (approx. 8 min TA). In future, the lower TA could improve clinical applicability of TSC measurements.

Rapid whole-heart CMR with single volume super-resolution

BACKGROUND

Three-dimensional, whole heart, balanced steady state free precession (WH-bSSFP) sequences provide delineation of intra-cardiac and vascular anatomy. However, they have long acquisition times. Here, we propose significant speed-ups using a deep-learning single volume super-resolution reconstruction, to recover high-resolution features from rapidly acquired low-resolution WH-bSSFP images.

METHODS

A 3D residual U-Net was trained using synthetic data, created from a library of 500 high-resolution WH-bSSFP images by simulating 50% slice resolution and 50% phase resolution. The trained network was validated with 25 synthetic test data sets. Additionally, prospective low-resolution data and high-resolution data were acquired in 40 patients. In the prospective data, vessel diameters, quantitative and qualitative image quality, and diagnostic scoring was compared between the low-resolution, super-resolution and reference high-resolution WH-bSSFP data.

RESULTS

The synthetic test data showed a significant increase in image quality of the low-resolution images after super-resolution reconstruction. Prospectively acquired low-resolution data was acquired ~× 3 faster than the prospective high-resolution data (173 s vs 488 s). Super-resolution reconstruction of the low-resolution data took < 1 s per volume. Qualitative image scores showed super-resolved images had better edge sharpness, fewer residual artefacts and less image distortion than low-resolution images, with similar scores to high-resolution data. Quantitative image scores showed super-resolved images had significantly better edge sharpness than low-resolution or high-resolution images, with significantly better signal-to-noise ratio than high-resolution data. Vessel diameters measurements showed over-estimation in the low-resolution measurements, compared to the high-resolution data. No significant differences and no bias was found in the super-resolution measurements in any of the great vessels. However, a small but significant for the underestimation was found in the proximal left coronary artery diameter measurement from super-resolution data. Diagnostic scoring showed that although super-resolution did not improve accuracy of diagnosis, it did improve diagnostic confidence compared to low-resolution imaging.

CONCLUSION

This paper demonstrates the potential of using a residual U-Net for super-resolution reconstruction of rapidly acquired low-resolution whole heart bSSFP data within a clinical setting. We were able to train the network using synthetic training data from retrospective high-resolution whole heart data. The resulting network can be applied very quickly, making these techniques particularly appealing within busy clinical workflow. Thus, we believe that this technique may help speed up whole heart CMR in clinical practice.

Deep complex convolutional network for fast reconstruction of 3D late gadolinium enhancement cardiac MRI

Several deep-learning models have been proposed to shorten MRI scan time. Prior deep-learning models that utilize real-valued kernels have limited capability to learn rich representations of complex MRI data. In this work, we utilize a complex-valued convolutional network (ℂNet) for fast reconstruction of highly under-sampled MRI data and evaluate its ability to rapidly reconstruct 3D late gadolinium enhancement (LGE) data. ℂNet preserves the complex nature and optimal combination of real and imaginary components of MRI data throughout the reconstruction process by utilizing complex-valued convolution, novel radial batch normalization, and complex activation function layers in a U-Net architecture. A prospectively under-sampled 3D LGE cardiac MRI dataset of 219 patients (17 003 images) at acceleration rates R = 3 through R = 5 was used to evaluate ℂNet. The dataset was further retrospectively under-sampled to a maximum of R = 8 to simulate higher acceleration rates. We created three reconstructions of the 3D LGE dataset using (1) ℂNet, (2) a compressed-sensing-based low-dimensional-structure self-learning and thresholding algorithm (LOST), and (3) a real-valued U-Net (realNet) with the same number of parameters as ℂNet. LOST-reconstructed data were considered the reference for training and evaluation of all models. The reconstructed images were quantitatively evaluated using mean-squared error (MSE) and the structural similarity index measure (SSIM), and subjectively evaluated by three independent readers. Quantitatively, ℂNet-reconstructed images had significantly improved MSE and SSIM values compared with realNet (MSE, 0.077 versus 0.091; SSIM, 0.876 versus 0.733, respectively; p < 0.01). Subjective quality assessment showed that ℂNet-reconstructed image quality was similar to that of compressed sensing and significantly better than that of realNet. ℂNet reconstruction was also more than 300 times faster than compressed sensing. Retrospective under-sampled images demonstrate the potential of ℂNet at higher acceleration rates. ℂNet enables fast reconstruction of highly accelerated 3D MRI with superior performance to real-valued networks, and achieves faster reconstruction than compressed sensing.

3D whole-heart isotropic sub-millimeter resolution coronary magnetic resonance angiography with non-rigid motion-compensated PROST

BACKGROUND

To enable free-breathing whole-heart sub-millimeter resolution coronary magnetic resonance angiography (CMRA) in a clinically feasible scan time by combining low-rank patch-based undersampled reconstruction (3D-PROST) with a highly accelerated non-rigid motion correction framework.

METHODS

Non-rigid motion corrected CMRA combined with 2D image-based navigators has been previously proposed to enable 100% respiratory scan efficiency in modestly undersampled acquisitions. Achieving sub-millimeter isotropic resolution with such techniques still requires prohibitively long acquisition times. We propose to combine 3D-PROST reconstruction with a highly accelerated non-rigid motion correction framework to achieve sub-millimeter resolution CMRA in less than 10 min. Ten healthy subjects and eight patients with suspected coronary artery disease underwent 4-5-fold accelerated free-breathing whole-heart CMRA with 0.9 mm3 isotropic resolution. Vessel sharpness, vessel length and image quality obtained with the proposed non-rigid (NR) PROST approach were compared against translational correction only (TC-PROST) and a previously proposed NR motion-compensated technique (non-rigid SENSE) in healthy subjects. For the patient study, image quality scoring and visual comparison with coronary computed tomography angiography (CCTA) were performed.

RESULTS

Average scan times [min:s] were 6:01 ± 0:59 (healthy subjects) and 8:29 ± 1:41 (patients). In healthy subjects, vessel sharpness of the left anterior descending (LAD) and right (RCA) coronary arteries were improved with the proposed non-rigid PROST (LAD: 51.2 ± 8.8%, RCA: 61.2 ± 9.1%) in comparison to TC-PROST (LAD: 43.8 ± 5.1%, P = 0.051, RCA: 54.3 ± 8.3%, P = 0.218) and non-rigid SENSE (LAD: 46.1 ± 5.8%, P = 0.223, RCA: 56.7 ± 9.6%, P = 0.50), although differences were not statistically significant. The average visual image quality score was significantly higher for NR-PROST (LAD: 3.2 ± 0.6, RCA: 3.3 ± 0.7) compared with TC-PROST (LAD: 2.1 ± 0.6, P = 0.018, RCA: 2.0 ± 0.7, P = 0.014) and non-rigid SENSE (LAD: 2.3 ± 0.5, P = 0.008, RCA: 2.5 ± 0.7, P = 0.016). In patients, the proposed approach showed good delineation of the coronaries, in agreement with CCTA, with image quality scores and vessel sharpness similar to that of healthy subjects.

CONCLUSIONS

We demonstrate the feasibility of combining high undersampling factors with non-rigid motion-compensated reconstruction to obtain high-quality sub-millimeter isotropic CMRA images in ~ 8 min. Validation in a larger cohort of patients with coronary artery disease is now warranted.

From Compressed-Sensing to Artificial Intelligence-Based Cardiac MRI Reconstruction

Cardiac magnetic resonance (CMR) imaging is an important tool for the non-invasive assessment of cardiovascular disease. However, CMR suffers from long acquisition times due to the need of obtaining images with high temporal and spatial resolution, different contrasts, and/or whole-heart coverage. In addition, both cardiac and respiratory-induced motion of the heart during the acquisition need to be accounted for, further increasing the scan time. Several undersampling reconstruction techniques have been proposed during the last decades to speed up CMR acquisition. These techniques rely on acquiring less data than needed and estimating the non-acquired data exploiting some sort of prior information. Parallel imaging and compressed sensing undersampling reconstruction techniques have revolutionized the field, enabling 2- to 3-fold scan time accelerations to become standard in clinical practice. Recent scientific advances in CMR reconstruction hinge on the thriving field of artificial intelligence. Machine learning reconstruction approaches have been recently proposed to learn the non-linear optimization process employed in CMR reconstruction. Unlike analytical methods for which the reconstruction problem is explicitly defined into the optimization process, machine learning techniques make use of large data sets to learn the key reconstruction parameters and priors. In particular, deep learning techniques promise to use deep neural networks (DNN) to learn the reconstruction process from existing datasets in advance, providing a fast and efficient reconstruction that can be applied to all newly acquired data. However, before machine learning and DNN can realize their full potentials and enter widespread clinical routine for CMR image reconstruction, there are several technical hurdles that need to be addressed. In this article, we provide an overview of the recent developments in the area of artificial intelligence for CMR image reconstruction. The underlying assumptions of established techniques such as compressed sensing and low-rank reconstruction are briefly summarized, while a greater focus is given to recent advances in dictionary learning and deep learning based CMR reconstruction. In particular, approaches that exploit neural networks as implicit or explicit priors are discussed for 2D dynamic cardiac imaging and 3D whole-heart CMR imaging. Current limitations, challenges, and potential future directions of these techniques are also discussed.

Accelerated coronary MRI with sRAKI: A database-free self-consistent neural network k-space reconstruction for arbitrary undersampling

Purpose

To accelerate coronary MRI acquisitions with arbitrary undersampling patterns by using a novel reconstruction algorithm that applies coil self-consistency using subject-specific neural networks.

Methods

Self-consistent robust artificial-neural-networks for k-space interpolation (sRAKI) performs iterative parallel imaging reconstruction by enforcing self-consistency among coils. The approach bears similarity to SPIRiT, but extends the linear convolutions in SPIRiT to nonlinear interpolation using convolutional neural networks (CNNs). These CNNs are trained individually for each scan using the scan-specific autocalibrating signal (ACS) data. Reconstruction is performed by imposing the learned self-consistency and data-consistency, which enables sRAKI to support random undersampling patterns. Fully-sampled targeted right coronary artery MRI was acquired in six healthy subjects. The data were retrospectively undersampled, and reconstructed using SPIRiT, l₁-SPIRiT and sRAKI for acceleration rates of 2 to 5. Additionally, prospectively undersampled whole-heart coronary MRI was acquired to further evaluate reconstruction performance.

Results

sRAKI reduces noise amplification and blurring artifacts compared with SPIRiT and l₁-SPIRiT, especially at high acceleration rates in targeted coronary MRI. Quantitative analysis shows that sRAKI outperforms these techniques in terms of normalized mean-squared-error (~44% and ~21% over SPIRiT and l1-SPIRiT at rate 5) and vessel sharpness (~10% and ~20% over SPIRiT and l₁-SPIRiT at rate 5). Whole-heart data shows the sharpest coronary arteries when resolved using sRAKI, with 11% and 15% improvement in vessel sharpness over SPIRiT and l₁-SPIRiT, respectively.

Conclusion

sRAKI is a database-free neural network-based reconstruction technique that may further accelerate coronary MRI with arbitrary undersampling patterns, while improving noise resilience over linear parallel imaging and image sharpness over l₁ regularization techniques.

Free-running cardiac magnetic resonance fingerprinting: Joint T1/T2 map and Cine imaging

PURPOSE

To develop and evaluate a novel non-ECG triggered 2D magnetic resonance fingerprinting (MRF) sequence allowing for simultaneous myocardial T₁ and T₂ mapping and cardiac Cine imaging.

METHODS

Cardiac MRF (cMRF) has been recently proposed to provide joint T₁/T₂ myocardial mapping by triggering the acquisition to mid-diastole and relying on a subject-dependent dictionary of MR signal evolutions to generate the maps. In this work, we propose a novel "free-running" (non-ECG triggered) cMRF framework for simultaneous myocardial T₁ and T₂ mapping and cardiac Cine imaging in a single scan. Free-running cMRF is based on a transient state bSSFP acquisition with tiny golden angle radial readouts, varying flip angle and multiple adiabatic inversion pulses. The acquired data is retrospectively gated into several cardiac phases, which are reconstructed with an approach that combines parallel imaging, low rank modelling and patch-based high-order tensor regularization. Free-running cMRF was evaluated in a standardized phantom and ten healthy subjects. Comparison with reference spin-echo, MOLLI, SASHA, T₂-GRASE and Cine was performed.

RESULTS

T₁ and T₂ values obtained with the proposed approach were in good agreement with reference phantom values (ICC(A,1) > 0.99). Reported values for myocardium septum T₁ were 1043 ± 48 ms, 1150 ± 100 ms and 1160 ± 79 ms for MOLLI, SASHA and free-running cMRF respectively and for T₂ of 51.7 ± 4.1 ms and 44.6 ± 4.1 ms for T₂-GRASE and free-running cMRF respectively. Good agreement was observed between free-running cMRF and conventional Cine 2D ejection fraction (bias = -0.83%).

CONCLUSION

The proposed free-running cardiac MRF approach allows for simultaneous assessment of myocardial T₁ and T₂ and Cine imaging in a single scan.

Image Reconstruction: From Sparsity to Data-adaptive Methods and Machine Learning

The field of medical image reconstruction has seen roughly four types of methods. The first type tended to be analytical methods, such as filtered back-projection (FBP) for X-ray computed tomography (CT) and the inverse Fourier transform for magnetic resonance imaging (MRI), based on simple mathematical models for the imaging systems. These methods are typically fast, but have suboptimal properties such as poor resolution-noise trade-off for CT. A second type is iterative reconstruction methods based on more complete models for the imaging system physics and, where appropriate, models for the sensor statistics. These iterative methods improved image quality by reducing noise and artifacts. The FDA-approved methods among these have been based on relatively simple regularization models. A third type of methods has been designed to accommodate modified data acquisition methods, such as reduced sampling in MRI and CT to reduce scan time or radiation dose. These methods typically involve mathematical image models involving assumptions such as sparsity or low-rank. A fourth type of methods replaces mathematically designed models of signals and systems with data-driven or adaptive models inspired by the field of machine learning. This paper focuses on the two most recent trends in medical image reconstruction: methods based on sparsity or low-rank models, and data-driven methods based on machine learning techniques.

Deep-Learning Methods for Parallel Magnetic Resonance Imaging Reconstruction: A Survey of the Current Approaches, Trends, and Issues

Following the success of deep learning in a wide range of applications, neural network-based machine learning techniques have received interest as a means of accelerating magnetic resonance imaging (MRI). A number of ideas inspired by deep learning techniques from computer vision and image processing have been successfully applied to non-linear image reconstruction in the spirit of compressed sensing for both low dose computed tomography and accelerated MRI. The additional integration of multi-coil information to recover missing k-space lines in the MRI reconstruction process, is still studied less frequently, even though it is the de-facto standard for currently used accelerated MR acquisitions. This manuscript provides an overview of the recent machine learning approaches that have been proposed specifically for improving parallel imaging. A general background introduction to parallel MRI is given that is structured around the classical view of image space and k-space based methods. Both linear and non-linear methods are covered, followed by a discussion of recent efforts to further improve parallel imaging using machine learning, and specifically using artificial neural networks. Image-domain based techniques that introduce improved regularizers are covered as well as k-space based methods, where the focus is on better interpolation strategies using neural networks. Issues and open problems are discussed as well as recent efforts for producing open datasets and benchmarks for the community.

Machine learning in cardiovascular magnetic resonance: basic concepts and applications

Machine learning (ML) is making a dramatic impact on cardiovascular magnetic resonance (CMR) in many ways. This review seeks to highlight the major areas in CMR where ML, and deep learning in particular, can assist clinicians and engineers in improving imaging efficiency, quality, image analysis and interpretation, as well as patient evaluation. We discuss recent developments in the field of ML relevant to CMR in the areas of image acquisition & reconstruction, image analysis, diagnostic evaluation and derivation of prognostic information. To date, the main impact of ML in CMR has been to significantly reduce the time required for image segmentation and analysis. Accurate and reproducible fully automated quantification of left and right ventricular mass and volume is now available in commercial products. Active research areas include reduction of image acquisition and reconstruction time, improving spatial and temporal resolution, and analysis of perfusion and myocardial mapping. Although large cohort studies are providing valuable data sets for ML training, care must be taken in extending applications to specific patient groups. Since ML algorithms can fail in unpredictable ways, it is important to mitigate this by open source publication of computational processes and datasets. Furthermore, controlled trials are needed to evaluate methods across multiple centers and patient groups.

Low-Rank Tensor Models for Improved Multi-Dimensional MRI: Application to Dynamic Cardiac T 1 Mapping

Multi-dimensional, multi-contrast magnetic resonance imaging (MRI) has become increasingly available for comprehensive and time-efficient evaluation of various pathologies, providing large amounts of data and offering new opportunities for improved image reconstructions. Recently, a cardiac phase-resolved myocardial T 1 mapping method has been introduced to provide dynamic information on tissue viability. Improved spatio-temporal resolution in clinically acceptable scan times is highly desirable but requires high acceleration factors. Tensors are well-suited to describe inter-dimensional hidden structures in such multi-dimensional datasets. In this study, we sought to utilize and compare different tensor decomposition methods, without the use of auxiliary navigator data. We explored multiple processing approaches in order to enable high-resolution cardiac phase-resolved myocardial T 1 mapping. Eight different low-rank tensor approximation and processing approaches were evaluated using quantitative analysis of accuracy and precision in T 1 maps acquired in six healthy volunteers. All methods provided comparable T 1 values. However, the precision was significantly improved using local processing, as well as a direct tensor rank approximation. Low-rank tensor approximation approaches are well-suited to enable dynamic T 1 mapping at high spatio-temporal resolutions.

ACCELERATED CORONARY MRI USING 3D SPIRIT-RAKI WITH SPARSITY REGULARIZATION

Coronary MRI is a non-invasive radiation-free imaging tool for the diagnosis of coronary artery disease. One of its limitations is the long scan time, due to the need for high resolution imaging in the presence of respiratory and cardiac motions. Machine learning (ML) methods have been recently utilized to accelerate MRI. In particular, a scan-specific ML technique, called Robust Artifical-neural-network for k-space Interpolation (RAKI) has shown promise in cardiac MRI. However, it requires uniform undersampling. In this study, we sought to extend this approach to arbitrary sampling patterns, using coil self-consistency. This technique, called SPIRiT-RAKI, utilizes scan-specific convolutional neural networks to nonlinearly enforce coil self-consistency. Additionally, regularization terms can also be incorporated. SPIRiT-RAKI was used to accelerate right coronary MRI. Reconstructions were compared to SPIRiT for different undersampling patterns and acceleration rates. Results show SPIRiT-RAKI reduces residual aliasing and blurring artifacts compared to SPIRiT.

High-dimensionality undersampled patch-based reconstruction (HD-PROST) for accelerated multi-contrast MRI

PURPOSE

To develop a new high-dimensionality undersampled patch-based reconstruction (HD-PROST) for highly accelerated 2D and 3D multi-contrast MRI.

METHODS

HD-PROST jointly reconstructs multi-contrast MR images by exploiting the highly redundant information, on a local and non-local scale, and the strong correlation shared between the multiple contrast images. This is achieved by enforcing multi-dimensional low-rank in the undersampled images. 2D magnetic resonance fingerprinting (MRF) phantom and in vivo brain acquisitions were performed to evaluate the performance of HD-PROST for highly accelerated simultaneous T1 and T2 mapping. Additional in vivo experiments for reconstructing multiple undersampled 3D magnetization transfer (MT)-weighted images were conducted to illustrate the impact of HD-PROST for high-resolution multi-contrast 3D imaging.

RESULTS

In the 2D MRF phantom study, HD-PROST provided accurate and precise estimation of the T1 and T2 values in comparison to gold standard spin echo acquisitions. HD-PROST achieved good quality maps for the in vivo 2D MRF experiments in comparison to conventional low-rank inversion reconstruction. T1 and T2 values of white matter and gray matter were in good agreement with those reported in the literature for MRF acquisitions with reduced number of time point images (500 time point images, ~2.5 s scan time). For in vivo MT-weighted 3D acquisitions (6 different contrasts), HD-PROST achieved similar image quality than the fully sampled reference image for an undersampling factor of 6.5-fold.

CONCLUSION

HD-PROST enables multi-contrast 2D and 3D MR images in a short acquisition time without compromising image quality. Ultimately, this technique may increase the potential of conventional parameter mapping.

Compressed sensing MRI: a review from signal processing perspective

Magnetic resonance imaging (MRI) is an inherently slow imaging modality, since it acquires multi-dimensional k-space data through 1-D free induction decay or echo signals. This often limits the use of MRI, especially for high resolution or dynamic imaging. Accordingly, many investigators has developed various acceleration techniques to allow fast MR imaging. For the last two decades, one of the most important breakthroughs in this direction is the introduction of compressed sensing (CS) that allows accurate reconstruction from sparsely sampled k-space data. The recent FDA approval of compressed sensing products for clinical scans clearly reflect the maturity of this technology. Therefore, this paper reviews the basic idea of CS and how this technology have been evolved for various MR imaging problems.

Scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction: Database-free deep learning for fast imaging

PURPOSE

To develop an improved k-space reconstruction method using scan-specific deep learning that is trained on autocalibration signal (ACS) data.

THEORY

Robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction trains convolutional neural networks on ACS data. This enables nonlinear estimation of missing k-space lines from acquired k-space data with improved noise resilience, as opposed to conventional linear k-space interpolation-based methods, such as GRAPPA, which are based on linear convolutional kernels.

METHODS

The training algorithm is implemented using a mean square error loss function over the target points in the ACS region, using a gradient descent algorithm. The neural network contains 3 layers of convolutional operators, with 2 of these including nonlinear activation functions. The noise performance and reconstruction quality of the RAKI method was compared with GRAPPA in phantom, as well as in neurological and cardiac in vivo data sets.

RESULTS

Phantom imaging shows that the proposed RAKI method outperforms GRAPPA at high (≥4) acceleration rates, both visually and quantitatively. Quantitative cardiac imaging shows improved noise resilience at high acceleration rates (rate 4:23% and rate 5:48%) over GRAPPA. The same trend of improved noise resilience is also observed in high-resolution brain imaging at high acceleration rates.

CONCLUSION

The RAKI method offers a training database-free deep learning approach for MRI reconstruction, with the potential to improve many existing reconstruction approaches, and is compatible with conventional data acquisition protocols.

Three-dimensional holographic visualization of high-resolution myocardial scar on HoloLens

Visualization of the complex 3D architecture of myocardial scar could improve guidance of radio-frequency ablation in the treatment of ventricular tachycardia (VT). In this study, we sought to develop a framework for 3D holographic visualization of myocardial scar, imaged using late gadolinium enhancement (LGE), on the augmented reality HoloLens. 3D holographic LGE model was built using the high-resolution 3D LGE image. Smooth endo/epicardial surface meshes were generated using Poisson surface reconstruction. For voxel-wise 3D scar model, every scarred voxel was rendered into a cube which carries the actual resolution of the LGE sequence. For surface scar model, scar information was projected on the endocardial surface mesh. Rendered layers were blended with different transparency and color, and visualized on HoloLens. A pilot animal study was performed where 3D holographic visualization of the scar was performed in 5 swines who underwent controlled infarction and electroanatomic mapping to identify VT substrate. 3D holographic visualization enabled assessment of the complex 3D scar architecture with touchless interaction in a sterile environment. Endoscopic view allowed visualization of scar from the ventricular chambers. Upon completion of the animal study, operator and mapping specialist independently completed the perceived usefulness questionnaire in the six-item usefulness scale. Operator and mapping specialist found it useful (usefulness rating: operator, 5.8; mapping specialist, 5.5; 1–7 scale) to have scar information during the intervention. HoloLens 3D LGE provides a true 3D perception of the complex scar architecture with immersive experience to visualize scar in an interactive and interpretable 3D approach, which may facilitate MR-guided VT ablation.

Content-aware compressive magnetic resonance image reconstruction

This paper describes an adaptive approach to regularizing model-based reconstructions in magnetic resonance imaging to account for local structure or image content. In conjunction with common models like wavelet and total variation sparsity, this content-aware regularization avoids oversmoothing or compromising image features while suppressing noise and incoherent aliasing from accelerated imaging. To evaluate this regularization approach, the experiments reconstruct images from single- and multi-channel, Cartesian and non-Cartesian, brain and cardiac data. These reconstructions combine common analysis-form regularizers and autocalibrating parallel imaging (when applicable). In most cases, the results show widespread improvement in structural similarity and peak-signal-to-error ratio relative to the fully sampled images. These results suggest that this content-aware regularization can preserve local image structures such as edges while providing denoising power superior to sparsity-promoting or sparsity-reweighted regularization.

Isotropic Reconstruction of MR Images Using 3D Patch-Based Self-Similarity Learning

Isotropic three-dimensional (3D) acquisition is a challenging task in magnetic resonance imaging (MRI). Particularly in cardiac MRI, due to hardware and time limitations, current 3D acquisitions are limited by low-resolution, especially in the through-plane direction, leading to poor image quality in that dimension. To overcome this problem, super-resolution (SR) techniques have been proposed to reconstruct a single isotropic 3D volume from multiple anisotropic acquisitions. Previously, local regularization techniques such as total variation have been applied to limit noise amplification while preserving sharp edges and small features in the images. In this paper, inspired by the recent progress in patch-based reconstruction, we propose a novel isotropic 3D reconstruction scheme that integrates non-local and self-similarity information from 3D patch neighborhoods. By grouping 3D patches with similar structures, we enforce the natural sparsity of MR images, which can be expressed by a low-rank structure, leading to robust image reconstruction with high signal-to-noise ratio efficiency. An Augmented Lagrangian formulation of the problem is proposed to efficiently decompose the optimization into a low-rank volume denoising and a SR reconstruction. Experimental results in simulations, brain imaging and clinical cardiac MRI, demonstrate that the proposed joint SR and self-similarity learning framework outperforms current state-of-the-art methods. The proposed reconstruction of isotropic 3D volumes may be particularly useful for cardiac applications, such as myocardial infarction scar assessment by late gadolinium enhancement MRI.

Multi-scale locally low-rank noise reduction for high-resolution dynamic quantitative cardiac MRI

Evaluation of myocardial T₁ times is conventionally limited to a single temporal snapshot of the cardiac cycle, leaving the dependence between functional and tissue characterization unexplored. We recently proposed a technique that alleviates this limitation by acquiring dynamic quantitative myocardial T₁ maps. However, tradeoffs between temporal resolution, scan duration and SNR limit the spatial resolution. In this work, we propose a multi-scale locally low rank noise reduction approach without parameter-tuning to enable high acceleration rates in the acquisition, facilitating superior spatial and temporal resolutions in dynamic myocardial T₁ mapping.

Five-minute whole-heart coronary MRA with sub-millimeter isotropic resolution, 100% respiratory scan efficiency, and 3D-PROST reconstruction

Purpose

To enable whole‐heart 3D coronary magnetic resonance angiography (CMRA) with isotropic sub‐millimeter resolution in a clinically feasible scan time by combining respiratory motion correction with highly accelerated variable density sampling in concert with a novel 3D patch‐based undersampled reconstruction (3D‐PROST).

Methods

An undersampled variable density spiral‐like Cartesian trajectory was combined with 2D image‐based navigators to achieve 100% respiratory efficiency and predictable scan time. 3D‐PROST reconstruction integrates structural information from 3D patch neighborhoods through sparse representation, thereby exploiting the redundancy of the 3D anatomy of the coronary arteries in an efficient low‐rank formulation. The proposed framework was evaluated in a static resolution phantom and in 10 healthy subjects with isotropic resolutions of 1.2 mm³ and 0.9 mm³ and undersampling factors of ×5 and ×9. 3D‐PROST was compared against fully sampled (1.2 mm³ only), conventional parallel imaging, and compressed sensing reconstructions.

Results

Phantom and in vivo (1.2 mm³) reconstructions were in excellent agreement with the reference fully sampled image. In vivo average acquisition times (min:s) were 7:57 ± 1:18 (×5) and 4:35 ± 0:44 (×9) for 0.9 mm³ resolution. Sub‐millimeter 3D‐PROST resulted in excellent depiction of the left and right coronary arteries including small branch vessels, leading to further improvements in vessel sharpness and visible vessel length in comparison with conventional reconstruction techniques. Image quality rated by 2 experts demonstrated that 3D‐PROST provides good image quality and is robust even at high acceleration factors.

Conclusion

The proposed approach enables free‐breathing whole‐heart 3D CMRA with isotropic sub‐millimeter resolution in <5 min and achieves improved coronary artery visualization in a short and predictable scan time.

Gray blood late gadolinium enhancement cardiovascular magnetic resonance for improved detection of myocardial scar

BACKGROUND

Low scar-to-blood contrast in late gadolinium enhanced (LGE) MRI limits the visualization of scars adjacent to the blood pool. Nulling the blood signal improves scar detection but results in lack of contrast between myocardium and blood, which makes clinical evaluation of LGE images more difficult.

METHODS

GB-LGE contrast is achieved through partial suppression of the blood signal using T2 magnetization preparation between the inversion pulse and acquisition. The timing parameters of GB-LGE sequence are determined by optimizing a cost-function representing the desired tissue contrast. The proposed 3D GB-LGE sequence was evaluated using phantoms, human subjects (n = 45) and a swine model of myocardial infarction (n = 5). Two independent readers subjectively evaluated the image quality and ability to identify and localize scarring in GB-LGE compared to black-blood LGE (BB-LGE) (i.e., with complete blood nulling) and conventional (bright-blood) LGE.

RESULTS

GB-LGE contrast was successfully generated in phantoms and all in-vivo scans. The scar-to-blood contrast was improved in GB-LGE compared to conventional LGE in humans (1.1 ± 0.5 vs. 0.6 ± 0.4, P < 0.001) and in animals (1.5 ± 0.2 vs. -0.03 ± 0.2). In patients, GB-LGE detected more tissue scarring compared to BB-LGE and conventional LGE. The subjective scores of the GB-LGE ability for localizing LV scar and detecting papillary scar were improved as compared with both BB-LGE (P < 0.024) and conventional LGE (P < 0.001). In the swine infarction model, GB-LGE scores for the ability to localize LV scar scores were consistently higher than those of both BB-LGE and conventional-LGE.

CONCLUSION

GB-LGE imaging improves the ability to identify and localize myocardial scarring compared to both BB-LGE and conventional LGE. Further studies are warranted to histologically validate GB-LGE.

Increased myocardial native T1 relaxation time in patients with nonischemic dilated cardiomyopathy with complex ventricular arrhythmia

PURPOSE

To study the relationship between diffuse myocardial fibrosis and complex ventricular arrhythmias (ComVA) in patients with nonischemic dilated cardiomyopathy (NICM). We hypothesized that NICM patients with ComVA would have a higher native myocardial T1 time, suggesting more extensive myocardial diffuse fibrosis.

MATERIALS AND METHODS

We prospectively enrolled NICM patients with a history of ComVA (n = 50) and age-matched NICM patients without ComVA (n = 57). Imaging was performed at 1.5T with a protocol that included cine magnetic resonance imaging (MRI) for left ventricular (LV) function, late gadolinium enhancement (LGE) for focal scar, and native T1 mapping for diffuse fibrosis assessment.

RESULTS

Global native T1 time was significantly higher in patients with NICM with ComVA when compared to patients with NICM without ComVA (1131 ± 42 vs. 1107 ± 45 msec, P = 0.006), and this finding remained after excluding segments with scar on LGE (1124 ± 36 vs. 1102 ± 44 msec, P = 0.006). Native T1 was similar in NICM patients with and without the presence of LGE (1121 ± 39 vs. 1117 ± 48 msec, P = 0.68) and mildly correlated with LV end-diastolic volume index (r = 0.27, P = 0.005), LV end-systolic volume index (r = 0.24, P = 0.01), and LV ejection fraction (r = -0.28, P = 0.003). Native T1 value for each 10-msec increment was an independent predictor of ComVA (odds ratio 1.14, 95% confidence interval 1.03-1.25; P = 0.008) beyond LV function and LGE.

CONCLUSION

NICM patients with ComVA have higher native T1 compared to NICM without any documented ComVA. Native myocardial T1 is independently associated with ComVA, after adjusting for LV function and LGE.

LEVEL OF EVIDENCE

2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;47:779-786. In memoriam: The authors are grateful for Dr. Josephson's inspiring guidance and contributions to this study.

Motion guided Spatiotemporal Sparsity for high quality 4D-CBCT reconstruction

Conventional cone-beam computed tomography is often deteriorated by respiratory motion blur, which negatively affects target delineation. On the other side, the four dimensional cone-beam computed tomography (4D-CBCT) can be considered to describe tumor and organ motion. But for current on-board CBCT imaging system, the slow rotation speed limits the projection number at each phase, and the associated reconstructions are contaminated by noise and streak artifacts using the conventional algorithm. To address the problem, we propose a novel framework to reconstruct 4D-CBCT from the under-sampled measurements—Motion guided Spatiotemporal Sparsity (MgSS). In this algorithm, we try to divide the CBCT images at each phase into cubes (3D blocks) and track the cubes with estimated motion field vectors through phase, then apply regional spatiotemporal sparsity on the tracked cubes. Specifically, we recast the tracked cubes into four-dimensional matrix, and use the higher order singular value decomposition (HOSVD) technique to analyze the regional spatiotemporal sparsity. Subsequently, the blocky spatiotemporal sparsity is incorporated into a cost function for the image reconstruction. The phantom simulation and real patient data are used to evaluate this algorithm. Results show that the MgSS algorithm achieved improved 4D-CBCT image quality with less noise and artifacts compared to the conventional algorithms.

A review of GPU-based medical image reconstruction

Tomographic image reconstruction is a computationally demanding task, even more so when advanced models are used to describe a more complete and accurate picture of the image formation process. Such advanced modeling and reconstruction algorithms can lead to better images, often with less dose, but at the price of long calculation times that are hardly compatible with clinical workflows. Fortunately, reconstruction tasks can often be executed advantageously on Graphics Processing Units (GPUs), which are exploited as massively parallel computational engines. This review paper focuses on recent developments made in GPU-based medical image reconstruction, from a CT, PET, SPECT, MRI and US perspective. Strategies and approaches to get the most out of GPUs in image reconstruction are presented as well as innovative applications arising from an increased computing capacity. The future of GPU-based image reconstruction is also envisioned, based on current trends in high-performance computing.

Clinical performance of high-resolution late gadolinium enhancement imaging with compressed sensing

PURPOSE

To evaluate diagnostic image quality of 3D late gadolinium enhancement (LGE) with high isotropic spatial resolution (∼1.4 mm³ ) images reconstructed from randomly undersampled k-space using LOw-dimensional-structure Self-learning and Thresholding (LOST).

MATERIALS AND METHODS

We prospectively enrolled 270 patients (181 men; 55 ± 14 years) referred for myocardial viability assessment. 3D LGE with isotropic spatial resolution of 1.4 ± 0.1 mm³ was acquired at 1.5T using a LOST acceleration rate of 3 to 5. In a subset of 121 patients, 3D LGE or phase-sensitive LGE were acquired with parallel imaging with an acceleration rate of 2 for comparison. Two readers evaluated image quality using a scale of 1 (poor) to 4 (excellent) and assessed for scar presence. The McNemar test statistic was used to compare the proportion of detected scar between the two sequences. We assessed the association between image quality and characteristics (age, gender, torso dimension, weight, heart rate), using generalized linear models.

RESULTS

Overall, LGE detection proportions for 3D LGE with LOST were similar between readers 1 and 2 (16.30% vs. 18.15%). For image quality, readers gave 85.9% and 80.0%, respectively, for images categorized as good or excellent. Overall proportion of scar presence was not statistically different from conventional 3D LGE (28% vs. 33% [P = 0.17] for reader 1 and 26% vs. 31% [P = 0.37] for reader 2). Increasing subject heart rate was associated with lower image quality (estimated slope = -0.009 (P = 0.001)).

CONCLUSION

High-resolution 3D LGE with LOST yields good to excellent image quality in >80% of patients and identifies patients with LV scar at the same rate as conventional 3D LGE.

LEVEL OF EVIDENCE

2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2017;46:1829-1838.