Jointly estimating parametric maps of multiple diffusion models from undersampled q-space data: A comparison of three deep learning approaches.

PURPOSE: While advanced diffusion techniques have been found valuable in many studies, their clinical availability has been hampered partly due to their long scan times. Moreover, each diffusion technique can only extract a few relevant microstructural features. Using multiple diffusion methods may help to better understand the brain microstructure, which requires multiple expensive model fittings. In this work, we compare deep learning (DL) approaches to jointly estimate parametric maps of multiple diffusion representations/models from highly undersampled q-space data. METHODS: We implement three DL approaches to jointly estimate parametric maps of diffusion tensor imaging (DTI), diffusion kurtosis imaging (DKI), neurite orientation dispersion and density imaging (NODDI), and multi-compartment spherical mean technique (SMT). A per-voxel q-space deep learning (1D-qDL), a per-slice convolutional neural network (2D-CNN), and a 3D-patch-based microstructure estimation with sparse coding using a separable dictionary (MESC-SD) network are considered. RESULTS: The accuracy of estimated diffusion maps depends on the q-space undersampling, the selected network architecture, and the region and the parameter of interest. The smallest errors are observed for the MESC-SD network architecture (less than 10 % normalized RMSE in most brain regions). CONCLUSION: Our experiments show that DL methods are very efficient tools to simultaneously estimate several diffusion maps from undersampled q-space data. These methods can significantly reduce both the scan ( ∼ 6-fold) and processing times ( ∼ 25-fold) for estimating advanced parametric diffusion maps while achieving a reasonable accuracy.

Quantitative 3D myocardial perfusion with an efficient arterial input function.

PURPOSE: The purpose of this study was to further develop and combine several innovative sequence designs to achieve quantitative 3D myocardial perfusion. These developments include an optimized 3D stack-of-stars readout (150 ms per beat), efficient acquisition of a 2D arterial input function, tailored saturation pulse design, and potential whole heart coverage during quantitative stress perfusion. THEORY AND METHODS: All studies were performed free-breathing on a Prisma 3T MRI scanner. Phantom validation was used to verify sequence accuracy. A total of 21 subjects (3 patients with known disease) were scanned, 12 with a rest only protocol and 9 with both stress (regadenoson) and rest protocols. First pass quantitative perfusion was performed with gadoteridol (0.075 mmol/kg). RESULTS: Implementation and quantitative perfusion results are shown for healthy subjects and subjects with known coronary disease. Average rest perfusion for the 15 included healthy subjects was 0.79 ± 0.19 mL/g/min, the average stress perfusion for 6 healthy subject studies was 2.44 ± 0.61 mL/g/min, and the average global myocardial perfusion reserve ratio for 6 healthy subjects was 3.10 ± 0.24. Perfusion deficits for 3 patients with ischemia are shown. Average resting heart rate was 59 ± 7 bpm and the average stress heart rate was 81 ± 10 bpm. CONCLUSION: This work demonstrates that a quantitative 3D myocardial perfusion sequence with the acquisition of a 2D arterial input function is feasible at high stress heart rates such as during stress. T1 values and gadolinium concentrations of the sequence match the reference standard well in a phantom, and myocardial rest and stress perfusion and myocardial perfusion reserve values are consistent with those published in literature.

Simultaneous NODDI and GFA parameter map generation from subsampled q-space imaging using deep learning.

PURPOSE: To develop a robust multidimensional deep-learning based method to simultaneously generate accurate neurite orientation dispersion and density imaging (NODDI) and generalized fractional anisotropy (GFA) parameter maps from undersampled q-space datasets for use in stroke imaging. METHODS: Traditional diffusion spectrum imaging (DSI) capable of producing accurate NODDI and GFA parameter maps requires hundreds of q-space samples which renders the scan time clinically untenable. A convolutional neural network (CNN) was trained to generated NODDI and GFA parameter maps simultaneously from 10× undersampled q-space data. A total of 48 DSI scans from 15 stroke patients and 14 normal subjects were acquired for training, validating, and testing this method. The proposed network was compared to previously proposed voxel-wise machine learning based approaches for q-space imaging. Network-generated images were used to predict stroke functional outcome measures. RESULTS: The proposed network achieves significant performance advantages compared to previously proposed machine learning approaches, showing significant improvements across image quality metrics. Generating these parameter maps using CNNs also comes with the computational benefits of only needing to generate and train a single network instead of multiple networks for each parameter type. Post-stroke outcome prediction metrics do not appreciably change when using images generated from this proposed technique. Over three test participants, the predicted stroke functional outcome scores were within 1-6% of the clinical evaluations. CONCLUSIONS: Estimates of NODDI and GFA parameters estimated simultaneously with a deep learning network from highly undersampled q-space data were improved compared to other state-of-the-art methods providing a 10-fold reduction scan time compared to conventional methods.

Optimization and validation of accelerated golden-angle radial sparse MRI reconstruction with self-calibrating GRAPPA operator gridding.

PURPOSE: Golden-angle radial sparse parallel (GRASP) MRI reconstruction requires gridding and regridding to transform data between radial and Cartesian k-space. These operations are repeatedly performed in each iteration, which makes the reconstruction computationally demanding. This work aimed to accelerate GRASP reconstruction using self-calibrating GRAPPA operator gridding (GROG) and to validate its performance in clinical imaging. METHODS: GROG is an alternative gridding approach based on parallel imaging, in which k-space data acquired on a non-Cartesian grid are shifted onto a Cartesian k-space grid using information from multicoil arrays. For iterative non-Cartesian image reconstruction, GROG is performed only once as a preprocessing step. Therefore, the subsequent iterative reconstruction can be performed directly in Cartesian space, which significantly reduces computational burden. Here, a framework combining GROG with GRASP (GROG-GRASP) is first optimized and then compared with standard GRASP reconstruction in 22 prostate patients. RESULTS: GROG-GRASP achieved approximately 4.2-fold reduction in reconstruction time compared with GRASP (∼333 min versus ∼78 min) while maintaining image quality (structural similarity index ≈ 0.97 and root mean square error ≈ 0.007). Visual image quality assessment by two experienced radiologists did not show significant differences between the two reconstruction schemes. With a graphics processing unit implementation, image reconstruction time can be further reduced to approximately 14 min. CONCLUSION: The GRASP reconstruction can be substantially accelerated using GROG. This framework is promising toward broader clinical application of GRASP and other iterative non-Cartesian reconstruction methods. Magn Reson Med 80:286-293, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Validation of highly accelerated real-time cardiac cine MRI with radial k-space sampling and compressed sensing in patients at 1.5T and 3T.

PURPOSE: To validate an optimal 12-fold accelerated real-time cine MRI pulse sequence with radial k-space sampling and compressed sensing (CS) in patients at 1.5T and 3T. METHODS: We used two strategies to reduce image artifacts arising from gradient delays and eddy currents in radial k-space sampling with balanced steady-state free precession readout. We validated this pulse sequence against a standard breath-hold cine sequence in two patient cohorts: a myocardial infarction (n = 16) group at 1.5T and chronic kidney disease group (n = 18) at 3T. Two readers independently performed visual analysis of 68 cine sets in four categories (myocardial definition, temporal fidelity, artifact, noise) on a 5-point Likert scale (1 = nondiagnostic, 2 = poor, 3 = adequate or moderate, 4 = good, 5 = excellent). Another reader calculated left ventricular (LV) functional parameters, including ejection fraction. RESULTS: Compared with standard cine, real-time cine produced nonsignificantly different visually assessed scores, except for the following categories: 1) temporal fidelity scores were significantly lower (P = 0.013) for real-time cine at both field strengths, 2) artifacts scores were significantly higher (P = 0.013) for real-time cine at both field strengths, and 3) noise scores were significantly (P = 0.013) higher for real-time cine at 1.5T. Standard and real-time cine pulse sequences produced LV functional parameters that were in good agreement (e.g., absolute mean difference in ejection fraction <4%). CONCLUSION: This study demonstrates that an optimal 12-fold, accelerated, real-time cine MRI pulse sequence using radial k-space sampling and CS produces good to excellent visual scores and relatively accurate LV functional parameters in patients at 1.5T and 3T. Magn Reson Med 79:2745-2751, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

A review of 3D first-pass, whole-heart, myocardial perfusion cardiovascular magnetic resonance.

A comprehensive review is undertaken of the methods available for 3D whole-heart first-pass perfusion (FPP) and their application to date, with particular focus on possible acceleration techniques. Following a summary of the parameters typically desired of 3D FPP methods, the review explains the mechanisms of key acceleration techniques and their potential use in FPP for attaining 3D acquisitions. The mechanisms include rapid sequences, non-Cartesian k-space trajectories, reduced k-space acquisitions, parallel imaging reconstructions and compressed sensing. An attempt is made to explain, rather than simply state, the varying methods with the hope that it will give an appreciation of the different components making up a 3D FPP protocol. Basic estimates demonstrating the required total acceleration factors in typical 3D FPP cases are included, providing context for the extent that each acceleration method can contribute to the required imaging speed, as well as potential limitations in present 3D FPP literature. Although many 3D FPP methods are too early in development for the type of clinical trials required to show any clear benefit over current 2D FPP methods, the review includes the small but growing quantity of clinical research work already using 3D FPP, alongside the more technical work. Broader challenges concerning FPP such as quantitative analysis are not covered, but challenges with particular impact on 3D FPP methods, particularly with regards to motion effects, are discussed along with anticipated future work in the field.

Sparse BLIP: BLind Iterative Parallel imaging reconstruction using compressed sensing.

PURPOSE: To develop a sensitivity-based parallel imaging reconstruction method to reconstruct iteratively both the coil sensitivities and MR image simultaneously based on their prior information. METHODS: Parallel magnetic resonance imaging reconstruction problem can be formulated as a multichannel sampling problem where solutions are sought analytically. However, the channel functions given by the coil sensitivities in parallel imaging are not known exactly and the estimation error usually leads to artifacts. In this study, we propose a new reconstruction algorithm, termed Sparse BLind Iterative Parallel, for blind iterative parallel imaging reconstruction using compressed sensing. The proposed algorithm reconstructs both the sensitivity functions and the image simultaneously from undersampled data. It enforces the sparseness constraint in the image as done in compressed sensing, but is different from compressed sensing in that the sensing matrix is unknown and additional constraint is enforced on the sensitivities as well. Both phantom and in vivo imaging experiments were carried out with retrospective undersampling to evaluate the performance of the proposed method. RESULTS: Experiments show improvement in Sparse BLind Iterative Parallel reconstruction when compared with Sparse SENSE, JSENSE, IRGN-TV, and L1-SPIRiT reconstructions with the same number of measurements. CONCLUSION: The proposed Sparse BLind Iterative Parallel algorithm reduces the reconstruction errors when compared to the state-of-the-art parallel imaging methods.

Evaluation of highly accelerated real-time cardiac cine MRI in tachycardia.

Electrocardiogram (ECG)-gated breath-hold cine MRI is considered to be the gold standard test for the assessment of cardiac function. However, it may fail in patients with arrhythmia, impaired breath-hold capacity and poor ECG gating. Although ungated real-time cine MRI may mitigate these problems, commercially available real-time cine MRI pulse sequences using parallel imaging typically yield relatively poor spatiotemporal resolution because of their low image acquisition efficiency. As an extension of our previous work, the purpose of this study was to evaluate the diagnostic quality and accuracy of eight-fold-accelerated real-time cine MRI with compressed sensing (CS) for the quantification of cardiac function in tachycardia, where it is challenging for real-time cine MRI to provide sufficient spatiotemporal resolution. We evaluated the performances of eight-fold-accelerated cine MRI with CS, three-fold-accelerated real-time cine MRI with temporal generalized autocalibrating partially parallel acquisitions (TGRAPPA) and ECG-gated breath-hold cine MRI in 21 large animals with tachycardia (mean heart rate, 104 beats per minute) at 3T. For each cine MRI method, two expert readers evaluated the diagnostic quality in four categories (image quality, temporal fidelity of wall motion, artifacts and apparent noise) using a Likert scale (1-5, worst to best). One reader evaluated the left ventricular functional parameters. The diagnostic quality scores were significantly different between the three cine pulse sequences, except for the artifact level between CS and TGRAPPA real-time cine MRI. Both ECG-gated breath-hold cine MRI and eight-fold accelerated real-time cine MRI yielded all four scores of ≥ 3.0 (acceptable), whereas three-fold-accelerated real-time cine MRI yielded all scores below 3.0, except for artifact (3.0). The left ventricular ejection fraction (LVEF) measurements agreed better between ECG-gated cine MRI and eight-fold-accelerated real-time cine MRI (mean difference, -1.6%) than between ECG-gated cine MRI and three-fold-accelerated real-time cine MRI (mean difference, -5.7%). Eight-fold-accelerated real-time cine MRI with CS yields acceptable diagnostic quality and relatively accurate LVEF measurements in the challenging setting of tachycardia.

Ablation-induced left atrial mechanical dysfunction recovers in weeks after ablation.

BACKGROUND: The immediate impact of catheter ablation on left atrial mechanical function and the timeline for its recovery in patients undergoing ablation for atrial fibrillation (AF) remain uncertain. The mechanical function response to catheter ablation in patients with different AF types is poorly understood. METHODS: A total of 113 AF patients were included in this retrospective study. Each patient had three magnetic resonance imaging (MRI) studies in sinus rhythm: one pre-ablation, one immediate post-ablation (within 2 days after ablation), and one post-ablation follow-up MRI (≤ 3 months). We used feature tracking in the MRI cine images to determine peak longitudinal atrial strain (PLAS). We evaluated the change in strain from pre-ablation, immediately after ablation to post-ablation follow-up in a short-term study (< 50 days) and a 3-month study (3 months after ablation). RESULTS: The PLAS exhibited a notable reduction immediately after ablation, compared to both pre-ablation levels and those observed in follow-up studies conducted at short-term (11.1 ± 9.0 days) and 3-month (69.6 ± 39.6 days) intervals. However, there was no difference between follow-up and pre-ablation PLAS. The PLAS returned to 95% pre-ablation level within 10 days. Paroxysmal AF patients had significantly higher pre-ablation PLAS than persistent AF patients in pre-ablation MRIs. Both type AF patients had significantly lower immediate post-ablation PLAS compared with pre-ablation and post-ablation PLAS. CONCLUSION: The present study suggested a significant drop in PLAS immediately after ablation. Left atrial mechanical function recovered within 10 days after ablation. The drop in PLAS did not show a substantial difference between paroxysmal and persistent AF patients.

Model-based reconstruction of undersampled diffusion tensor k-space data.

The practical utility of diffusion tensor imaging, especially for 3D high-resolution spin warp experiments of ex vivo specimens, has been hampered by long acquisition times. To accelerate the acquisition, a compressed sensing framework that uses a model-based formulation to reconstruct diffusion tensor fields from undersampled k-space data was presented and evaluated. Accuracies in brain specimen white matter fiber orientation, fractional anisotropy, and mean diffusivity mapping were compared with alternative methods achievable using the same scan time via reduced image resolution, fewer diffusion encoding directions, standard compressed sensing, or asymmetrical sampling reconstruction. The efficiency of the proposed approach was also compared with fully sampled cases across a range of the number of diffusion encoding directions. In general, the proposed approach was found to reduce the image blurring and noise and to provide more accurate fiber orientation, fractional anisotropy, and mean diffusivity measurements compared with the alternative methods. Moreover, depending on the degree of undersampling used and the diffusion tensor imaging parameter examined, the measurement accuracy of the proposed scheme was equivalent to fully sampled diffusion tensor imaging datasets that consist of 33-67% more encoding directions and require proportionally longer scan times. The findings show model-based compressed sensing to be promising for improving the resolution, accuracy, or scan time of diffusion tensor imaging.

Highly accelerated real-time cardiac cine MRI using k-t SPARSE-SENSE.

For patients with impaired breath-hold capacity and/or arrhythmias, real-time cine MRI may be more clinically useful than breath-hold cine MRI. However, commercially available real-time cine MRI methods using parallel imaging typically yield relatively poor spatio-temporal resolution due to their low image acquisition speed. We sought to achieve relatively high spatial resolution (∼2.5 × 2.5 mm(2)) and temporal resolution (∼40 ms), to produce high-quality real-time cine MR images that could be applied clinically for wall motion assessment and measurement of left ventricular function. In this work, we present an eightfold accelerated real-time cardiac cine MRI pulse sequence using a combination of compressed sensing and parallel imaging (k-t SPARSE-SENSE). Compared with reference, breath-hold cine MRI, our eightfold accelerated real-time cine MRI produced significantly worse qualitative grades (1-5 scale), but its image quality and temporal fidelity scores were above 3.0 (adequate) and artifacts and noise scores were below 3.0 (moderate), suggesting that acceptable diagnostic image quality can be achieved. Additionally, both eightfold accelerated real-time cine and breath-hold cine MRI yielded comparable left ventricular function measurements, with coefficient of variation <10% for left ventricular volumes. Our proposed eightfold accelerated real-time cine MRI with k-t SPARSE-SENSE is a promising modality for rapid imaging of myocardial function.

Rapid ungated myocardial perfusion cardiovascular magnetic resonance: preliminary diagnostic accuracy.

BACKGROUND: Myocardial perfusion cardiovascular magnetic resonance (CMR) is a well-established method for detection of ischemic heart disease. However, ECG gating problems can result in image degradation and non-diagnostic scans, particularly in patients with arrhythmias. METHODS: A turboFLASH saturation recovery pulse sequence was used without any ECG triggering. One saturation pulse followed by 4-5 slices of undersampled radial k-space images was acquired rapidly, on the order of 40-50 msec per image. The acquisition of the set of 4-5 slices was continuously repeated approximately 4 times per second. An iterative constrained reconstruction method was used to reconstruct the ungated images. The ungated perfusion images were post-processed into three different sets of images (ungated, self-gated to near systole, and self-gated to near diastole). To test the ungated approach and compare the different processing methods, 8 patients scheduled for coronary angiography underwent stress and rest perfusion imaging with the ungated acquisition. Six patients had a history of atrial fibrillation (AF). Three blinded readers assessed image quality and presence/absence of disease. RESULTS: All 8 subjects successfully completed the perfusion CMR protocol and 7/8 underwent coronary angiography. Three patients were in atrial fibrillation during CMR. Overall, the CMR images were of high quality as assessed by the three readers. There was little difference in image quality between patients in AF compared to those in sinus rhythm (3.6±0.7 vs. 3.3±0.5). Stress/rest perfusion imaging showed normal perfusion in 4 patients, fixed perfusion defects in 2 patients, and reversible perfusion defects in 2 patients, corresponding with angiographic results. Pooled results from the independent readers gave a sensitivity of 0.92 (CI 0.65-0.99) and specificity of 0.92 (CI 0.65-0.99) for the detection of coronary artery disease using ungated perfusion imaging. The same sensitivity, and a specificity of 1 (CI 0.76-1), was achieved when the images were self-gated after acquisition into near systole or near diastole. CONCLUSIONS: Ungated radial dynamic perfusion CMR can give high quality imaging in patients in sinus rhythm and during atrial fibrillation. In this small cohort, high diagnostic accuracy was possible with this rapid perfusion imaging sequence. An ungated approach simplifies the acquisition and could expand the role of perfusion CMR to include patients with arrhythmia and those with gating problems.

Comparison of Prospective and Retrospective Gated 4D Flow Cardiac MR Image Acquisitions in the Carotid Bifurcation.

PURPOSE: To evaluate the agreement of 4D flow cMRI-derived bulk flow features and fluid (blood) velocities in the carotid bifurcation using prospective and retrospective gating techniques. METHODS: Prospective and retrospective ECG-gated three-dimensional (3D) cine phase-contrast cardiac MRI with three-direction velocity encoding (i.e., 4D flow cMRI) data were acquired in ten carotid bifurcations from men (n = 3) and women (n = 2) that were cardiovascular disease-free. MRI sequence parameters were held constant across all scans except temporal resolution values differed. Velocity data were extracted from the fluid domain and evaluated across the entire volume or at defined anatomic planes (common, internal, external carotid arteries). Qualitative agreement between gating techniques was performed by visualizing flow streamlines and topographical images, and statistical comparisons between gating techniques were performed across the fluid volume and defined anatomic regions. RESULTS: Agreement in the kinematic data (e.g., bulk flow features and velocity data) were observed in the prospectively and retrospectively gated acquisitions. Voxel differences in time-averaged, peak systolic, and diastolic-averaged velocity magnitudes between gating techniques across all volunteers were 2.7%, 1.2%, and 6.4%, respectively. No significant differences in velocity magnitudes or components ([Formula: see text], [Formula: see text], [Formula: see text]) were observed. Importantly, retrospective acquisitions captured increased retrograde flow in the internal carotid artery (i.e., carotid sinus) compared to prospective acquisitions (10.4 ± 6.3% vs. 4.6 ± 5.3%; [Formula: see text] < 0.05). CONCLUSION: Prospective and retrospective ECG-gated 4D flow cMRI acquisitions provide comparable evaluations of fluid velocities, including velocity vector components, in the carotid bifurcation. However, the increased temporal coverage of retrospective acquisitions depicts increased retrograde flow patterns (i.e., disturbed flow) not captured by the prospective gating technique.

k-t ISD: dynamic cardiac MR imaging using compressed sensing with iterative support detection.

Compressed sensing (CS) has been used in dynamic cardiac MRI to reduce the data acquisition time. The sparseness of the dynamic image series in the spatial- and temporal-frequency (x-f) domain has been exploited in existing works. In this article, we propose a new k-t iterative support detection (k-t ISD) method to improve the CS reconstruction for dynamic cardiac MRI by incorporating additional information on the support of the dynamic image in x-f space based on the theory of CS with partially known support. The proposed method uses an iterative procedure for alternating between image reconstruction and support detection in x-f space. At each iteration, a truncated ℓ(1) minimization is applied to obtain the reconstructed image in x-f space using the support information from the previous iteration. Subsequently, by thresholding the reconstruction, we update the support information to be used in the next iteration. Experimental results demonstrate that the proposed k-t ISD method improves the reconstruction quality of dynamic cardiac MRI over the basic CS method in which support information is not exploited.

Myocardial perfusion acquisition without magnetization preparation or gating.

Current myocardial perfusion MRI acquisitions are performed with a saturation recovery sequence, in large part to minimize sensitivity to arrhythmia. A new approach is proposed here where the images are acquired ungated at steady state without use of a saturation pulse. The data are acquired continuously and reach steady state after the first few images. A confluence of advances has made this new paradigm of an ungated steady-state acquisition possible-very rapid undersampled readouts with new reconstruction technologies permit enough measurements that continuous acquisition becomes a feasible approach. Gating can be applied retrospectively from a logged electrocardiogram (ECG) or with self-gating methods. In this work, simulations and measurements in a concentration phantom are used to demonstrate that similar contrast and signal can be obtained with the standard saturation recovery and the proposed spoiled gradient echo (SPGR) acquisition. Specifically, for a flip angle of 14° and a saturation recovery time of 80 ms, similar signals are acquired over a range of T(1) s that reflect realistic myocardial tissue concentrations. Preliminary results in one subject are presented to show the potential of this new approach. The method may allow for cine cardiac perfusion and more signal-to-noise ratio-efficient acquisitions.

Myocardial perfusion MRI with an undersampled 3D stack-of-stars sequence.

PURPOSE: To determine the feasibility of three-dimensional (3D) hybrid radial (stack-of-stars) MRI with spatiotemporal total variation (TV) constrained reconstruction for dynamic contrast enhanced myocardial perfusion imaging. METHODS: An ECG-triggered saturation recovery turboFLASH sequence with undersampled stack-of-stars sampling with spatiotemporal TV constrained reconstruction was developed for dynamic contrast enhanced myocardial perfusion imaging. Simulations were performed to study the dependence of the approach to steady state on flip angle and saturation recovery time for this stack-of-stars acquisition. Phantom studies were used to show the effect of the flip angle selection and imperfect spoiling on image qualities. Studies were done in three humans to test the feasibility of the approach for myocardial perfusion imaging. RESULTS: The simulation and phantom studies showed that imperfect spoiling and magnetization changes during the readout were a function of flip angle and nonoptimized selection of flip angle could degrade the images. Low flip angle acquisitions in the human subjects result in images with good quality similar to multislice radial 2D images. CONCLUSIONS: 3D stack-of-stars sampling with spatiotemporal TV constrained reconstruction provides a promising alternative for myocardial perfusion imaging.

Accelerated cardiac T1 mapping with recurrent networks and cyclic, model-based loss.

BACKGROUND: Using the spin-lattice relaxation time (T1) as a biomarker, the myocardium can be quantitatively characterized using cardiac T1 mapping. The modified Look-Locker inversion (MOLLI) recovery sequences have become the standard clinical method for cardiac T1 mapping. However, the MOLLI sequences require an 11-heartbeat breath-hold that can be difficult for subjects, particularly during exercise or pharmacologically induced stress. Although shorter cardiac T1 mapping sequences have been proposed, these methods suffer from reduced precision. As such, there is an unmet need for accelerated cardiac T1 mapping. PURPOSE: To accelerate cardiac T1 mapping MOLLI sequences by using neural networks to estimate T1 maps using a reduced number of T1-weighted images and their corresponding inversion times. MATERIALS AND METHODS: In this retrospective study, 911 pre-contrast T1 mapping datasets from 202 subjects (128 males, 56 ± 15 years; 74 females, 54 ± 17 years) and 574 T1 mapping post-contrast datasets from 193 subjects (122 males, 57 ± 15 years; 71 females, 54 ± 17 years) were acquired using the MOLLI-5(3)3 sequence and the MOLLI-4(1)3(1)2 sequence, respectively. All acquisition protocols used similar scan parameters: T R = 2.2 ms $TR\; = \;2.2\;{\rm{ms}}$ , T E = 1.12 ms $TE\; = \;1.12\;{\rm{ms}}$ , and F A = 35 ∘ $FA\; = \;35^\circ $ , gadoteridol (ProHance, Bracco Diagnostics) dose ∼ 0.075 mmol / kg $\sim 0.075\;\;{\rm{mmol/kg}}$ . A bidirectional multilayered long short-term memory (LSTM) network with fully connected output and cyclic model-based loss was used to estimate T1 maps from the first three T1-weighted images and their corresponding inversion times for pre- and post-contrast T1 mapping. The performance of the proposed architecture was compared to the three-parameter T1 recovery model using the same reduction of the number of T1-weighted images and inversion times. Reference T1 maps were generated from the scanner using the full MOLLI sequences and the three-parameter T1 recovery model. Correlation and Bland-Altman plots were used to evaluate network performance in which each point represents averaged regions of interest in the myocardium corresponding to the standard American Heart Association 16-segment model. The precision of the network was examined using consecutively repeated scans. Stress and rest pre-contrast MOLLI studies as well as various disease test cases, including amyloidosis, hypertrophic cardiomyopathy, and sarcoidosis were also examined. Paired t-tests were used to determine statistical significance with p < 0.05 $p < 0.05$ . RESULTS: Our proposed network demonstrated similar T1 estimations to the standard MOLLI sequences (pre-contrast: 1260 ± 94 ms $1260 \pm 94\;{\rm{ms}}$ vs. 1254 ± 91 ms $1254 \pm 91\;{\rm{ms}}$ with p = 0.13 $p\; = \;0.13$ ; post-contrast: 484 ± 92 ms $484 \pm 92\;{\rm{ms}}$ vs. 493 ± 91 ms $493 \pm 91\;{\rm{ms}}$ with p = 0.07 $p\; = \;0.07$ ). The precision of standard MOLLI sequences was well preserved with the proposed network architecture ( 24 ± 28 ms $24 \pm 28\;\;{\rm{ms}}$ vs. 18 ± 13 ms $18 \pm 13\;{\rm{ms}}$ ). Network-generated T1 reactivities are similar to stress and rest pre-contrast MOLLI studies ( 5.1 ± 4.0 % $5.1 \pm 4.0\;\% $ vs. 4.9 ± 4.4 % $4.9 \pm 4.4\;\% $ with p = 0.84 $p\; = \;0.84$ ). Amyloidosis T1 maps generated using the proposed network are also similar to the reference T1 maps (pre-contrast: 1243 ± 140 ms $1243 \pm 140\;\;{\rm{ms}}$ vs. 1231 ± 137 ms $1231 \pm 137\;{\rm{ms}}$ with p = 0.60 $p\; = \;0.60$ ; post-contrast: 348 ± 26 ms $348 \pm 26\;{\rm{ms}}$ vs. 346 ± 27 ms $346 \pm 27\;{\rm{ms}}$ with p = 0.89 $p\; = \;0.89$ ). CONCLUSIONS: A bidirectional multilayered LSTM network with fully connected output and cyclic model-based loss was used to generate high-quality pre- and post-contrast T1 maps using the first three T1-weighted images and their corresponding inversion times. This work demonstrates that combining deep learning with cardiac T1 mapping can potentially accelerate standard MOLLI sequences from 11 to 3 heartbeats.

Constrained estimation of the arterial input function for myocardial perfusion cardiovascular magnetic resonance.

Accurate quantification of myocardial perfusion remains challenging due to saturation of the arterial input function at high contrast concentrations. A method for estimating the arterial input function directly from tissue curves in the myocardium that avoids these difficulties is presented. In this constrained alternating minimization with model (CAMM) algorithm, a portion of the left ventricular blood pool signal is also used to constrain the estimation process. Extensive computer simulations assessing the accuracy of kinetic parameter estimation were performed. In 5000 noise realizations, the use of the AIF given by the estimation method returned kinetic parameters with mean Ktrans error of -2% and mean kep error of 0.4%. Twenty in vivo resting perfusion datasets were also processed with this method, and pharmacokinetic parameter values derived from the blind AIF were compared with those derived from a dual-bolus measured AIF. For 17 of the 20 datasets, there were no statistically significant differences in Ktrans estimates, and in aggregate the kinetic parameters were not significantly different from the dual-bolus method. The cardiac constrained alternating minimization with model method presented here provides a promising approach to quantifying perfusion of myocardial tissue with a single injection of contrast agent and without a special pulse sequence though further work is needed to validate the approach in a clinical setting.

A framework for generalized reference image reconstruction methods including HYPR-LR, PR-FOCUSS, and k-t FOCUSS.

PURPOSE: To investigate the relationships among highly constrained back projection (HYPR)-LR, projection reconstruction focal underdetermined system solver (PR-FOCUSS), and k-t FOCUSS by showing how each method relates to a generalized reference image reconstruction method. That is, the generalized series model employs a fixed reference image and multiplicative corrections-that model is extended here to consider reference images more broadly, both in image space and in transform spaces (x-t and x-f spaces), and that can evolve with iteration. MATERIALS AND METHODS: Theoretical relationships between the methods were derived. Computer simulations were done to compare HYPR-LR to one iteration of PR-FOCUSS. The generalized reference approaches applied in the x-t or x-f domain were compared using computer simulation, five cardiac cine imaging datasets, and six myocardial perfusion datasets. RESULTS: PR-FOCUSS and HYPR-LR gave comparable errors, with PR-FOCUSS slightly outperforming HYPR-LR. The baseline image is important to the performance of k-t FOCUSS and x-t FOCUSS, as demonstrated by results from cardiac cine imaging. For cardiac perfusion reconstructions with the use of a temporal average image as the baseline image, k-t FOCUSS gave lower errors than x-t FOCUSS. CONCLUSION: HYPR-LR and PR-FOCUSS are closely related: both work for radial sampling and use reference images in the x-t domain; HYPR-LR is an approximate implementation of the generalized reference framework, while PR-FOCUSS is a conjugate gradient implementation of the generalized reference framework. The superiority of generalized reference approaches applied in the x-t or x-f domain was sensitive to the characteristics of the acquired data and to the baseline image used.

Three-dimensional late gadolinium enhancement imaging of the left atrium with a hybrid radial acquisition and compressed sensing.

PURPOSE: To develop and test a hybrid radial (stack of stars) acquisition and compressed sensing reconstruction for efficient late gadolinium enhancement (LGE) imaging of the left atrium. MATERIALS AND METHODS: Two hybrid radial acquisition schemes, kx-ky-first and kz-first, are tested using the signal equation for an inversion recovery sequence with simulated data. Undersampled data reconstructions are then performed using a compressed sensing approach with a three-dimensional total variation constraint. The data acquisition and reconstruction framework is tested on five atrial fibrillation patients after treatment by radio-frequency ablation. The hybrid radial data are acquired with free breathing without respiratory navigation. RESULTS: The kz-first radial acquisition gave improved image quality as compared to a kx-ky-first scheme. Compressed sensing reconstructions improved the overall quality of undersampled radial LGE images. An image quality metric that takes into account the signal, noise, artifact, and blur for the radial images was 35% (±17%) higher than the corresponding Cartesian acquisitions. Total acquisition time for 36 slices with 1.25 × 1.25 × 2.5 mm(3) resolution was under 3 min for the proposed scheme. CONCLUSION: Hybrid radial LGE imaging of the LA with compressed sensing is a promising approach for obtaining images efficiently and offers more robust image quality than Cartesian acquisitions that were acquired without a respiratory navigator signal.