Evaluation of a Deep Learning Reconstruction for High-Quality T2-Weighted Breast Magnetic Resonance Imaging.

Deep learning (DL) reconstruction techniques to improve MR image quality are becoming commercially available with the hope that they will be applicable to multiple imaging application sites and acquisition protocols. However, before clinical implementation, these methods must be validated for specific use cases. In this work, the quality of standard-of-care (SOC) T2w and a high-spatial-resolution (HR) imaging of the breast were assessed both with and without prototype DL reconstruction. Studies were performed using data collected from phantoms, 20 retrospectively collected SOC patient exams, and 56 prospectively acquired SOC and HR patient exams. Image quality was quantitatively assessed via signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and edge sharpness. Qualitatively, all in vivo images were scored by either two or four radiologist readers using 5-point Likert scales in the following categories: artifacts, perceived sharpness, perceived SNR, and overall quality. Differences in reader scores were tested for significance. Reader preference and perception of signal intensity changes were also assessed. Application of the DL resulted in higher average SNR (1.2-2.8 times), CNR (1.0-1.8 times), and image sharpness (1.2-1.7 times). Qualitatively, the SOC acquisition with DL resulted in significantly improved image quality scores in all categories compared to non-DL images. HR acquisition with DL significantly increased SNR, sharpness, and overall quality compared to both the non-DL SOC and the non-DL HR images. The acquisition time for the HR data only required a 20% increase compared to the SOC acquisition and readers typically preferred DL images over non-DL counterparts. Overall, the DL reconstruction demonstrated improved T2w image quality in clinical breast MRI.

An Anthropomorphic Digital Reference Object (DRO) for Simulation and Analysis of Breast DCE MRI Techniques.

Advances in accelerated magnetic resonance imaging (MRI) continue to push the bounds on achievable spatial and temporal resolution while maintaining a clinically acceptable image quality. Validation tools, including numerical simulations, are needed to characterize the repeatability and reproducibility of such methods for use in quantitative imaging applications. We describe the development of a simulation framework for analyzing and optimizing accelerated MRI acquisition and reconstruction techniques used in dynamic contrast enhanced (DCE) breast imaging. The simulation framework, in the form of a digital reference object (DRO), consists of four modules that control different aspects of the simulation, including the appearance and physiological behavior of the breast tissue as well as the MRI acquisition settings, to produce simulated k-space data for a DCE breast exam. The DRO design and functionality are described along with simulation examples provided to show potential applications of the DRO. The included simulation results demonstrate the ability of the DRO to simulate a variety of effects including the creation of simulated lesions, tissue enhancement modeled by the generalized kinetic model, T1-relaxation, fat signal precession and saturation, acquisition SNR, and changes in temporal resolution.

Comparison of data-driven and general temporal constraints on compressed sensing for breast DCE MRI.

PURPOSE: Current breast DCE-MRI strategies provide high sensitivity for cancer detection but are known to be insufficient in fully capturing rapidly changing contrast kinetics at high spatial resolution across both breasts. Advanced acquisition and reconstruction strategies aim to improve spatial and temporal resolution and increase specificity for disease characterization. In this work, we evaluate the spatial and temporal fidelity of a modified data-driven low-rank-based model (known as MOCCO, model consistency condition) compressed-sensing (CS) reconstruction compared to CS with temporal total variation with radial acquisition for high spatial-temporal breast DCE MRI. METHODS: Reconstruction performance was characterized using numerical simulations of a golden-angle stack-of-stars breast DCE-MRI acquisition at 5-second temporal resolution. Specifically, MOCCO was compared with CS total variation and conventional SENSE reconstructions. The temporal model for MOCCO was prelearned over the source data, whereas CS total variation was performed using a first-order temporal gradient sparsity transform. RESULTS: The MOCCO reconstruction was able to capture rapid lesion kinetics while providing high image quality across a range of optimal regularization values. It also recovered kinetics in small lesions (1.5 mm) in line-profile analysis and error images, whereas g-factor maps showed relatively low and constant values with no significant artifacts. The CS-TV method demonstrated either recovery of high spatial resolution with reduced temporal accuracy using large regularization values, or recovery of rapid lesion kinetics with reduced image quality using low regularization values. CONCLUSION: Simulations demonstrated that MOCCO with radial acquisition provides a robust imaging technique for improving temporal fidelity, while maintaining high spatial resolution and image quality in the setting of bilateral breast DCE MRI.

Thrombus-mimicking artifacts in two-point Dixon MRI: Prevalence, appearance, and severity.

PURPOSE: To evaluate the incidence and severity of potentially thrombus mimicking, flow-induced misallocation artifacts in a clinical setting. Two-point "Dixon" fat-water separation methods, with bipolar readout gradients, may suffer from flow-induced fat-water misallocation artifacts. If these artifacts occur within blood vessels, they may mimic thrombus. MATERIALS AND METHODS: Two-point Dixon coronal and axial images acquired in 102 consecutive patients were retrospectively evaluated for the presence of flow-induced artifacts in arteries and veins. Artifacts were graded on a 3-point scale (none, mild, severe) by two independent readers. Interreader agreement was evaluated with kappa statistics. RESULTS: Reader 1 reported 63 artifacts in 46 (45%) of the cases (severe in 19 cases, 18.6%). Reader 2 reported 51 artifacts in 43 (42.2%) of the cases (severe in 18 cases, 17.6%). Misallocation of fat and water was apparent in all datasets with severe artifacts, whereas variable signal intensity changes in water and fat images were observed in mild artifacts. Interreader agreement was good for artifacts appearing in coronal images (κ = 0.7) and fair for artifact appearance in axial images (κ = 0.24). CONCLUSION: Our study shows a high incidence of flow-induced mild and severe artifacts in a two-point Dixon method with bipolar readout gradients. This artifact should not be misinterpreted as intravascular thrombus. LEVEL OF EVIDENCE: 3 J. Magn. Reson. Imaging 2017;45:229-236.

Novel High Spatiotemporal Resolution Versus Standard-of-Care Dynamic Contrast-Enhanced Breast MRI: Comparison of Image Quality.

OBJECTIVE: Currently, dynamic contrast-enhanced (DCE) breast magnetic resonance imaging (MRI) prioritizes spatial resolution over temporal resolution given the limitations of acquisition techniques. The purpose of our intrapatient study was to assess the ability of a novel high spatial and high temporal resolution DCE breast MRI method to maintain image quality compared with the clinical standard-of-care (SOC) MRI. MATERIALS AND METHODS: Thirty patients, each demonstrating a focal area of enhancement (29 benign, 1 cancer) on their SOC MRI, consented to undergo a research DCE breast MRI on a second date. For the research DCE MRI, a method (DIfferential Subsampling with Cartesian Ordering [DISCO]) using pseudorandom k-space sampling, view sharing reconstruction, 2-point Dixon fat-water separation, and parallel imaging was used to produce images with an effective temporal resolution 6 times faster than the SOC MRI (27 vs 168 seconds, respectively). Both the SOC and DISCO MRI scans were acquired with matching spatial resolutions of 0.8 × 0.8 × 1.6 mm. Image quality (distortion/artifacts, resolution, fat suppression, lesion conspicuity, perceived signal-to-noise ratio, and overall image quality) was scored by 3 radiologists in a blinded reader study. RESULTS: Differences in image quality scores between the DISCO and SOC images were all less than 0.8 on a 10-point scale, and both methods were assessed as providing diagnostic image quality in all cases. DISCO images with the same high spatial resolution, but 6 times the effective temporal resolution as the SOC MRI scans, were produced, yielding 20 postcontrast time points with DISCO compared with 3 for the SOC MRI, over the same total time interval. CONCLUSIONS: DISCO provided comparable image quality compared with the SOC MRI, while also providing 6 times faster effective temporal resolution and the same high spatial resolution.

Flow-induced signal misallocation artifacts in two-point fat-water chemical shift MRI.

PURPOSE: Two-point fat-water separation methods are increasingly being used for chest and abdominal MRI and have recently been introduced for use in MR angiography of the lower extremities. With these methods, flowing spins can accumulate unintended phase shifts between the echo times. The purpose of this study is to demonstrate that these phase shifts can lead to inaccurate signals in the water and fat images. THEORY AND METHODS: In vitro experiments were conducted at 1.5T and 3.0T using a stenosis-mimicking phantom and a computer-controlled pump to image a range of physiologically relevant velocities. RESULTS: In the phantom images acquired using bipolar readout gradients, fat-water signal inaccuracies were visible in regions of flow, with increasing severity as the flow rate was increased. Additionally, similar effects were observed in regions of high flow in clinical chest and liver exams. In the phantom images, the effect was eliminated by using a dual-pass method without bipolar readout gradients. CONCLUSION: When using fat-water separation methods with bipolar readout gradients, phase shifts caused by the motion of spins can lead to signal inaccuracies in the fat and water images. These artifacts can be mitigated by using approaches that do not use bipolar readout gradients.

Combined dynamic contrast-enhanced liver MRI and MRA using interleaved variable density sampling.

PURPOSE: To develop and evaluate a method for volumetric contrast-enhanced MRI of the liver, with high spatial and temporal resolutions, for combined dynamic imaging and MR angiography (MRA) using a single injection of contrast agent. METHODS: An interleaved variable density (IVD) undersampling pattern was implemented in combination with a real-time-triggered, time-resolved, dual-echo 3D spoiled gradient echo sequence. Parallel imaging autocalibration lines were acquired only once during the first time frame. Imaging was performed in 10 subjects with focal nodular hyperplasia (FNH) and compared with their clinical MRI. The angiographic phase of the proposed method was compared with a dedicated MR angiogram acquired during a second injection of contrast. RESULTS: A total of 21 FNH, three cavernous hemangiomas, and 109 arterial segments were visualized in 10 subjects. The temporally resolved images depicted the characteristic arterial enhancement pattern of the lesions with a 4-s update rate. Images were graded as having significantly higher quality compared with the clinical MRI. Angiograms produced from the IVD method provided noninferior diagnostic assessment compared with the dedicated MR angiogram. CONCLUSION: Using an undersampled IVD imaging method, we have demonstrated the feasibility of obtaining high spatial and temporal resolution dynamic contrast-enhanced imaging and simultaneous MRA of the liver.

Application of direct virtual coil to dynamic contrast-enhanced MRI and MR angiography with data-driven parallel imaging.

PURPOSE: To demonstrate the feasibility of direct virtual coil (DVC) in the setting of 4D dynamic imaging used in multiple clinical applications. THEORY AND METHODS: Three dynamic imaging applications were chosen: pulmonary perfusion, liver perfusion, and peripheral MR angiography (MRA), with 18, 11, and 10 subjects, respectively. After view-sharing, the k-space data were reconstructed twice: once with channel-by-channel (CBC) followed by sum-of-squares coil combination and once with DVC. Images reconstructed using CBC and DVC were compared and scored based on overall image quality by two experienced radiologists using a five-point scale. RESULTS: The CBC and DVC showed similar image quality in image domain. Time course measurements also showed good agreement in the temporal domain. CBC and DVC images were scored as equivalent for all pulmonary perfusion cases, all liver perfusion cases, and four of the 10 peripheral MRA cases. For the remaining six peripheral MRA cases, DVC were scored as slightly better (not clinically significant) than the CBC images by Radiologist A and as equivalent by Radiologist B. CONCLUSION: For dynamic contrast-enhanced MR applications, it is clinically feasible to reduce image reconstruction time while maintaining image quality and time course measurement using the DVC technique.

Use of a computer-controlled motion phantom to investigate the temporal and spatial fidelity of HYPR processing.

PURPOSE: In this work, we investigate the spatial and temporal fidelity of highly constrained backPRojection (HYPR) processing using a computer-controlled motion phantom. The goal of this experimental set-up was to provide not only well-defined temporal dynamics and spatial characteristics of the motion phantom, but also circumstances that imitate in vivo scenarios. METHODS: The phantom was designed to represent an artery flanked on both sides by vein. Both arterial and venous components have different temporal dynamics but are confluent, which corresponds to a difficult scenario for HYPR. Spatial and temporal fidelity was investigated by measuring signal intensity profiles through the phantom both orthogonal to as well as along the direction of motion. RESULTS: Spatial fidelity profiles measured from the HYPR processed images yielded full-width-at-half-maximum values very similar to those measured in non-HYPR-processed images. Furthermore, there was no significant spreading of the motion phantom leading edge in HYPR processed images. CONCLUSION: Although HYPR processing has certain characteristic artifacts that are discussed, the technique can be used to improve image quality of highly undersampled time frame images with minimal loss of spatial or temporal fidelity.

Pulmonary perfusion MRI using interleaved variable density sampling and HighlY constrained cartesian reconstruction (HYCR).

PURPOSE: To demonstrate the feasibility of performing single breathhold, noncardiac gated, ultrafast, high spatial-temporal resolution whole chest MR pulmonary perfusion imaging in humans. MATERIALS AND METHODS: Eight subjects (five male, three female) were scanned with the proposed method on a 3 Tesla clinical scanner using a 32-channel phased-array coil. Seven (88%) were healthy volunteers, and one was a patient volunteer with sarcoidosis. The peak lung enhancement phase for each subject was scored for gravitational effect, peak parenchymal enhancement and severity of artifacts by three cardiothoracic radiologists independently. RESULTS: All studies were successfully performed by MR technologists without any additional training. Mean parenchymal signal was very good, measuring 0.78 ± 0.13 (continuous scale, 0 = "none" → 1 = "excellent"). Mean level of motion artifacts was low, measuring 0.13 ± 0.08 (continuous scale, 0 = "none" → 1 = "severe"). CONCLUSION: It is feasible to perform single breathhold, noncardiac gated, ultrafast, high spatial-temporal resolution whole chest MR pulmonary perfusion imaging in humans.

Interleaved variable density sampling with a constrained parallel imaging reconstruction for dynamic contrast-enhanced MR angiography.

For MR applications such as contrast-enhanced MR angiography, it is desirable to achieve simultaneously high spatial and temporal resolution. The current clinical standard uses view-sharing methods combined with parallel imaging; however, this approach still provides limited spatial and temporal resolution. To improve on the clinical standard, we present an interleaved variable density (IVD) sampling method that pseudorandomly undersamples each individual frame of a 3D Cartesian ky-kz plane combined with parallel imaging acceleration. From this dataset, time-resolved images are reconstructed with a method that combines parallel imaging with a multiplicative constraint. Total acceleration factors on the order of 20 are achieved for contrast-enhanced MR angiography of the lower extremities, and improvements in temporal fidelity of the depiction of the contrast bolus passage are demonstrated relative to the clinical standard.

Time resolved contrast enhanced intracranial MRA using a single dose delivered as sequential injections and highly constrained projection reconstruction (HYPR CE).

Time-resolved contrast-enhanced magnetic resonance angiography of the brain is challenging due to the need for rapid imaging and high spatial resolution. Moreover, the significant dispersion of the intravenous contrast bolus as it passes through the heart and lungs increases the overlap between arterial and venous structures, regardless of the acquisition speed and reconstruction window. An innovative technique is presented that divides a single dose contrast into two injections. Initially a small volume of contrast material (2-3 mL) is used to acquiring time-resolved weighting images with a high frame rate (2 frames/s) during the first pass of the contrast agent. The remaining contrast material is used to obtain a high resolution whole brain contrast-enhanced (CE) magnetic resonance angiography (0.57 × 0.57 × 1 mm(3) ) that is used as the spatial constraint for Local Highly Constrained Projection Reconstruction (HYPR LR) reconstruction. After HYPR reconstruction, the final dynamic images (HYPR CE) have both high temporal and spatial resolution. Furthermore, studies of contrast kinetics demonstrate that the shorter bolus length from the reduced contrast volume used for the first injection significantly improves the arterial and venous separation.

Automated vessel segmentation using cross-correlation and pooled covariance matrix analysis.

Time-resolved contrast-enhanced magnetic resonance angiography (CE-MRA) provides contrast dynamics in the vasculature and allows vessel segmentation based on temporal correlation analysis. Here we present an automated vessel segmentation algorithm including automated generation of regions of interest (ROIs), cross-correlation and pooled sample covariance matrix analysis. The dynamic images are divided into multiple equal-sized regions. In each region, ROIs for artery, vein and background are generated using an iterative thresholding algorithm based on the contrast arrival time map and contrast enhancement map. Region-specific multi-feature cross-correlation analysis and pooled covariance matrix analysis are performed to calculate the Mahalanobis distances (MDs), which are used to automatically separate arteries from veins. This segmentation algorithm is applied to a dual-phase dynamic imaging acquisition scheme where low-resolution time-resolved images are acquired during the dynamic phase followed by high-frequency data acquisition at the steady-state phase. The segmented low-resolution arterial and venous images are then combined with the high-frequency data in k-space and inverse Fourier transformed to form the final segmented arterial and venous images. Results from volunteer and patient studies demonstrate the advantages of this automated vessel segmentation and dual phase data acquisition technique.

CE-MRA of the lower extremities using HYPR stack-of-stars.

PURPOSE: To investigate the properties of HYPR (HighlY constrained back PRojection) processing-the temporal fidelity and the improvements of spatial/temporal resolution-for contrast-enhanced MR angiography in a pilot study of the lower extremities in healthy volunteers. MATERIALS AND METHODS: HYPR processing with a radial three-dimensional (3D) stack-of-stars acquisition was investigated for contrast-enhanced MR angiography of the lower extremities in 15 healthy volunteers. HYPR images were compared with control images acquired using a fast, multiphase, 2D Cartesian method to verify the temporal fidelity of HYPR. HYPR protocols were developed for achieving either a high frame update rate or a minimal slice thickness by adjusting the acquisition parameters. HYPR images were compared with images obtained using 3D TRICKS, a widely used protocol in dynamic 3D MRA. RESULTS: HYPR images showed good temporal agreement with 2D control images. In comparison with TRICKS, HYPR stack-of-stars demonstrated higher spatial and temporal resolution. High radial undersampling factors for each time frame were permitted, typically approximately 50 to 100 compared with fully sampled radial imaging. CONCLUSION: In this feasibility study, HYPR processing has been demonstrated to improve the spatial or temporal resolution in peripheral CE-MRA.

MR hysterosalpingography with an angiographic time-resolved 3D pulse sequence: assessment of tubal patency.

OBJECTIVE: The purpose of our study was to determine if tubal patency can be assessed by MR hysterosalpingography (HSG) using a clinically available MR angiographic sequence (3D time-resolved imaging of contrast kinetics [TRICKS]). This capability would enhance the value of MRI in women with infertility. CONCLUSION: MR HSG effectively shows tubal patency and can be considered when both conventional HSG and standard MRI are necessary for the evaluation of women with infertility, such as in women with suspected uterine anomalies or extrauterine disease.

Evaluation of temporal and spatial characteristics of 2D HYPR processing using simulations.

Highly constrained back-projection (HYPR) is a data acquisition and reconstruction method that provides very rapid frame update rates and very high spatial resolution for a time series of images while maintaining a good signal-to-noise ratio and high image quality. In this study we used simulations to evaluate the temporal and spatial characteristics of images produced using the HYPR algorithm. The simulations demonstrate that spatial accuracy is well maintained in the images and the temporal changes in signal intensity are represented with high fidelity. The waveforms representing signal intensity as a function of time obtained from regions-of-interest placed in simulated objects track the true curves very well, with variations from the truth occurring only when objects with very different temporal behavior are very close to each other. However, even when objects with different temporal characteristics are touching, their influences on each other are small.

Whole-body MR angiography using variable density sampling and dual-injection bolus-chase acquisition.

Conventional bolus-chase acquisition generates peripheral runoff images using a single injection of the contrast material. Low spatial resolution, small slice coverage and venous contamination are major problems especially in the distal stations. A technique is presented herein in which whole-body magnetic resonance angiography is performed using a dual-contrast-injection four-station acquisition protocol. Bolus sharing was performed between two stations: the abdomen and calf stations share the first bolus injection, while the thorax and thigh stations share the second bolus injection. The combination of variable density sampling and elliptical centric acquisition order was applied to the abdomen and thorax stations. The scan time was extended to generate high spatial resolution arterial phase images with broad slice coverage for the calf and thigh stations. The feasibility of this technique was demonstrated using phantom and in vivo human volunteer studies.

Imaging of lung ventilation and respiratory dynamics in a single ventilation cycle using hyperpolarized He-3 MRI.

PURPOSE: To image respiratory dynamics and three-dimensional (3D) ventilation during inhalation, breath-hold, and exhalation for evaluation of obstructive lung disease using a single dose of hyperpolarized (HP) He-3 during MRI. MATERIALS AND METHODS: A single 2D-3D projections inside Z encoding (PRIZE)-2D acquisition was performed that consisted of a rapid 2D radial acquisition phase during inhalation of the HP He-3, a 3D acquisition phase during a breath-hold interval, and finally the same 2D radial acquisition during a forced exhalation maneuver followed by tidal breathing. The 3D PRIZE acquisition was comprised of radial sampling in the coronal plane and Fourier encoding in the patient's anterior-posterior direction. Nine patients with mild/moderate to severe asthma were studied (two individuals were studied twice) using this technique. RESULTS: Breath-hold and dynamic imaging results showed physiological abnormalities and were compared with results from standard spirometry, body plethysmography, and computed tomography (CT). Dynamic images depicted regions of differential gas clearance and trapping observed during and after forced exhalation that were corroborated as regions of air trapping on CT imaging. CONCLUSION: The 2D-3D PRIZE-2D acquisition allowed for 3D depiction of ventilation during a breath-hold, as well as detection of gas trapping. Imaging results were confirmed with spirometry, body plethysmography, and CT.