Faculty Bibliography

Small variations in left-ventricular preload due to respiration produce measurable changes in cardiac function in normal subjects. We show that this mechanism is altered in patients with reduced ejection fraction (EF), hypertrophy, or volume-loaded right ventricle (RV). We propose a multi-dimensional retrospective image reconstruction, based on an adaptive, soft classification of data into respiratory and cardiac phases, to study these effects.

PURPOSE: Free-breathing whole-heart coronary MR angiography (MRA) commonly uses navigators to gate respiratory motion, resulting in lengthy and unpredictable acquisition times. Conversely, self-navigation has 100% scan efficiency, but requires motion correction over a broad range of respiratory displacements, which may introduce image artifacts. We propose replacing navigators and self-navigation with a respiratory motion-resolved reconstruction approach. METHODS: Using a respiratory signal extracted directly from the imaging data, individual signal-readouts are binned according to their respiratory states. The resultant series of undersampled images are reconstructed using an extradimensional golden-angle radial sparse parallel imaging (XD-GRASP) algorithm, which exploits sparsity along the respiratory dimension. Whole-heart coronary MRA was performed in 11 volunteers and four patients with the proposed methodology. Image quality was compared with that obtained with one-dimensional respiratory self-navigation. RESULTS: Respiratory-resolved reconstruction effectively suppressed respiratory motion artifacts. The quality score for XD-GRASP reconstructions was greater than or equal to self-navigation in 80/88 coronary segments, reaching diagnostic quality in 61/88 segments versus 41/88. Coronary sharpness and length were always superior for the respiratory-resolved datasets, reaching statistical significance (P < 0.05) in most cases. CONCLUSION: XD-GRASP represents an attractive alternative for handling respiratory motion in free-breathing whole heart MRI and provides an effective alternative to self-navigation. Magn Reson Med, 2016. (c) 2016 Wiley Periodicals, Inc.

The introduction of compressed sensing for increasing imaging speed in magnetic resonance imaging (MRI) has raised significant interest among researchers and clinicians, and has initiated a large body of research across multiple clinical applications over the last decade. Compressed sensing aims to reconstruct unaliased images from fewer measurements than are traditionally required in MRI by exploiting image compressibility or sparsity. Moreover, appropriate combinations of compressed sensing with previously introduced fast imaging approaches, such as parallel imaging, have demonstrated further improved performance. The advent of compressed sensing marks the prelude to a new era of rapid MRI, where the focus of data acquisition has changed from sampling based on the nominal number of voxels and/or frames to sampling based on the desired information content. This article presents a brief overview of the application of compressed sensing techniques in body MRI, where imaging speed is crucial due to the presence of respiratory motion along with stringent constraints on spatial and temporal resolution. The first section provides an overview of the basic compressed sensing methodology, including the notion of sparsity, incoherence, and nonlinear reconstruction. The second section reviews state-of-the-art compressed sensing techniques that have been demonstrated for various clinical body MRI applications. In the final section, the article discusses current challenges and future opportunities. LEVEL OF EVIDENCE: 5 J. Magn. Reson. Imaging 2016.

BACKGROUND: Arrhythmia can significantly alter the image quality of cardiovascular magnetic resonance (CMR); automatic detection and sorting of the most frequent types of arrhythmias during the CMR acquisition could potentially improve image quality. New CMR techniques, such as non-Cartesian CMR, can allow self-gating: from cardiac motion-related signal changes, we can detect cardiac cycles without an electrocardiogram. We can further use this data to obtain a surrogate for RR intervals (valley intervals: VV). Our purpose was to evaluate the feasibility of an automated method for classification of non-arrhythmic (NA) (regular cycles) and arrhythmic patients (A) (irregular cycles), and for sorting of common arrhythmia patterns between atrial fibrillation (AF) and premature ventricular contraction (PVC), using the cardiac motion-related signal obtained during self-gated free-breathing radial cardiac cine CMR with compressed sensing reconstruction (XD-GRASP). METHODS: One hundred eleven patients underwent cardiac XD-GRASP CMR between October 2015 and February 2016; 33 were included for retrospective analysis with the proposed method (6 AF, 8 PVC, 19 NA; by recent ECG). We analyzed the VV, using pooled statistics (histograms) and sequential analysis (Poincare plots), including the median (medVV), the weighted mean (meanVV), the total number of VV values (VVval), and the total range (VVTR) and half range (VVHR) of the cumulative frequency distribution of VV, including the median to half range (medVV/VVHR) and the half range to total range (VVHR/VVTR) ratios. We designed a simple algorithm for using the VV results to differentiate A from NA, and AF from PVC. RESULTS: Between NA and A, meanVV, VVval, VVTR, VVHR, medVV/VVHR and VVHR/VVTR ratios were significantly different (p values = 0.00014, 0.0027, 0.000028, 5x10-9, 0.002, respectively). Between AF and PVC, meanVV, VVval and medVV/VVHR ratio were significantly different (p values = 0.018, 0.007, 0.044, respectively). Using our algorithm, sensitivity, specificity, and accuracy were 93 %, 95 % and 94 % to discriminate between NA and A, and 83 %, 71 %, and 77 % to discriminate between AF and PVC, respectively; areas under the ROC curve were 0.93 and 0.89. CONCLUSIONS: Our study shows we can reliably detect arrhythmias and differentiate AF from PVC, using self-gated cardiac cine XD-GRASP CMR.

PURPOSE: To develop a novel framework for free-breathing MRI called XD-GRASP, which sorts dynamic data into extra motion-state dimensions using the self-navigation properties of radial imaging and reconstructs the multidimensional dataset using compressed sensing. METHODS: Radial k-space data are continuously acquired using the golden-angle sampling scheme and sorted into multiple motion-states based on respiratory and/or cardiac motion signals derived directly from the data. The resulting undersampled multidimensional dataset is reconstructed using a compressed sensing approach that exploits sparsity along the new dynamic dimensions. The performance of XD-GRASP is demonstrated for free-breathing three-dimensional (3D) abdominal imaging, two-dimensional (2D) cardiac cine imaging and 3D dynamic contrast-enhanced (DCE) MRI of the liver, comparing against reconstructions without motion sorting in both healthy volunteers and patients. RESULTS: XD-GRASP separates respiratory motion from cardiac motion in cardiac imaging, and respiratory motion from contrast enhancement in liver DCE-MRI, which improves image quality and reduces motion-blurring artifacts. CONCLUSION: XD-GRASP represents a new use of sparsity for motion compensation and a novel way to handle motions in the context of a continuous acquisition paradigm. Instead of removing or correcting motion, extra motion-state dimensions are reconstructed, which improves image quality and also offers new physiological information of potential clinical value. Magn Reson Med, 2015. (c) 2015 Wiley Periodicals, Inc.

BACKGROUND: To evaluate the influence of temporal sparsity regularization and radial undersampling on compressed sensing reconstruction of dynamic contrast-enhanced (DCE) MRI, using the iterative Golden-angle RAdial Sparse Parallel (iGRASP) MRI technique in the setting of breast cancer evaluation. METHODS: DCE-MRI examinations of the breast (n = 7) were conducted using iGRASP at 3 Tesla. Images were reconstructed with five different radial undersampling schemes corresponding to temporal resolutions between 2 and 13.4 s/frame and with four different weights for temporal sparsity regularization (lambda = 0.1, 0.5, 2, and 6 times of noise level). Image similarity to time-averaged reference images was assessed by two breast radiologists and using quantitative metrics. Temporal similarity was measured in terms of wash-in slope and contrast kinetic model parameters. RESULTS: iGRASP images reconstructed with lambda = 2 and 5.1 s/frame had significantly (P < 0.05) higher similarity to time-averaged reference images than the images with other reconstruction parameters (mutual information (MI) >5%), in agreement with the assessment of two breast radiologists. Higher undersampling (temporal resolution < 5.1 s/frame) required stronger temporal sparsity regularization (lambda >/= 2) to remove streaking aliasing artifacts (MI > 23% between lambda = 2 and 0.5). The difference between the kinetic-model transfer rates of benign and malignant groups decreased as temporal resolution decreased (82% between 2 and 13.4 s/frame). CONCLUSION: This study demonstrates objective spatial and temporal similarity measures can be used to assess the influence of sparsity constraint and undersampling in compressed sensing DCE-MRI and also shows that the iGRASP method provides the flexibility of optimizing these reconstruction parameters in the postprocessing stage using the same acquired data. J. Magn. Reson. Imaging 2015.

Background: Three-dimensional (3D) virtual models are valuable tools that may help to better understand complex cardiovascular anatomy and facilitate surgical planning in patients with congenital heart disease (CHD). Although computed tomography (CT) images are used most commonly to create these models [1,2], Magnetic Resonance Imaging (MRI) may be an attractive alternative, since it offers superior soft-tissue characterization and flexible image contrast mechanisms, and avoids the use of ionizing radiation. However, segmentation on MRI images is inherently challenging due to noise/artifacts, magnetic field inhomogeneity, and relatively lower spatial resolution compared to CT. The purpose of this study was to evaluate the image quality and assess the feasibility of creating virtual 3D heart models using a novel prototype 3D whole heart self-navigated radial MRI technique. Methods: Free-breathing self-navigated whole heart MRI was performed on three pediatric patients: two with complex CHD (average age=17 months) and one with normal cardiac anatomy (age=17years), using a 3D radial, non-slice-selective, T2-prepared, fat-saturated bSSFP sequence on a 1.5T MRI scanner (MAGNETOM Aera, Siemens, Germany). The acquisition window (~50-55 ms) was placed in mid-diastole and was adapted for different heart rates. Imaging parameters were as follows: TR/TE=3.1/1.56 ms, FOV=200 mm3, voxel size=1 mm3, FA=115degree, and acquisition time=5-6 minutes (~12000 radial lines). Respiratory motion correction and image reconstruction was performed on the scanner as described in [3]. For comparison, conventional non-gated 3D FLASH or navigator-gated 3D bSSFP sequences were also performed. All results were blinded and randomized for image quality assessment by one pediatric cardiologist and one cardiac radiologist using a five-point scale (1=non-diagnostic, 2=poor, 3=adequate, 4=good, 5=excellent). Statistical analysis was performed to compare mean scores. DICOM images were imported to a 3D workstation (Mimics, Materialise, Leuven, Belgium) for 3D postprocessing. The cardiovascular anatomy was first segmented using a combination of automated and manual techniques; and volume rendering was performed to depict the anatomy of interest. Results: The free-breathing self-navigated 3D radial acquisition provided significantly improved image quality and myocardial wall-blood contrast (Figure 1). Mean scores were 4.58 and 2.67 for the 3D radial and FLASH/ bSSFP sequences respectively (p = 0.003). The cardiovascular anatomy was well depicted on all virtual 3D models (Figure 2). Conclusions: 3D virtual models are frequently being created to understand complex anatomy, influence surgical planning, and provide intra-operative guidance for patients with CHD. This novel free-breathing, self-navigated whole heart 3D radial sequence provided excellent image quality as compared to existing routine MR sequences. Furthermore, the (Figure Presented) superb image quality provided using this novel sequence makes it an excellent choice for the creation of 3D models

Background: Conventional cine MRI provides data on the variation of cardiac dimensions across the cardiac cycle; cardiac function analysis primarily focuses on the difference between end-diastolic (ED) and endsystolic (ES) dimensions of the left and right ventricles (LV and RV). With cardiorespiratory-synchronized (CRS) CMR, there is an additional effective dimension of information available, related to the effect of the respiratory cycle phase on cardiac dimensions. However, there are currently no established ways to analyze this potentially useful additional physiological data. We have developed a set of tools for the functional analysis of CRS CMR, particularly for the study of the respiratory effects on LV-RV interaction, and derived some initial normative values for the results. Methods: We have developed a set of interactive CMR function analysis programs. Images from CRS CMR are organized in a two-dimensional matrix, sorted by cardiac and respiratory cycle phases. The user can interactively position an analysis line across the ventricles in a representative image; this line can then be automatically tracked across the other cardiac and respiratory phases. The intensity profile along the line is then used to automatically track the corresponding positions of the edges of the LV and RV free walls and the interventricular septum (IVS). A variety of absolute and normalized variables can be derived from these varying positions, including ED and ES dimensions, and displayed as functional images over the cardiac and respiratory cycle dimensions. CRS CMR was performed with a sparsity-based method (XD-GRASP), using continuous acquisition of radial k-space samples with golden-angle increments and retrospective cardiac and respiratory phase sorting in reconstruction. An initial set of CRS CMR data from 9 normal subjects (age 28.33 +/- 5.85) was analyzed, as well as from 3 patients (age 40 +/- 9.66, one with HCM). Results: On visual inspection of the images, it is apparent that there is a clear shift in the relative position of the IVS over the respiratory cycle, to the left in inspiration and to the right in expiration, reflecting the LV-RV interaction; this is much more prominent near ED than ES. For the normal subjects, in midlevel short-axis views, the respiratory-related absolute shift in IVS position was 1.07-3.23 mm at ED and 0.69-2.14 mm at ES; corresponding values normalized to ED dimension were 2.65-7.08 pp and 1.99-5.18 pp. The ED-ES difference for the normalized shift ranges was -1.9-4.35 pp (median 1.35, first quartile 0.68). For the HCM patient, the difference between the shift ranges was 0.79 pp. Linear regression when plotting NCD against NEDD (reflecting the Frank-Starling relationship and giving an estimate of contractility) was 0.68 +/- 0.11 in the normal subjects. Conclusions: Novel physiologic data on LV-RV interaction can be derived from CRS CMR; this seems to show consistent ranges in normal subjects, and may provide useful information on disease-related changes in cardiac function. (Figure Presented)

OBJECTIVE: This study aimed to demonstrate feasibility of free-breathing radial acquisition with respiratory motion-resolved compressed sensing reconstruction [extra-dimensional golden-angle radial sparse parallel imaging (XD-GRASP)] for multiphase dynamic gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced liver imaging, and to compare image quality to compressed sensing reconstruction with respiratory motion-averaging (GRASP) and prior conventional breath-held Cartesian-sampled data sets [BH volume interpolated breath-hold examination (VIBE)] in same patients. SUBJECTS AND METHODS: In this Health Insurance Portability and Accountability Act-compliant prospective study, 16 subjects underwent free-breathing continuous radial acquisition during Gd-EOB-DTPA injection and had prior BH-VIBE available. Acquired data were reconstructed using motion-averaging GRASP approach in which consecutive 84 spokes were grouped in each contrast-enhanced phase for a temporal resolution of approximately 14 seconds. Additionally, respiratory motion-resolved reconstruction was performed from the same k-space data by sorting each contrast-enhanced phase into multiple respiratory motion states using compressed sensing algorithm named XD-GRASP, which exploits sparsity along both the contrast-enhancement and respiratory-state dimensions.Contrast-enhanced dynamic multiphase XD-GRASP, GRASP, and BH-VIBE images were anonymized, pooled together in a random order, and presented to 2 board-certified radiologists for independent evaluation of image quality, with higher score indicating more optimal examination. RESULTS: The XD-GRASP reconstructions had significantly (all P < 0.05) higher overall image quality scores compared to GRASP for early arterial (reader 1: 4.3 +/- 0.6 vs 3.31 +/- 0.6; reader 2: 3.81 +/- 0.8 vs 3.38 +/- 0.9) and late arterial (reader 1: 4.5 +/- 0.6 vs 3.63 +/- 0.6; reader 2: 3.56 +/- 0.5 vs 2.88 +/- 0.7) phases of enhancement for both readers. The XD-GRASP also had higher overall image quality score in portal venous phase, which was significant for reader 1 (4.44 +/- 0.5 vs 3.75 +/- 0.8; P = 0.002). In addition, the XD-GRASP had higher overall image quality score compared to BH-VIBE for early (reader 1: 4.3 +/- 0.6 vs 3.88 +/- 0.6; reader 2: 3.81 +/- 0.8 vs 3.50 +/- 1.0) and late (reader 1: 4.5 +/- 0.6 vs 3.44 +/- 0.6; reader 2: 3.56 +/- 0.5 vs 2.94 +/- 0.9) arterial phases. CONCLUSION: Free-breathing motion-resolved XD-GRASP reconstructions provide diagnostic high-quality multiphase images in patients undergoing Gd-EOB-DTPA-enhanced liver examination.

PURPOSE: To demonstrate dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) of the prostate with both high spatial and temporal resolution via a combination of golden-angle radial k-space sampling, compressed sensing, and parallel-imaging reconstruction (GRASP), and to compare image quality and lesion depiction between GRASP and conventional DCE in prostate cancer patients. MATERIALS AND METHODS: Twenty prostate cancer patients underwent two 3T prostate MRI examinations on separate dates, one using standard DCE (spatial resolution 3.0 x 1.9 x 1.9 mm, temporal resolution 5.5 sec) and the other using GRASP (spatial resolution 3.0 x 1.1 x 1.1 mm, temporal resolution 2.3 sec). Two radiologists assessed measures of image quality and dominant lesion size. The experienced reader recorded differences in contrast arrival times between the dominant lesion and benign prostate. RESULTS: Compared with standard DCE, GRASP demonstrated significantly better clarity of the capsule, peripheral/transition zone boundary, urethra, and periprostatic vessels; image sharpness; and lesion conspicuity for both readers (P < 0.001-0.020). GRASP showed improved interreader correlation for lesion size (GRASP: r = 0.691-0.824, standard: r = 0.495-0.542). In 8/20 cases, only GRASP showed earlier contrast arrival in tumor than benign; in no case did only standard DCE show earlier contrast arrival in tumor. CONCLUSION: High spatiotemporal resolution prostate DCE is possible with GRASP, which has the potential to improve image quality and lesion depiction as compared with standard DCE.J. Magn. Reson. Imaging 2014. (c) 2014 Wiley Periodicals, Inc.