Three-dimensional Phase-contrast Magnetic Resonance Imaging Research Articles

A comparison of machine learning methods for recovering noisy and missing 4D flow MRI data.

Experimental blood flow measurement techniques are invaluable for a better understanding of cardiovascular disease formation, progression, and treatment. One of the emerging methods is time-resolved three-dimensional phase-contrast magnetic resonance imaging (4D flow MRI), which enables noninvasive time-dependent velocity measurements within large vessels. However, several limitations hinder the usability of 4D flow MRI and other experimental methods for quantitative hemodynamics analysis. These mainly include measurement noise, corrupt or missing data, low spatiotemporal resolution, and other artifacts. Traditional filtering is routinely applied for denoising experimental blood flow data without any detailed discussion on why it is preferred over other methods. In this study, filtering is compared to different singular value decomposition (SVD)-based machine learning and autoencoder-type deep learning methods for denoising and filling in missing data (imputation). An artificially corrupted and voxelized computational fluid dynamics (CFD) simulation as well as in vitro 4D flow MRI data are used to test the methods. SVD-based algorithms achieve excellent results for the idealized case but severely struggle when applied to in vitro data. The autoencoders are shown to be versatile and applicable to all investigated cases. For denoising, the in vitro 4D flow MRI data, the denoising autoencoder (DAE), and the Noise2Noise (N2N) autoencoder produced better reconstructions than filtering both qualitatively and quantitatively. Deep learning methods such as N2N can result in noise-free velocity fields even though they did not use clean data during training. This work presents one of the first comprehensive assessments and comparisons of various classical and modern machine-learning methods for enhancing corrupt cardiovascular flow data in diseased arteries for both synthetic and experimental test cases.

Impact of training data composition on the generalizability of convolutional neural network aortic cross-section segmentation in four-dimensional magnetic resonance flow imaging

BackgroundTime-resolved, three-dimensional phase-contrast magnetic resonance imaging (4D flow MRI) plays an important role in assessing cardiovascular diseases. However, the manual or semi-automatic segmentation of aortic vessel boundaries in 4D flow data introduces variability and limits reproducibility of aortic hemodynamics visualization and quantitative flow-related parameter computation. This paper explores the potential of deep learning to improve 4D flow MRI segmentation by developing models for automatic segmentation and analyzes the impact of the training data on the generalization of the model across different sites, scanner vendors, sequences, and pathologies. MethodsThe study population consists of 260 4D flow MRI datasets, including subjects without known aortic pathology, healthy volunteers, and patients with bicuspid aortic valve (BAV) examined at different hospitals. The dataset was split to train segmentation models on subsets with different representations of characteristics such as pathology, gender, age, scanner model, vendor, and field strength. An enhanced 3D U-net convolutional neural network (CNN) architecture with residual units was trained for 2D+t aortic cross-sectional segmentation. The model performance was evaluated using Dice score, Hausdorff distance, and average symmetric surface distance on test data, datasets with characteristics not represented in the training set (model-specific), and an overall evaluation set. Standard diagnostic flow parameters were computed and compared with manual segmentation results using Bland-Altman analysis and interclass correlation. ResultsThe representation of technical factors such as scanner vendor and field strength in the training dataset had the strongest influence on the overall segmentation performance. Age had a greater impact than gender. Models solely trained on BAV patients’ datasets performed well on datasets of healthy subjects but not vice versa. ConclusionThis study highlights the importance of considering a heterogeneous dataset for the training of widely applicable automatic CNN segmentations in 4D flow MRI, with a particular focus on the inclusion of different pathologies and technical aspects of data acquisition.

Three-dimensional phase-contrast magnetic resonance imaging for the detention of a small communicating defect in a patient with a spinal extradural arachnoid cyst: illustrative case.

Spinal extradural arachnoid cysts are thought to be pouches that communicate with the intraspinal subarachnoid space through a dural defect. The treatment for these cysts is resection of the cyst wall followed by obliteration of the communicating defect, which is often elusive. The authors report the case of a 22-year-old man with an extradural arachnoid cyst with claudication and progressive motor weakness. Regular magnetic resonance imaging (MRI) and computed tomography did not reveal the location of the defect in the cyst. However, three-dimensional (3D) phase-contrast MRI clearly indicated the location of the defect and the flow of cerebrospinal fluid into the cyst. These findings allowed the authors to perform the least invasive surgery; the patient recovered motor function and could walk more smoothly. 3D phase-contrast MRI can reveal a subtle dural defect in patients with spinal extradural arachnoid cysts.

Systolic flow displacement using 3D magnetic resonance imaging in an experimental model of ascending aorta aneurysm: impact of rheological factors.

The impact of systolic flow displacement on the development and progression of ascending aorta dilatation in aortic valve disease is a matter of controversy. Our objective was to study the association between rheological stimuli and development of aortic dilatation in a large animal model of supravalvular aortic stenosis and eccentric flow. Twenty-four pigs weighing 10-14 kg were randomly allocated (ratio 2:1) to either restrictive ascending aortic banding or sham operation. Aortic diameter and systolic flow displacement were assessed by three-dimensional phase-contrast magnetic resonance imaging at 6 and 18 weeks after surgery. Twenty pigs (n = 14, banded vs n = 6, sham) completed full imaging protocol and were included in the analysis. After the last follow-up, a subset of 14 animals was sacrificed for histological analysis. All banded animals developed significant progressive aortic dilatation both at 6 and 18 weeks, compared with sham-operated pigs: 34.3 ± 4.8 vs 21.4 ± 2.7 mm at 6 weeks (P < 0.001); and 50.0 ± 8.4 vs 38.0 ± 8.3 mm at 18 weeks (P = 0.002). The peak gradient at 6 weeks showed a trend to positively correlate with aortic diameter at 18 weeks (R = 0.50, P = 0.06), whereas the systolic flow displacement at 6 weeks correlated better with aortic diameter at 18 weeks (R = 0.59, P = 0.02). The aortic wall thickness was significantly decreased in the anterior aortic section in banded, compared with sham-operated, pigs (1.5 ± 0.4 vs 2.0 ± 0.1 mm, respectively; P = 0.03). In addition, banded pigs showed a higher degree of cystic medial necrosis and elastin fibre fragmentation, compared with sham-operated animals. In this preclinical model of supravalvular aortic stenosis and eccentric flow, we found that systolic flow displacement at earlier stages is positively correlated with the degree of aortic dilatation during follow-up as assessed by three-dimensional phase-contrast magnetic resonance imaging. If our findings are confirmed in further studies, this imaging parameter might be useful to identify those subjects with aortic valve disease who are at risk of developing aortic dilatation at a later stage.

Noninvasive pressure difference mapping derived from 4D flow MRI in patients with unrepaired and repaired aortic coarctation.

To develop a method for computing and visualizing pressure differences derived from time-resolved velocity-encoded three-dimensional phase-contrast magnetic resonance imaging (4D flow MRI) and to compare pressure difference maps of patients with unrepaired and repaired aortic coarctation to young healthy volunteers. 4D flow MRI data of four patients with aortic coarctation either before or after repair (mean age 17 years, age range 3-28, one female, three males) and four young healthy volunteers without history of cardiovascular disease (mean age 24 years, age range 20-27, one female, three males) was acquired using a 1.5-T clinical MR scanner. Image analysis was performed with in-house developed image processing software. Relative pressures were computed based on the Navier-Stokes equation. A standardized method for intuitive visualization of pressure difference maps was developed and successfully applied to all included patients and volunteers. Young healthy volunteers exhibited smooth and regular distribution of relative pressures in the thoracic aorta at mid systole with very similar distribution in all analyzed volunteers. Patients demonstrated disturbed pressures compared to volunteers. Changes included a pressure drop at the aortic isthmus in all patients, increased relative pressures in the aortic arch in patients with residual narrowing after repair, and increased relative pressures in the descending aorta in a patient after patch aortoplasty. Pressure difference maps derived from 4D flow MRI can depict alterations of spatial pressure distribution in patients with repaired and unrepaired aortic coarctation. The technique might allow identifying pathophysiological conditions underlying complications after aortic coarctation repair.

Neoaortic Root Aneurysm After Arterial Switch Operation With Lecompte Maneuver

Fig 2. Ateries had undergone an arterial switch operation with the Lecompte maneuver and mechanical valve replacement for neoaortic valve insufficiency at 1 and 8 years of age, respectively. Recently, a neoaortic aneurysm was diagnosed by magnetic resonance imaging. Right (blue) and left (red) heart structures are shown in Fig 1A, and isolated left heart structures can be seen in Fig 1B. Three-dimensional phase-contrast magnetic resonance imaging data was used to visualize flow patterns. Vortical flow in the aneurysm was fast (red path lines) during systole and persisted as slow (blue) flow into late diastole. Blood flow was accelerated in the narrowed distal main pulmonary artery (Video 1). Because the aneurysm showed progressive growth, reaching a diameter of 7 cm, replacement of the ascending aorta was indicated. The main pulmonary artery (Fig 2, MPA) was transected and the aneurysm was incised (Fig 2, asterisk). Implantation of a mechanical valve conduit (Fig 2, MVC) became necessary because of pannus formation at the valve prosthesis (Fig 2, AC 1⁄4 aortic cannula; LVOT 1⁄4 left ventricular outflow tract; VC 1⁄4 venous cannula). Based on intraoperative observations, we assume that the aneurysm evolved because of

Imaging of the Thoracic Aorta with Time-Resolved Three-Dimensional Phase-Contrast MRI: A Review

Time-resolved three-dimensional (3D) phase-contrast (PC) magnetic resonance imaging (MRI), or four-dimensional (4D) flow, is able to provide robust 3D images with three-directional velocities. This review discusses the technique and application of 4D flow in the imaging of thoracic aortic pathologies. It has been instrumental in describing normal flow patterns throughout the cardiac cycle in the ascending and descending aorta and has shown the variety of flow patterns that exist in ascending aortic aneurysms.

Cerebral Ischemia during Carotid Artery Cross-Clamping: Predictive Value of Phase-Contrast Magnetic Resonance Imaging

The goal of this prospective study was to determine the utility of preoperative cerebral magnetic resonance imaging (MRI) in predicting cerebral ischemia during carotid artery cross-clamping for endarterectomy. Between January 2000 and December 2003, a total of 121 patients (95 men, 26 women) underwent three-dimensional phase-contrast MRI to assess collateral function prior to carotid endarterectomy. During regional anesthesia, patients were monitored to detect ischemic events and their timing in relation to cross-clamping and to determine mean intraoperative arterial pressure. These findings were then correlated with the collateral variations observed in the circle of Willis on preoperative MRI. Patients were classified into three groups according to neurological tolerance: normal tolerance (n = 106), immediate severe deficit (n = 9), and late deficit associated with arterial hypotension (n = 6). In the second group, a significant correlation was found between the absence of collateral circulation and neurological deficit (p < .0001). These results indicated that three-dimensional phase-contrast MRI is useful for predicting cerebral ischemia during carotid cross-clamping and selecting indications for shunting. Absence of visible collaterals of the circle of Willis on MRI is significantly predictive of early ischemia and an indication for systematic shunt placement.

Alterations in pulmonary artery flow patterns and shear stress determined with three-dimensional phase-contrast magnetic resonance imaging in fontan patients

Objective: This study compares in vivo pulmonary blood flow patterns and shear stresses in patients with either the direct atrium–pulmonary artery connection or the bicaval tunnel connection of the Fontan procedure to those in normal volunteers. Comparisons were made with the use of three-dimensional phase contrast magnetic resonance imaging. Methods: Three-dimensional velocities, flows, and pulmonary artery cross-sectional areas were measured in both pulmonary arteries of each subject. Axial, circumferential, and radial shear stresses were calculated with the use of velocities and estimates of viscosity. Results: The axial velocities were not significantly different between subject groups. However, the flows and cross-sectional areas were higher in the normal group than in the two patient groups in both pulmonary arteries. The group with the bicaval connection had circular swirling in the cross section of both pulmonary arteries, causing higher shear stresses than in the controls. The disorder caused by the connection of the atrium to the pulmonary artery caused an increase in some shear stresses over the controls, but not higher than those found in the group having a bicaval tunnel. Conclusions: We found that pulmonary flow was equally reduced compared with normal flow in both patient groups. This reduction in flow can be attributed in part to the reduced size of the pulmonary arteries in both patient groups without change in axial velocity. We also found higher shear stress acting on the wall of the vessels in the patients having a bicaval tunnel, which may alter endothelial function and affect the longevity of the repair. (J Thorac Cardiovasc Surg 1998;116:294-304)

Visualizing three-dimensional flow with simulated streamlines and three-dimensional phase-contrast MR imaging.

Three-dimensional (3D) velocity maps acquired with 3D phase-contrast magnetic resonance (MR) imaging contain information regarding complex motions that occur during imaging. A technique called simulated streamlines, which facilitates the display and comprehension of these velocity data, is presented. Single or multiple seed points may be identified within blood vessels of interest and tracked through the velocity field. The resulting trajectories are combined with a 3D MR angiogram and displayed with 3D volume visualization software. Mathematical analysis highlights potential applications and pitfalls of the technique, which was implemented both in phantoms and in vivo with excellent results. For example, single streamlines reveal helical flow patterns in aneurysms, and multiple streamlines seeded in the common carotid artery reveal branch filling-time relationships and slow filling of the carotid bulb. The technique is helpful in understanding these complex flow patterns.