WarpPINN: Cine-MR image registration with physics-informed neural networks.

The diagnosis of heart failure usually includes a global functional assessment, such as ejection fraction measured by magnetic resonance imaging. However, these metrics have low discriminate power to distinguish different cardiomyopathies, which may not affect the global function of the heart. Quantifying local deformations in the form of cardiac strain can provide helpful information, but it remains a challenge. In this work, we introduce WarpPINN, a physics-informed neural network to perform image registration to obtain local metrics of heart deformation. We apply this method to cine magnetic resonance images to estimate the motion during the cardiac cycle. We inform our neural network of the near-incompressibility of cardiac tissue by penalizing the Jacobian of the deformation field. The loss function has two components: an intensity-based similarity term between the reference and the warped template images, and a regularizer that represents the hyperelastic behavior of the tissue. The architecture of the neural network allows us to easily compute the strain via automatic differentiation to assess cardiac activity. We use Fourier feature mappings to overcome the spectral bias of neural networks, allowing us to capture discontinuities in the strain field. The algorithm is tested on synthetic examples and on a cine SSFP MRI benchmark of 15 healthy volunteers, where it is trained to learn the deformation mapping of each case. We outperform current methodologies in landmark tracking and provide physiological strain estimations in the radial and circumferential directions. WarpPINN provides precise measurements of local cardiac deformations that can be used for a better diagnosis of heart failure and can be used for general image registration tasks. Source code is available at https://github.com/fsahli/WarpPINN.

Fully Three-Dimensional Hemodynamic Characterization of Altered Blood Flow in Bicuspid Aortic Valve Patients With Respect to Aortic Dilatation: A Finite Element Approach.

Background and Purpose: Prognostic models based on cardiovascular hemodynamic parameters may bring new information for an early assessment of patients with bicuspid aortic valve (BAV), playing a key role in reducing the long-term risk of cardiovascular events. This work quantifies several three-dimensional hemodynamic parameters in different patients with BAV and ranks their relationships with aortic diameter. Materials and Methods: Using 4D-flow CMR data of 74 patients with BAV (49 right-left and 25 right-non-coronary) and 48 healthy volunteers, aortic 3D maps of seventeen 17 different hemodynamic parameters were quantified along the thoracic aorta. Patients with BAV were divided into two morphotype categories, BAV-Non-AAoD (where we include 18 non-dilated patients and 7 root-dilated patients) and BAV-AAoD (where we include the 49 patients with dilatation of the ascending aorta). Differences between volunteers and patients were evaluated using MANOVA with Pillai's trace statistic, Mann-Whitney U test, ROC curves, and minimum redundancy maximum relevance algorithm. Spearman's correlation was used to correlate the dilation with each hemodynamic parameter. Results: The flow eccentricity, backward velocity, velocity angle, regurgitation fraction, circumferential wall shear stress, axial vorticity, and axial circulation allowed to discriminate between volunteers and patients with BAV, even in the absence of dilation. In patients with BAV, the diameter presented a strong correlation (> |+/-0.7|) with the forward velocity and velocity angle, and a good correlation (> |+/-0.5|) with regurgitation fraction, wall shear stress, wall shear stress axial, and vorticity, also for morphotypes and phenotypes, some of them are correlated with the diameter. The velocity angle proved to be an excellent biomarker in the differentiation between volunteers and patients with BAV, BAV morphotypes, and BAV phenotypes, with an area under the curve bigger than 0.90, and higher predictor important scores. Conclusions: Through the application of a novel 3D quantification method, hemodynamic parameters related to flow direction, such as flow eccentricity, velocity angle, and regurgitation fraction, presented the best relationships with a local diameter and effectively differentiated patients with BAV from healthy volunteers.

Assessment of 4D flow MRI's quality by verifying its Navier-Stokes compatibility.

4D Flow Magnetic Resonance Imaging (MRI) is the state-of-the-art technique to comprehensively measure the complex spatio-temporal and multidirectional patterns of blood flow. However, it is subject to artifacts such as noise and aliasing, which due to the 3D and dynamic structure is difficult to detect in clinical practice. In this work, a new mathematical and computational model to determine the quality of 4D Flow MRI is presented. The model is derived by assuming the true velocity satisfies the incompressible Navier-Stokes equations and that can be decomposed by the measurements u→meas plus an extra field w→ . Therefore, a non-linear problem with w→ as unknown arises, which serves as a measure of data quality. A stabilized finite element formulation tailored to this problem is proposed and analyzed. Then, extensive numerical examples-using synthetic 4D Flow MRI data as well as real measurements on experimental phantom and subjects-illustrate the ability to use w→ for assessing the quality of 4D Flow MRI measurements over space and time.

Three-dimensional quantification of circulation using finite-element methods in four-dimensional flow MR data of the thoracic aorta.

PURPOSE: Three-dimensional (3D) quantification of circulation using a Finite Elements methodology. METHODS: We validate our 3D method using an in-silico arch model, for different mesh resolutions, image resolution and noise levels, and we compared this with a currently used 2D method. Finally, we evaluated the application of our methodology in 4D Flow MRI data of ascending aorta of six healthy volunteers, and six bicuspid aortic valve (BAV) patients, three with right and three with left handed flow, at peak systole. The in-vivo data was compared using a Mann-Whitney U-test between volunteers and patients (right and left handed flow). RESULTS: The robustness of our method throughout different image resolutions and noise levels showed subestimation of circulation less than 45 cm2 /s in comparison with the 55cm2 /s generated by the current 2D method. The circulation (mean ± SD) of the healthy volunteer group was 13.83 ± 28.78 cm2 /s, in BAV patients with right-handed flow 724.37 ± 317.53 cm2 /s, and BAV patients with left-handed flow -480.99 ± 387.29 cm2 /s. There were significant differences between healthy volunteers and BAV patients groups (P-value < .01), and also between BAV patients with a right-handed or left-handed helical flow and healthy volunteers (P-value < .01). CONCLUSION: We propose a novel 3D formulation to estimate the circulation in the thoracic aorta, which can be used to assess the differences between normal and diseased hemodynamic from 4D-Flow MRI data. This method also can correctly differentiate between the visually seen right- and left-handed helical flow, which suggests that this approach may have high clinical sensitivity, but requires confirmation in longitudinal studies with a large cohort.

An off-resonance insensitive orthogonal CSPAMM sequence (ORI-O-CSPAMM) for the acquisition of CSPAMM and MICSR grids in half scan time.

PURPOSE: To develop an Off-Resonance Insensitive Orthogonal CSPAMM sequence (ORI-O-CSPAMM) for the acquisition of CSPAMM and MICSR grids in half of the acquisition time. METHODS: Phantom and mid-level left ventricle short-axis tagged images were acquired using CSPAMM, ORI-CSPAMM, O-CSPAMM, and the proposed ORI-O-CPAMM sequences to interrogate and compare its behavior under off-resonance effects produced by vegetable oil and subcutaneous and epicardial fat. The images were compared in terms of signal and the capacity to obtain complex difference and MICSR images. RESULTS: Like ORI-CSPAMM, the proposed ORI-O-CSPAMM sequence removed almost completely the off-resonance artifacts produced during the tagging preparation. Tagging grids without DC components were obtained with ORI-O-CSPAMM using complex difference and MICSR from only two complementary images, which reduced the scan time to a half compared to CSPAMM and ORI-CSPAMM. The removal of off-resonance effects and the capacity to obtain MICSR images are advantages of ORI-O-CSPAMM over the O-CSPAMM sequence. CONCLUSION: We developed a novel and fast tagging sequence designed to remove off-resonance effects during the tagging preparation, and to obtain complex difference and MICSR grids in half of the scan time compared to CSPAMM and ORI-CSPAMM sequences, which could allow its application to clinical protocols.

A comprehensive comparison between shortest-path HARP refinement, SinMod, and DENSEanalysis processing tools applied to CSPAMM and DENSE images.

We addressed comprehensively the performance of Shortest-Path HARP Refinement (SP-HR), SinMod, and DENSEanalysis using 2D slices of synthetic CSPAMM and DENSE images with realistic contrasts obtained from 3D phantoms. The three motion estimation techniques were interrogated under ideal and no-ideal conditions (with MR induced artifacts, noise, and through-plane motion), considering several resolutions and noise levels. Under noisy conditions, and for isotropic pixel sizes of 1.5 mm and 3.0 mm in CSPAMM and DENSE images respectively, the nRMSE obtained for the circumferential and radial strain components were 10.7 ±â€¯10.8% and 25.5 ±â€¯14.8% using SP-HR, 11.9 ±â€¯2.5% and 29.3 ±â€¯6.5% using SinMod, and 6.4 ±â€¯2.0% and 18.2 ±â€¯4.6% using DENSEanalysis. Overall, the results showed that SP-HR tends to fail for large tissue motions, whereas SinMod and DENSEanalysis gave accurate displacement and strain field estimations, being the last which performed the best.

HARP-I: A Harmonic Phase Interpolation Method for the Estimation of Motion From Tagged MR Images.

We proposed a novel method called HARP-I, which enhances the estimation of motion from tagged Magnetic Resonance Imaging (MRI). The harmonic phase of the images is unwrapped and treated as noisy measurements of reference coordinates on a deformed domain, obtaining motion with high accuracy using Radial Basis Functions interpolations. Results were compared against Shortest Path HARP Refinement (SP-HR) and Sine-wave Modeling (SinMod), two harmonic image-based techniques for motion estimation from tagged images. HARP-I showed a favorable similarity with both methods under noise-free conditions, whereas a more robust performance was found in the presence of noise. Cardiac strain was better estimated using HARP-I at almost any motion level, giving strain maps with less artifacts. Additionally, HARP-I showed better temporal consistency as a new method was developed to fix phase jumps between frames. In conclusion, HARP-I showed to be a robust method for the estimation of motion and strain under ideal and non-ideal conditions.