Search

We propose a novel neural deformable model (NDM) targeting at the reconstruction and modeling of 3D bi-ventricular shape of the heart from 2D sparse cardiac magnetic resonance (CMR) imaging data. We model the bi-ventricular shape using blended deformable superquadrics, which are parameterized by a set of geometric parameter functions and are capable of deforming globally and locally. While global geometric parameter functions and deformations capture gross shape features from visual data, local deformations, parameterized as neural diffeomorphic point flows, can be learned to recover the detailed heart shape.Different from iterative optimization methods used in conventional deformable model formulations, NDMs can be trained to learn such geometric parameter functions, global and local deformations from a shape distribution manifold. Our NDM can learn to densify a sparse cardiac point cloud with arbitrary scales and generate high-quality triangular meshes automatically. It also enables the implicit learning of dense correspondences among different heart shape instances for accurate cardiac shape registration. Furthermore, the parameters of NDM are intuitive, and can be used by a physician without sophisticated post-processing. Experimental results on a large CMR dataset demonstrate the improved performance of NDM over conventional methods., Comment: Accepted by ICCV 2023

Joint 2D cardiac segmentation and 3D volume reconstruction are fundamental to building statistical cardiac anatomy models and understanding functional mechanisms from motion patterns. However, due to the low through-plane resolution of cine MR and high inter-subject variance, accurately segmenting cardiac images and reconstructing the 3D volume are challenging. In this study, we propose an end-to-end latent-space-based framework, DeepRecon, that generates multiple clinically essential outcomes, including accurate image segmentation, synthetic high-resolution 3D image, and 3D reconstructed volume. Our method identifies the optimal latent representation of the cine image that contains accurate semantic information for cardiac structures. In particular, our model jointly generates synthetic images with accurate semantic information and segmentation of the cardiac structures using the optimal latent representation. We further explore downstream applications of 3D shape reconstruction and 4D motion pattern adaptation by the different latent-space manipulation strategies.The simultaneously generated high-resolution images present a high interpretable value to assess the cardiac shape and motion.Experimental results demonstrate the effectiveness of our approach on multiple fronts including 2D segmentation, 3D reconstruction, downstream 4D motion pattern adaption performance., Comment: MICCAI2022

Cardiac tagging magnetic resonance imaging (t-MRI) is the gold standard for regional myocardium deformation and cardiac strain estimation. However, this technique has not been widely used in clinical diagnosis, as a result of the difficulty of motion tracking encountered with t-MRI images. In this paper, we propose a novel deep learning-based fully unsupervised method for in vivo motion tracking on t-MRI images. We first estimate the motion field (INF) between any two consecutive t-MRI frames by a bi-directional generative diffeomorphic registration neural network. Using this result, we then estimate the Lagrangian motion field between the reference frame and any other frame through a differentiable composition layer. By utilizing temporal information to perform reasonable estimations on spatio-temporal motion fields, this novel method provides a useful solution for motion tracking and image registration in dynamic medical imaging. Our method has been validated on a representative clinical t-MRI dataset; the experimental results show that our method is superior to conventional motion tracking methods in terms of landmark tracking accuracy and inference efficiency., Comment: CVPR 2021 Oral

To present and validate a new method for 4D flow quantification of vortex-ring mixing during early, rapid filling of the left ventricle (LV) as a potential index of diastolic dysfunction and heart failure.

Background: Respiratory gating is often used in 4D-flow acquisition to reduce motion artifacts. However, gating increases scan time. The aim of this study was to investigate if respiratory gating can be excluded from 4D flow acquisitions without affecting quantitative intracardiac parameters. Methods: Eight volunteers underwent CMR at 1.5 T with a 5-channel coil (5ch). Imaging included 2D flow measurements and whole-heart 4D flow with and without respiratory gating (Resp(+), Resp(-)). Stroke volume (SV), particle-trace volumes, kinetic energy, and vortex-ring volume were obtained from 4D flow-data. These parameters were compared between 5ch Resp(+) and 5ch Resp(-). In addition, 20 patients with heart failure were scanned using a 32-channel coil (32ch), and particle-trace volumes were compared to planimetric SV. Paired comparisons were performed using Wilcoxon's test and correlation analysis using Pearson r. Agreement was assessed as bias ± SD. Results: Stroke volume from 4D flow was lower compared to 2D flow both with and without respiratory gating (5ch Resp(+) 88 ± 18 vs 97 ± 24.0, p = 0.001; 5ch Resp(-) 86 ± 16 vs 97.1 ± 22.7, p < 0.01). There was a good correlation between Resp(+) and Resp(-) for particle-trace derived volumes (R2 = 0.82, 0.2 ± 9.4 ml), mean kinetic energy (R2 = 0.86, 0.07 ± 0.21 mJ), peak kinetic energy (R2 = 0.88, 0.14 ± 0.77 mJ), and vortex-ring volume (R2 = 0.70, -2.5 ± 9.4 ml). Furthermore, good correlation was found between particle-trace volume and planimetric SV in patients for 32ch Resp(-) (R2 = 0.62, -4.2 ± 17.6 ml) and in healthy volunteers for 5ch Resp(+) (R2 = 0.89, -11 ± 7 ml), and 5ch Resp(-) (R2 = 0.93, -7.5 ± 5.4 ml), Average scan duration for Resp(-) was shorter compared to Resp(+) (27 ± 9 min vs 61 ± 19 min, p < 0.05). Conclusions: Whole-heart 4D flow can be acquired with preserved quantitative results without respiratory gating, facilitating clinical use.

Background: Pulmonary involvement, manifested as pulmonary arterial hypertension or pulmonary fibrosis, is the most common cause of death in systemic sclerosis (SSc). We aimed to explore the feasibility of detecting early pulmonary involvement in SSc using recently developed non-invasive quantitative measures of pulmonary physiology using cardiovascular magnetic resonance (CMR). Methods: Twenty-seven SSc patients (9 men, 57 +/- 13 years) and 10 healthy controls (3 men, 54 +/- 9 years) underwent CMR to determine the pulmonary blood volume (PBV) and the PBV variation (PBVV) throughout the cardiac cycle. Patients underwent Doppler echocardiography, high-resolution computed tomography (HRCT), and pulmonary function testing by spirometry. Comparisons were performed using the unpaired t-test and linear regression analysis was performed with Pearson's correlation coefficient (r). Results: Compared to healthy controls, the PBV indexed to lung volume (PBVI) was lower in patients (16 +/- 4 vs 20 +/- 5%, p < 0.05). There was no difference in PBV (466 +/- 87 vs 471 +/- 122 mL, p = 0.91) or PBVV/stroke volume (45 +/- 10 vs 40 +/- 6%, p = 0.09). There were no significant correlations between PBVI and pulmonary artery pressure estimated by Doppler (p = 0.08) the lung's diffusion capacity for carbon monoxide (DLCO) (p = 0.09), vital capacity (p = 0.45), or pulmonary fibrosis by HRCT (p = 0.74). Conclusions: This study is the first to measure the PBV in humans using CMR. Compared to healthy controls, newly diagnosed SSc patients have a reduced amount of blood in the pulmonary vasculature (PBVI) but unchanged pulmonary vascular distensibility (PBVV/stroke volume). PBVI is unrelated to DLCO, pulmonary artery pressure, vital capacity, and the presence of pulmonary fibrosis. PBVI may be a novel parameter reflecting vascular lung involvement in early-stage SSc, and these findings may be consistent with pathophysiological changes of the pulmonary vasculature.

Recent studies suggest that vortex ring formation during left ventricular (LV) rapid filling is an optimized mechanism for blood transport, and that the volume of the vortex ring is an important measure. However, due to lack of quantitative methods, the volume of the vortex ring has not previously been studied. Lagrangian Coherent Structures (LCS) is a new flow analysis method, which enables in vivo quantification of vortex ring volume. Therefore, we aimed to investigate if vortex ring volume in the human LV can be reliably quantified using LCS and magnetic resonance velocity mapping (4D PC-MR). Flow velocities were measured using 4D PC-MR in 9 healthy volunteers and 4 patients with dilated ischemic cardiomyopathy. LV LCS were computed from flow velocities and manually delineated in all subjects. Vortex volume in the healthy volunteers was 51 ± 6% of the LV volume, and 21 ± 5% in the patients. Interobserver variability was -1 ± 13% and interstudy variability was -2 ± 12%. Compared to idealized flow experiments, the vortex rings showed additional complexity and asymmetry, related to endocardial trabeculation and papillary muscles. In conclusion, LCS and 4D PC-MR enables measurement of vortex ring volume during rapid filling of the LV.

Background: Three-dimensional time-resolved (4D) phase-contrast (PC) CMR can visualize and quantify cardiovascular flow but is hampered by long acquisition times. Acceleration with SENSE or k-t BLAST are two possibilities but results on validation are lacking, especially at 3 T. The aim of this study was therefore to validate quantitative in vivo cardiac 4D-acquisitions accelerated with parallel imaging and k-t BLAST at 1.5 T and 3 T with 2D-flow as the reference and to investigate if field strengths and type of acceleration have major effects on intracardiac flow visualization. Methods: The local ethical committee approved the study. 13 healthy volunteers were scanned at both 1.5 T and 3 T in random order with 2D-flow of the aorta and main pulmonary artery and two 4D-flow sequences of the heart accelerated with SENSE and k-t BLAST respectively. 2D-image planes were reconstructed at the aortic and pulmonary outflow. Flow curves were calculated and peak flows and stroke volumes (SV) compared to the results from 2D-flow acquisitions. Intra-cardiac flow was visualized using particle tracing and image quality based on the flow patterns of the particles was graded using a four-point scale. Results: Good accuracy of SV quantification was found using 3 T 4D-SENSE (r(2) = 0.86, -0.7 +/- 7.6%) and although a larger bias was found on 1.5 T (r(2) = 0.71, -3.6 +/- 14.8%), the difference was not significant (p = 0.46). Accuracy of 4D k-t BLAST for SV was lower (p < 0.01) on 1.5 T (r(2) = 0.65, -15.6 +/- 13.7%) compared to 3 T (r(2) = 0.64, -4.6 +/- 10.0%). Peak flow was lower with 4D-SENSE at both 3 T and 1.5 T compared to 2D-flow (p < 0.01) and even lower with 4D k-t BLAST at both scanners (p < 0.01). Intracardiac flow visualization did not differ between 1.5 T and 3 T (p = 0.09) or between 4D-SENSE or 4D k-t BLAST (p = 0.85). Conclusions: The present study showed that quantitative 4D flow accelerated with SENSE has good accuracy at 3 T and compares favourably to 1.5 T. 4D flow

Exacerbation of ST elevation associated with reperfusion has been reported in patients with myocardial infarction. However, the cause of the "reperfusion peak" and relation of its magnitude to the size of myocardial damage has not been explored. The aim of our study was to assess the correlation between the ST-dynamics during reperfusion, the myocardium at risk (MaR), and the infarct size (IS).

Background: Polymorphonuclear neutrophils, stimulated by the activated complement factor C5a, have been implicated in cardiac ischemia/reperfusion injury. ADC-1004 is a competitive C5a receptor antagonist that has been shown to inhibit complement related neutrophil activation. ADC-1004 shields the neutrophils from C5a activation before they enter the reperfused area, which could be a mechanistic advantage compared to previous C5a directed reperfusion therapies. We investigated if treatment with ADC-1004, according to a clinically applicable protocol, would reduce infarct size and microvascular obstruction in a large animal myocardial infarct model. Methods: In anesthetized pigs (42-53 kg), a percutaneous coronary intervention balloon was inflated in the left anterior descending artery for 40 minutes, followed by 4 hours of reperfusion. Twenty minutes after balloon inflation the pigs were randomized to an intravenous bolus administration of ADC-1004 (175 mg, n = 8) or saline (9 mg/ml, n = 8). Area at risk (AAR) was evaluated by ex vivo SPECT. Infarct size and microvascular obstruction were evaluated by ex vivo MRI. The observers were blinded to the treatment at randomization and analysis. Results: ADC-1004 treatment reduced infarct size by 21% (ADC-1004: 58.3 +/- 3.4 vs control: 74.1 +/- 2.9% AAR, p = 0.007). Microvascular obstruction was similar between the groups (ADC-1004: 2.2 +/- 1.2 vs control: 5.3 +/- 2.5% AAR, p = 0.23). The mean plasma concentration of ADC-1004 was 83 +/- 8 nM at sacrifice. There were no significant differences between the groups with respect to heart rate, mean arterial pressure, cardiac output and blood-gas data. Conclusions: ADC-1004 treatment reduces myocardial ischemia-reperfusion injury and represents a novel treatment strategy of myocardial infarct with potential clinical applicability.