Learning to Beat: Phenotype-Guided Latent Flow with Regional Motion Priors for Biventricular Motion Synthesis

Xuan Yang; Xiaohan Yuan; Hao Li; Lingyu Chen; Yanan Liu; Qingya Li; Lei Li

cs.CVAug 20, 2026

Learning to Beat: Phenotype-Guided Latent Flow with Regional Motion Priors for Biventricular Motion Synthesis

Authors: Xuan Yang, Xiaohan Yuan, Hao Li, Lingyu Chen, Yanan Liu, Qingya Li, Lei Li

Organizations: Department of Biomedical Engineering, National University of Singapore, Singapore · School of Automation, Southeast University, Nanjing, China · School of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China

Abstract

Full-cycle biventricular geometry is essential for characterizing cardiac function. However, dense and temporally consistent 3D+t biventricular meshes are not routinely available, whereas end-diastolic (ED) anatomy can often be obtained reliably. We therefore investigate full-cycle biventricular motion synthesis from a single ED mesh. This task is challenging because cardiac deformation is spatially heterogeneous and phenotype dependent, while conventional global generative models often obscure localized motion patterns. In this study, we propose a region-specific and phenotype-adaptive framework that integrates motion-informed functional parcellation with conditional latent flow. A functional partition learned from reconstructed motion organizes the ventricular surface into regions with coherent dynamics and enables topology-aware regional feature exchange. A phenotype-conditioned rectified-flow model subsequently maps the ED anatomy to full-cycle motion latents through fine-grained conditioning and prototype-routed motion adapters. An optional control branch further incorporates available motion descriptors for controllable synthesis. Experiments on ACDC, M&Ms, and M&Ms-2 demonstrate consistent improvements in geometric accuracy and functional fidelity. Under ED-only synthesis, our method achieves biventricular ASSD, HD95, and vRMSE of $1.49\pm0.34$ ~mm, $3.77\pm1.06$ ~mm, and $3.31\pm1.03$ ~mm, respectively, outperforming all competing methods. Complementary functional and robustness evaluations further demonstrate that the synthesized sequences preserve physiologically plausible ventricular dynamics and generalize across cohorts and disease phenotypes. The code will be released publicly upon acceptance of the manuscript for publication.

Figures & tables

Figure 1: Illustration of the task and its formulation. ED: end-diastolic.

Figure 2: Illustration of the proposed regional motion-informed and phenotype-guided biventricular motion synthesis framework. (a) Motion-driven functional regionalization. (b) Region-topology aware motion variational autoencoder (VAE). (c) Phenotype-guided latent flow for anatomy-to-motion generation. DyMeshVAE: dynamic mesh VAE; PE: positional encoding.

Figure 3: (a) Region-specific injection module. (b) Phenotype-conditioned rectified-flow block with optional functional motion control. FPS: farthest point sampling; MLP: multilayer perceptron.

Dataset	NOR	DCM	HCM	RVA	Total
ACDC	30	30	30	30	120
M&Ms	89	97	85	15	286
M&Ms-2	75	60	60	65	260
Total	194	187	175	110	666

Table 1: Composition of the study cohort across datasets and diagnostic groups. NOR: normal; DCM: dilated cardiomyopathy; HCM: hypertrophic cardiomyopathy; RVA: right-ventricular abnormality.

Method	ASSD (mm) $\downarrow$			HD95 (mm) $\downarrow$			vRMSE (mm) $\downarrow$
Method	LV	RV	BiV	LV	RV	BiV	LV	RV	BiV
CVAE ( 42 )	$2.16\pm 0.66$	$2.08\pm 0.63$	$2.15\pm 0.54$	$4.34\pm 1.30$	$4.78\pm 1.77$	$4.71\pm 1.54$	$3.86\pm 1.34$	$4.25\pm 1.54$	$4.11\pm 1.25$
ACTOR ( 35 )	$2.10\pm 0.64$	$2.00\pm 0.80$	$2.11\pm 0.60$	$4.28\pm 1.41$	$4.85\pm 2.33$	$4.93\pm 1.98$	$3.71\pm 1.39$	$4.14\pm 1.83$	$4.01\pm 1.52$
Action2Motion ( 15 )	$2.28\pm 0.71$	$2.18\pm 0.78$	$2.22\pm 0.64$	$4.81\pm 1.65$	$5.07\pm 2.18$	$5.16\pm 1.98$	$4.20\pm 1.53$	$4.51\pm 1.85$	$4.21\pm 1.52$
CHeart ( 38 )	$2.21\pm 0.65$	$2.04\pm 0.72$	$2.09\pm 0.53$	$4.48\pm 1.43$	$4.73\pm 1.98$	$4.97\pm 1.52$	$3.80\pm 1.37$	$4.10\pm 1.70$	$3.98\pm 1.22$
MeshHeart ( 37 )	$2.09\pm 0.65$	$2.01\pm 0.63$	$2.08\pm 0.53$	$4.23\pm 1.30$	$4.64\pm 1.77$	$4.68\pm 1.54$	$3.78\pm 1.34$	$4.16\pm 1.54$	$4.13\pm 1.25$

Table 2: Quantitative comparison under the ED-conditioned full-cycle biventricular motion synthesis. ASSD: average symmetric surface distance; HD95: 95th percentile Hausdorff distance; vRMSE: vertex-wise root mean square error; BiV: biventricular; LV: left ventricular; RV: right ventricular. Values are reported as the mean

\pm

standard deviation across test subjects. For each subject, each metric was first averaged across 20 independently generated motion sequences.

Figure 4: Qualitative comparison of ED-conditioned biventricular motion synthesis. Surface-error maps are shown for two representative cardiac phases from NOR, DCM, HCM, and RVA subjects. Each column corresponds to one competing method, and the color scale denotes the point-wise surface error relative to the reference mesh. The LV and RV volume curves on the right indicate the selected phases using vertical dashed lines.

Figure 5: Functional fidelity of the synthesized biventricular sequences. (a) Diagnostic group-stratified relative volume-change trajectories for the LV (top) and RV (bottom). Solid and dashed curves represent the reference and synthesized sequences, respectively, and shaded bands indicate inter-subject variability. (b) Agreement between reference and synthesized left- and right-ventricular ejection fractions (LVEF and RVEF), with the identity line shown for reference. (c) Normalized KL divergence and Wasserstein distance (WD) between the reference and synthesized left- and right-ventricular end-systolic volumes (LVESV and RVESV), LVEF, and RVEF distributions. Lower values indicate better distributional agreement. MAE: mean absolute error.

Figure 6: Regional analysis of motion-informed parcellation. (a) Surface-group vRMSE for the AHA-like partition and motion-informed parcellation with different numbers of regions. (b) Region-wise vRMSE reduction obtained by the $N_{R}=16$ motion-informed partition relative to the AHA-like partition, reported for each diagnostic group and the full cohort. Positive values indicate lower errors with motion-informed parcellation.

Partition	ASSD (mm) $\downarrow$	HD95 (mm) $\downarrow$	vRMSE (mm) $\downarrow$
AHA-like	$1.59\pm 0.44$	$4.11\pm 1.74$	$3.62\pm 1.43$
$N_{R}=8$	$1.50\pm 0.37$	$3.88\pm 1.39$	$3.40\pm 1.19$
$N_{R}=16$	$\mathbf{1.49\pm 0.34}$	$\mathbf{3.77\pm 1.06}$	$\mathbf{3.31\pm 1.03}$
$N_{R}=32$	$1.59\pm 0.38$	$4.19\pm 1.36$	$3.62\pm 1.21$

Table 3: Effect of regional partition strategies on biventricular motion synthesis.

Topology setting	ASSD (mm) $\downarrow$	HD95 (mm) $\downarrow$	vRMSE (mm) $\downarrow$
w/o MRAB	$1.56\pm 0.34$	$3.92\pm 1.27$	$3.51\pm 1.08$
$L_{\mathrm{hop}}=1$	$1.55\pm 0.42$	$3.81\pm 1.71$	$3.35\pm 1.38$
$L_{\mathrm{hop}}=2$	$1.49\pm 0.38$	$\mathbf{3.75\pm 1.55}$	$3.31\pm 1.24$
$L_{\mathrm{hop}}=3$	$1.49\pm 0.43$	$3.86\pm 1.80$	$3.37\pm 1.44$
$L_{\mathrm{hop}}=4$	$\mathbf{1.49\pm 0.34}$	$3.77\pm 1.06$	$\mathbf{3.31\pm 1.03}$
$L_{\mathrm{hop}}=5$	$1.55\pm 0.40$	$3.94\pm 1.45$	$3.41\pm 1.31$

Table 4: Effect of MRAB on biventricular motion synthesis performance. MRAB: multi-hop regional attention bias.

$c_{ED}$	Adapter	$c^{motion}$	ASSD (mm) $\downarrow$	HD95 (mm) $\downarrow$	vRMSE (mm) $\downarrow$
Condition	Route	Control	ASSD (mm) $\downarrow$	HD95 (mm) $\downarrow$	vRMSE (mm) $\downarrow$
$\times$	$\times$	$\times$	$1.64\pm 0.42$	$4.17\pm 1.78$	$3.60\pm 1.43$
$\checkmark^{G}$	$\times$	$\times$	$1.56\pm 0.37$	$4.09\pm 1.53$	$3.47\pm 1.25$
$\checkmark^{F}$	$\times$	$\times$	$1.51\pm 0.36$	$3.92\pm 1.12$	$3.36\pm 1.16$
$\times$	$\checkmark^{G}$	$\times$	$1.52\pm 0.36$	$3.99\pm 1.47$	$3.40\pm 1.16$
$\times$	$\checkmark^{F}$	$\times$	$1.54\pm 0.37$	$3.97\pm 1.44$	$3.43\pm 1.17$

Table 5: Ablation of phenotype conditioning, prototype routing, and optional functional control in the rectified-flow generator. Superscripts

G

and

F

denote the global descriptor

c_{ED}^{global}

and fine-grained descriptor

c_{ED}^{fine}

, respectively. The final row uses the additional motion descriptor

c^{motion}

and is not part of the primary ED-only setting.

Figure 7: Diagnostic-group distributions across prototype-routed motion adapters. Row-normalized adapter assignments are shown for routing using the anatomical latent together with (a) the global ED descriptor $c_{ED}^{global}$ or (b) the fine-grained ED descriptor $c_{ED}^{fine}$ . Each cell reports the percentage of subjects in a diagnostic group assigned to the corresponding adapter.

$N_{e}$	ASSD (mm) $\downarrow$	HD95 (mm) $\downarrow$	vRMSE (mm) $\downarrow$
$\mathbf{4}$	$1.49\pm 0.34$	$\mathbf{3.77\pm 1.06}$	$\mathbf{3.31\pm 1.03}$
6	$\mathbf{1.47\pm 0.38}$	$3.82\pm 1.58$	$3.36\pm 1.30$
8	$1.50\pm 0.40$	$3.94\pm 1.72$	$3.41\pm 1.38$

Table 6: Effect of the number of routing prototypes/ adapters

N_{e}

on biventricular motion synthesis performance.

Figure 8: Geometric errors across in-distribution (ID) and out-of-distribution (OOD) diagnostic groups. (a) Four ID groups shared by ACDC, M&Ms, and M&Ms-2. (b) Five OOD groups excluded from training: hypertensive heart disease (HHD) and ischemic heart disease (IHD) from M&Ms, and tetralogy of Fallot (FALL), interatrial communication (CIA), and tricuspid regurgitation (TRI) from M&Ms-2.

Figure 9: Illustration of full-cycle synthesized mesh sequences for three representative OOD cases and their corresponding LV and RV volume trajectories.

Figure 10: Controllable modulation of synthesized cardiac motion. (a) Mean LV and RV displacement trajectories for the reference, the ED-only synthesis, and predictions obtained using the optional functional-control branch. With $c^{motion}$ uses the subject-specific motion descriptor of the illustrated NOR case, whereas the DCM-mean and HCM-mean curves use the corresponding group-average descriptors. (b) Response of a representative NOR case to control scale ranging from 0 to 2, including the relationship between the control coefficient and peak normalized mean displacement and the corresponding spatial maps of the $c^{motion}$ -induced deformation at $s=0$ , $s=1$ , and $s=2$ .

Explore similar work

May 20, 2026

RePCM: Region-Specific and Phenotype-Adaptive Bi-Ventricular Cardiac Motion Synthesis

Cardiac motion over a cardiac cycle is crucial for quantifying regional function and is strongly affected by cardiovascular diseases. Since temporally dense mesh sequences are difficult to obtain in practice, we focus on leveraging the more accessible end-diastolic frame to infer a full-cycle sequence. Due to strong regional and disease-specific differences, traditional methods often oversmooth the data by relying on generative models that are optimized for global patterns. To address this problem, we propose Region-Aware and Phenotype-Adaptive Bi-Ventricular Cardiac Motion Synthesis (RePCM) for single frame Bi-ventricular mesh motion completion. In Stage I, a reconstruction network learns vertex wise motion descriptors and clustering yields a data driven functional partition, providing an explicit motion derived region structure. In Stage II, a Region-Specific Injection Module enforces masked, synchronized region exchange within a conditional VAE, preserving localized specific dynamics and restricting cross-region mixing. A Phenotype-Adaptive Mixture-of-Experts prior conditioned on ED shape uses anatomy-guided cues to model latent motion trends and capture inter-disease variability. Experiments on three datasets covering different cardiovascular diseases show consistent gains in geometric and functional metrics and improved preservation of region specific dynamics.

Xuan Yang, Xiaohan Yuan, Hao Li +3

School of Biomedical Engineering, National University of Singapore, Singapore · School of Automation, Southeast University, Nanjing, China · School of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China +1

Medical Image Synthesis

Ask PDF

Oct 6, 2026

One Frame, Full Heartbeat: ECG-Free 4D Cardiac Cine MRI Synthesis via Radial-Decomposed Flow Matching

Cine cardiovascular magnetic resonance (CMR) captures the cardiac cycle as a four-dimensional (4D) sequence, but standard acquisition requires electrocardiogram (ECG) gating and repeated breath holds. Visual realism alone does not establish accurate patient-specific ejection fraction (EF) or ventricular volumes. We present PhaseFlow3D, a generative framework that synthesizes a complete 4D cine sequence from a single end-diastolic (ED) three-dimensional (3D) volume without ECG. To capture asymmetric systolic and diastolic dynamics, it represents the cardiac cycle as a piecewise linear phase anchored at ED and end-systolic (ES) time points. At inference, a population-level canonical template supplies this phase without patient-specific temporal information. A phase-conditioned rectified flow model generates a cardiac motion trajectory in latent space. Radial Contraction Decomposition converts each latent state into a 3D displacement field, combining a physics-informed radial component for centripetal myocardial contraction with an image-conditioned residual for rotation and out-of-plane motion. Each frame is generated by directly warping the ED volume, bypassing variational autoencoder decoding. On the combined ACDC and M&Ms benchmark, PhaseFlow3D achieves the lowest EF mean absolute error, the only positive left-ventricular volume-curve

R^2

, and the best distributional quality among compared methods. Ablations confirm each component's contribution. Downstream evaluations demonstrate the utility of the synthesized sequences and displacement fields for segmentation, pathology classification, label propagation, and myocardial strain analysis.

Shiyi Wang, Ruochen Sun, Xiang Li +2

Department of Electrical and Computer Engineering, Johns Hopkins University, Baltimore, MD, USA · Gordon Center for Medical Imaging, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA · Center for Advanced Medical Computing and Analysis, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

Medical Image Synthesis Cardiac MRI 4D Reconstruction Physics-Informed Generative Modeling Flow-Based Generative Modeling

Ask PDF

Jun 2, 2026

Conditional Latent Diffusion Model with Fourier-based Motion Modelling for Virtual Population Synthesis

In-silico trials of medical devices require the generation of virtual populations of anatomies. In cardiovascular applications, virtual anatomy is typically represented as a 3D+t mesh sampled from a generative model. However, most existing mesh generators focus on static anatomy, while sequence models often lack explicit periodicity. To this end, we propose 4D F-MeshLDM, a conditional generative framework comprising a convolutional mesh VAE to encode meshes, a structural latent space that parameterises motion using a truncated Fourier series, and a diffusion prior that learns the latent distribution over Fourier coefficient tokens. By conditioning the diffusion process on clinical covariates via affine modulation, we enable controllable synthesis. Sampling tokens and performing inverse Fourier synthesis yield cycle-consistent latent trajectories, which can be decoded into 3D+t cardiac mesh sequences. Experiments on 5,000 UK Biobank subjects demonstrate that 4D F-MeshLDM outperforms state-of-the-art baselines in anatomical fidelity and achieves near-zero cycle closure error. Furthermore, the generated cohorts accurately preserve clinical functional indices, highlighting the potential of our framework for reliable in-silico cardiac trials.

Shaokun Lan, Haoran Dou, Jinghan Huang +5

Centre for Computational Imaging and Modelling in Medicine (CIMIM), University of Manchester, Manchester, UK

3D Mesh Generation Conditional Diffusion Models

Ask PDF