Comprehensive 4D flow MRI characterisation of left atrial hemodynamic flow components in hypertension and hypertrophic cardiomyopathy
Paula Casademunt1, Xabier Morales1, Ayah Elsayed2, Ada Doltra3,4, Marta Sitges3,4, Oscar Camara1
1: Physense, BCNMedtech, Department of Engineering, Universitat Pompeu Fabra, Barcelona, Spain; 2: Auckland Bioengineering Institute, University of Auckland, Auckland, New Zealand; 3: Cardiovascular Institute, Hospital Clínic, Universitat de Barcelona, Barcelona, Spain; 4: Institut d’Investigacions Biomèdiques Agust Pi i Sunyer (IDIBAPS), Barcelona, Spain
Introduction
The left atrium (LA) supports efficient ventricular filling by acting as reservoir, conduit, and booster pump. Chronic pressure overload conditions such as hypertension (HTN) and hypertrophic cardiomyopathy (HCM) induce maladaptive atrial remodelling, including dilation, fibrosis, and reduced compliance. These structural changes impair atrial function, promote stasis, and increase thromboembolic risk. Capturing such functional alterations remains challenging with conventional imaging: while four-dimensional (4D) flow magnetic resonance imaging (MRI) enables comprehensive velocity mapping, most analyses remain restricted to qualitative visualization or global indices. To address this, we implemented a semi-automated Lagrangian-tracking pipeline that decomposes LA blood flow into distinct functional components. This study represents the first use of the methodology on 4D flow MRI data to compare LA hemodynamics in HTN, HCM, and controls.
Materials & Methods
We analyzed 4D flow MRI datasets from 109 subjects (29 controls, 47 HTN, 33 HCM), and LA segmentation was performed using nnU-Net [1]. Lagrangian particle tracking categorised blood flow into conduit, reservoir, delayed ejection, retained, residual, and pulmonary venous (PV) backflow components, based on the pipeline developed by Back et al. [2]. For each, we quantified volumes, velocities, and kinetic energy (KE). Representative trajectories of each flow component were extracted via k-medoids clustering. PV and mitral valve velocity curves were derived using spherical sampling, producing pseudo-Doppler profiles, as proposed by Morales et al. [3]. Statistical comparisons employed Kruskal–Wallis and pairwise Mann–Whitney U tests, with functional MANOVA for temporal curves.
Results
Controls exhibited significantly larger conduit fractions (p < 0.001) and lower retained and residual fractions (p < 0.01) compared to both patient groups, indicating impaired LA emptying and a shift toward stasis-prone flow in HTN and HCM. PV backflow was reduced in diseased groups (p < 0.001). Pathline geometry revealed less tortuous reservoir trajectories, consistent with impaired vortex formation. Pseudo-Doppler analysis demonstrated attenuated PV S and D waves and diminished mitral E-wave velocities in both diseases, with compensatory increases in A-wave amplitudes. KE analysis showed marked alterations: conduit KE peaks were dampened and reservoir KE was temporally broadened in both diseased groups, indicative of impaired LA mechanical performance.
Discussion & Conclusions
Our results demonstrate that HTN and HCM drive a reproducible LA hemodynamic phenotype characterised by reduced conduit efficiency, increased stasis, and energetically inefficient blood transport. The semi-automated pipeline applied to 4D flow MRI provides a robust framework for quantifying intra-atrial flow organisation beyond global indices. These quantitative biomarkers complement established strain measures and may improve non-invasive risk stratification for thromboembolic events. Future longitudinal studies are warranted to evaluate their prognostic value.
|
