Cardiac magnetic resonance imaging (CMR) is gaining traction in the evaluation of pulmonary hypertension (PH) as clinicians seek non-invasive ways to understand pulmonary artery haemodynamics and monitor disease course. Right heart catheterisation (RHC) remains the reference for diagnosis, yet it is invasive and limited to specialist centres. A review of 38 publications (2012–2024) examines CMR flow techniques for pulmonary hypertension, their correlation with invasive haemodynamics and practical trade-offs. Across thirty-eight studies published between 2012 and 2024, investigators applied two-dimensional (2D) and four-dimensional (4D) phase-contrast approaches alongside black blood imaging to characterise flow, relate it to mean pulmonary arterial pressure (mPAP) and pulmonary vascular resistance (PVR), and evaluate treatment response. The findings underscore strong associations between CMR-derived parameters and RHC measurements, with 4D flow bringing additional value for complex flow patterns.
Correlation With Invasive Haemodynamics
Multiple cohorts demonstrated that 2D flow parameters relate to invasive markers of disease severity. Mean pulmonary artery velocity, mean blood flow and velocity transfer functions showed moderate to strong associations with RHC-derived PVR. In pulmonary arterial hypertension (PAH), mean pulmonary artery flow velocity exhibited an inverse correlation of −0.88 with PVR, indicating lower velocities with higher vascular resistance. Additional 2D metrics, including minimum area and peak velocity, were used to approximate mPAP with varying strength. Beyond cross-sectional assessment, 2D peak velocity increased after pulmonary endarterectomy in chronic thromboembolic pulmonary hypertension (CTEPH), suggesting improved pulmonary flow post-treatment.
4D flow offered richer haemodynamic characterisation, particularly through the visualisation and quantification of vortical blood flow related to elevated pressures. In mixed PH cohorts, estimates of mPAP derived from visual detection of a pulmonary artery vortex closely matched catheter measurements, with correlations approaching unity. Vortex duration correlated with mPAP and PVR, providing a non-invasive indicator of pressure load across PH subgroups. The presence and extent of vortices in the main and right pulmonary arteries were strongly associated with PVR. Agreement analyses indicated that 2D and 4D methods yielded comparable stroke volume and pulmonary artery area measurements, highlighting consistency between techniques for key volumetric indices while 4D flow captured complex patterns more comprehensively.
Must Read: The Role of DL Reconstruction in Spectral CT Pulmonary Angiography
Black blood imaging, designed to suppress signal from flowing blood, contributed complementary information through qualitative assessment of slow or turbulent flow. A visual scoring system for slow flow correlated with both mPAP and PVR and added value to regression models predicting haemodynamics. Although this technique does not provide quantitative velocity vectors or directionality, its ability to highlight regions of disturbed flow supported its use alongside phase-contrast approaches.
Diagnostic Performance Across Flow Methods
Diagnostic accuracy varied by technique and parameter but was consistently high across several indices. In 2D flow applications, mean pulmonary artery flow velocity below 10 cm/s achieved perfect specificity in one cohort comparing PH and non-PH participants, while acceleration time delivered sensitivity of 90% and an area under the curve (AUC) of 0.92 in a PAH-focused sample. Relative area change thresholds also provided balanced sensitivity and specificity in mixed PH populations. These findings point to practical 2D cut-offs that can assist with ruling in or ruling out PH when interpreted in clinical context.
4D flow parameters further strengthened diagnostic classification. Vortex duration thresholds yielded sensitivity up to 100% and specificity up to 96%, with an AUC of 0.99 in identifying PH. Quantitative descriptors such as helicity also distinguished PH from controls with high accuracy. Importantly, 4D flow maintained excellent agreement with 2D measures for maximum and minimum pulmonary artery areas and stroke volume, supporting its reliability while offering enhanced pathophysiological insight. In head-to-head comparisons where available, 4D flow’s multidirectional sampling and retrospective plane prescription mitigated underestimation of peak velocities seen with single-plane 2D acquisitions, which can be affected by plane misalignment relative to true flow direction.
Qualitative black blood imaging delivered clinically useful performance through slow-flow artefact scoring, achieving sensitivity and specificity in the mid-80% range in large retrospective cohorts of suspected PH. When integrated into multivariable models, the presence of slow blood flow improved diagnostic accuracy compared with models based on phase-contrast features alone, underscoring the complementary nature of structural and flow-sensitive contrasts.
Methodological Constraints and Practical Considerations
Across the included literature, risk of bias varied, with a sizeable proportion of studies rated unsatisfactory and none providing sample size justification. Many cohorts were single-centred with small sample sizes, raising concerns about generalisability and statistical power. Time intervals between RHC and CMR ranged from minutes to a year, which could introduce discrepancies due to disease progression, particularly in severe PH. Exclusion of conference abstracts, preprints and non-English articles may also have contributed to publication bias within the evidence base.
Technical and operational factors influence method selection. 2D flow remains widely used because of shorter scan times and straightforward analysis, making it suitable for routine follow-up and treatment monitoring. However, reliance on predefined imaging planes can underestimate peak velocities if alignment is suboptimal. By contrast, 4D flow allows comprehensive volumetric coverage and retrospective plane placement, capturing complex patterns such as vortices that relate closely to mPAP and PVR. These advantages come with longer acquisition times, typically 10–20 minutes, and higher post-processing demands that may be challenging in patients with breathing difficulties or in time-constrained workflows. Temporal resolution may also limit precision for pressure estimation in some settings. Black blood imaging offers detailed visualisation of vessel walls and slow flow regions but remains qualitative, introducing potential observer variability and limiting reproducibility.
Guideline context recognises the value of CMR in PH assessment while stopping short of recommending it as a stand-alone diagnostic test. Within that framework, the reviewed evidence suggests that 2D parameters can inform severity estimation and response to intervention, 4D signatures can enhance pressure estimation and diagnostic discrimination, and black blood features can augment prediction models. The choice of technique should align with the clinical question, patient stability and resource availability, with consideration for integrating complementary sequences to balance speed, accuracy and interpretability.
Across diverse PH populations, CMR flow parameters show strong associations with invasive haemodynamics and deliver high diagnostic accuracy, with 4D flow providing particular value for complex pulmonary artery dynamics. 2D flow offers practical metrics for monitoring and treatment assessment, while black blood imaging contributes qualitative markers that strengthen predictive models. Methodological limitations, acquisition demands and variability in protocols temper interpretation, but the aggregated findings support a growing role for CMR flow assessment in the investigation, prognostication and follow-up of PH. These insights can inform imaging pathway design, modality selection and longitudinal evaluation strategies alongside established invasive standards.
Source: British Journal of Radiology
Image Credit: iStock
