Estimation of mean pulmonary artery pressure by cardiovascular magnetic resonance four-dimensional flow and compressed sensing

BackgroundFour-dimensional (4D) phase-contrast cardiovascular magnetic resonance (CMR) allows for precise non-invasive estimation of mean pulmonary artery pressure (mPAP) by estimating the duration of pathological vortex persistence in the main pulmonary artery. This has previously been achieved with compressed sensing acceleration of a multiple two-dimensional (CS-M2D) flow sequence, but acquisition using a true time-resolved 3D excitation (CS-4D) offers theoretical advantages including spatiotemporal coherence. This study aimed to validate a state-of-the-art CS-4D sequence with a previously utilized CS-M2D sequence for estimating mPAP, and compare both to right heart catheterization (RHC). MethodsThe study included patients clinically referred for CMR (n=45), of which a subgroup (n=20) had prior mPAP of >16 mmHg confirmed by RHC. CMR was performed at 1.5T using CS-M2D and CS-4D sequences covering the main pulmonary artery. mPAP was estimated using a previously published linear relationship between vortex duration and mPAP. Agreement between CS-M2D and CS-4D estimates was quantified, including analysis of intra- and interobserver variabilities. The diagnostic performance of CS-M2D and CS-4D in predicting mPAP was further compared to gold-standard RHC. ResultsCS-M2D and CS-4D both had average scan durations under 3 minutes (175{+/-}36 and 135{+/-}34 seconds, respectively). Estimated mPAP by CS-4D and CS-M2D were strongly correlated (R2=0.93, p<0.001), with negligible mean{+/-}SD bias (0.0{+/-}2.7 mmHg) and good reproducibility. There was excellent agreement with RHC for both CS-M2D (R2=0.92, p<0.001, bias 0.6{+/-}3.1 mmHg) and CS-4D (R2=0.86, p<0.001, bias 1.1{+/-}4.5 mmHg). ConclusionsCS-4D and CS-M2D sequences effectively yield interchangeable non-invasive estimations of mPAP, with excellent agreement compared to invasive RHC. They can both be acquired in a scan time applicable to clinical workflow, offering a promising tool for non-invasive mPAP estimation in clinical practice.