1351077 - Imaging Regional Pressure Changes in Aortic Dissection – Non-Invasive Mapping by 4D Flow MRI

Background: Treatment plans for patients with type B aortic dissection (TBAD) rely on predicting false lumen (FL) growth as a key indicator of future rupture risk [1]. However, current anatomic predictors only weakly predict FL growth [2]. We recently proposed a method to predict FL growth by estimating relative pressure drop across the FL using 4D flow MRI [3], which correlated well with clinically observed growth rates. Use of relative pressure mapping from 4D flow MRI in aortic dissection has yet to be validated, however. Moreover, the complexity of pressure distributions in aortic dissection remains unknown. Herein, we validated two algorithms that calculate relative pressure and its spatial distribution, comparing 4D flow MRI and catheter-based pressure measurements from 3D-printed TBAD models [4]. Results were then compared within a TBAD patient cohort, exploring the use of 3D pressure mapping in vivo.

Methods: For validation, we used an in vitro flow setup with an embedded compliant 3D-printed TBAD geometry derived from a clinical CT exam with applied physiologically accurate boundary conditions [3]. 4D flow MRI was acquired across 3 models: one baseline model (BL) and two with reduced FL entry (BL_ent) and exit (BL_ext) tears, respectively. After acquisition, BL and BL_ent were spatiotemporally upsampled by a factor of 2 with sinc interpolation. Twelve invasive pressure measurements were also acquired at consistent locations across each model (Fig. 1). Relative pressures were calculated from the 4D flow images using two methods: vWERP (virtual work-energy relative pressure) and STE (stokes estimator) and compared directly to invasive measurements. Further, the regional pressure mapping of STE was applied to 4D flow MRI in a cohort of TBAD cases with a range of FL growth rates.

Results: vWERP and STE both showed very high agreement with invasive catheter measurements (Fig 2.). vWERP correlations were tight (r² = 0.91), although with a slight underestimation at peak relative pressures (slope = 0.75, mean bias = 0.07 ± 1.9 mmHg). STE correlations were similarly tight (r² = 0.91), with slightly larger underestimation at peak relative pressure (slope = 0.68, mean bias = 0.01 ± 1.9 mmHg). STE relative pressure maps for the clinical cases supported the utility of the approach, demonstrating both higher magnitude and extent of FL pressures in a case with clinically observed rapid FL growth rate (Fig 3.).

Conclusion: Image-based vWERP and STE show strong agreement with in vitro pressure data in a standard TBAD anatomy, although with a tendency to mildly underestimate peak relative pressures. The reason for this underestimation is uncertain, although we suspect insufficient spatiotemporal sampling plays a role [5]. Nevertheless, the data demonstrate the potential of 4D flow MRI for quantitative hemodynamic estimations of FL pressure, with the first-ever implementation of STE in an in vivo setting showing an intuitive agreement between FL growth rate and image-derived pressure metrics.