(E-170) On the non-invasive estimation of end diastolic pressure volume relationship in the right ventricle.

Introduction:: Functional measurements of the diastolic performance of the right ventricle (RV) have been correlated with disease outcomes in pathologies such as pulmonary hypertension, and heart failure. The stiffening of the myocardium through fibrosis and hypertrophy is a significant contributor to RV diastolic dysfunction. The “gold standard” predictor of RV stiffness is the measurement of the RV end-diastolic pressure volume relationship (RV-EDPVR). However, the clinical utility of this metric is primarily hampered by the requirement of vena cava occlusion, rendering it impractical in humans. We hypothesize that a reliable anatomy-independent RV-EDPVR can be estimated through the successful recapitulation of the end-diastolic behavior of the biventricular cardiac function. Klotz et. al [1] suggested a methodology to estimate a universal single beat EDPVR in the left ventricle (LV) through a combination of experimental, and statistical methods. This estimation has been shown to provide an accurate anatomy-independent assessment of the LV-EDPVR. In this work, we implement an in-silico approach via the implementation of finite element models to train a regression model and predict a universal RV-EDPVR.

Materials and Methods:: Finite element simulations: Ex-vivo cardiac magnetic resonance images of 25 rats were used to synthesize the biventricular geometries (Fig. 1). The sample size was increased to 2000 through Latin hypercube sampling for various geometric, material, and architectural properties through an in-house framework [2]. Subsequently, forward simulations were performed by ramping the LV and RV chamber pressure from 0 to 30, and 6 mmHg, respectively. Normalization: To improve the accuracy of the predicted EDPVR, the spread of the variance was reduced through normalization < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] >, with < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] >and < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] > denoting the ventricular volume at the minimum and maximum ventricular pressure. Thus, the resulting normalized EDPVR takes the form, < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] > Single-point EDPVR estimation: The data was used to train a linear regression model, with select excluded datasets used to validate the predictivity of the framework. With the calculated universal constants < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] > and < ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] >the individual EDPVR can thus be obtained by using the EDP and EDV.


Results, Conclusions, and Discussions:: In our preliminary investigations, the regression framework was used to determine the LV-EDPVR for each geometry. The regression model was compared to the Klotz curve estimation (< ![if !msEquation] >< ![if !vml] >< ![endif] >< ![endif] > = 27.78, 2.76) (Fig. 2A), with the predicted universal constants proving to be within an error of 0.1%. The model was validated using arbitrary pressure and volume measurements (Fig. 2B), with the model showing promise in estimating a universal curve for the LV-EDPVR. The following steps will involve the extension of the regression training and testing sets to include the RV data, with varying standards for normalization and validation to account for the non-convexity of its geometry.

A universal estimation of the RV-EDPVR is crucial in the establishing the diastolic performance of the RV under normal and pathological condition. Through this complete in-silico single beat estimation, the necessity of invasive methods such as catheterization can be avoided, thus increasing the feasibility of RV-EDPVR as functional marker in clinical diagnosis.


Acknowledgements (Optional): :

References (Optional): : 1] Klotz S., et al., Am. Journal of Physiol.-Heart and Circ. Physiol. 2006; 0363-(6135)1522-1539. [2] Babaei H., et al., Sci. Reports. 2022; 12-(1) 2045-2322.