Project Summary/Abstract Heart failure with preserved ejection fraction (HFpEF) is characterized by a stiff heart which cannot fill adequately with blood, without increasing filling (left atrial) pressure, which in turn results in pulmonary congestion. Patients with HFpEF have unexplained breathlessness upon exercise. Its hemodynamic correlate is elevated cardiac pressures. HFpEF is common, represents 50% of all HF, and has similar mortality rates as HF with reduced ejection fraction (50% in five years). A major challenge for HFpEF is that diagnosis by echocardiography is challenging, with many inconclusive studies, and accuracy of about 68% to predict high pressures, with especially low sensitivity. Therapies for HFpEF are lacking, and their development would be aided by better identification of HFpEF. MRI is not able to diagnose HFpEF at all. To overcome these challenges, we propose to develop MRI methods for diastolic function, including an approach to measuring tricuspid regurgitant flow velocity, which is used to estimate pressures in the right heart, a key metric in diastolic function evaluation. This method uses deep learning processing of long-axis cines to generate a dynamic slice plane prescription, which is used in valve-tracking phase-contrast to measure tricuspid regurgitant flow. This same technique, suitably modified can measure all or many of the diastolic parameters (E, A, e' etc). To this end, we develop a valve-tracking PC approach which is carefully tested in phantoms and human subjects, comparing regurgitation velocities by MRI vs. the standard. We also develop methods for diastolic evaluation with 4D flow, including bSSFP 4D flow, with an innovative approach. Then the importance of exercise is studied to unmask diastolic dysfunction. All of these techniques are applied in patient cohorts where excellent ground truth data is available, first in patients with correlative echocardiography and MRI (n=40), secondly in patients with implanted pulmonary artery pressure sensors (n=20, imaged at two time points and using rest/stress, and thirdly in patients with invasive right heart catheterization to measure all pressure data (n=40). We hypothesize that MRI can measure TR pressure gradients accurately, and can detect changes in diastolic parameters with exercise. We hypothesize that a single breath-held valve-tracking phase- contrast acquisition could be used for diastolic evaluation of all needed parameters. We envision that every cardiac MRI exam can employ a few additional breath-holds to fully evaluate diastolic function, thus reaching its potential for diagnosis of HFpEF.