Feature tracking for assessment of diastolic function by cardiovascular magnetic resonance imaging

Aim: To assess the agreement of cardiovascular magnetic resonance imaging (CMRI) feature-tracking (FT) parameters with echocardiography to diagnose diastolic dysfunction; to determine whether a similar parameter to mitral inflow early diastolic velocity to early diastolic tissue velocity ratio (E/e') can increase accuracy of imaging by dividing the phase contrast (PC) mitral inflow E-wave (E) with a CMRI-FT parameter; to compare the agreement between CMRI-FT and PC diastolic function assessment using echocardiography.

Materials and methods: Patients (n=71; 43 abnormal diastolic function) undergoing both CMRI and echocardiography independently were included. Echocardiography was the reference standard. CMRI-FT analysed the short and long axis cine contours. PC images of mitral inflow, tissue velocity, pulmonary vein flow, and left atrial area were assessed.

Results: Using CMRI-FT, the area under the curve (AUC) for identifying diastolic dysfunction was >0.80 for radial and circumferential strain, systolic strain rate (SSR), and early diastolic strain rate (DSR). For cases with CMRI-determined left ventricular ejection fraction (LVEF) ≥50% (n=38), circumferential DSR was the only parameter with good accuracy (AUC=0.87; cut-off 0.93/s). E/circumferential DSR ratio and longitudinal strain had high accuracy in all patients (AUC=0.88 and 0.93 respectively) and CMRI-determined LVEF ≥50% (AUC=0.81; cut-off 76.7). Circumferential DSR showed the highest agreement with echocardiography (higher than E/circumferential DSR and PC assessment) in all cases (kappa 0.75; p<0001) and cases with CMRI LVEF ≥50% (kappa 0.73; p<0.0001).

Conclusions: CMRI-FT circumferential DSR showed the highest accuracy for determining diastolic dysfunction with good agreement with echocardiography. Circumferential DSR had higher accuracy than E/circumferential DSR and PC.