Automated estimation of regional mean transition times and radial velocities from cine magnetic resonance images: Evaluation in normal subjects

Abstract

To assess regional ventricular function via an accurate and automated definition of functional parameters. An automated method is proposed that estimates reliable regional normalized mean transition times (F(mc)) and mean radial velocities (V(m)) from cine images. This approach combines a quantitative parametric imaging method and an automated detection of the endocardial border, which is robust to the presence of papillary muscles and nonhomogeneities within the left ventricular cavity. Steady-state free-precession cine-magnetic resonance imaging (MRI) of 36 healthy subjects was analyzed. The quality of the automated segmentation was assessed on a subgroup of 20 subjects by comparing the results with the manual contours traced by an expert. The comparison of functional parameters estimated consequently using the automated and the manual contours yielded (y = 0.959x + 0.016, R = 0.964) for F(mc) and (y = 0.883x + 0.505, R = 0.935) for V(m). On the entire group, F(mc) was equal to 0.392 +/- 0.069 and V(m) to 5.4 +/- 2.3 cm/s. Increasing values of the temporal parameter from the apex to the base and larger values in the septal wall than in lateral wall were demonstrated and were in accordance with the physiology. The proposed method ensures an automated and robust assessment of regional wall motion parameters, which could be clinically useful.