Determination of aortic compliance from magnetic resonance images using an automatic active contour model

The possibility of monitoring changes in aortic elasticity in humans has important applications for clinical trials because it estimates the efficacy of plaque-reducing therapies. The elasticity is usually quantified by compliance measurements. Therefore, the relative temporal change in the vessel cross-sectional area throughout the cardiac cycle has to be determined. In this work we determined and compared the compliance between three magnetic resonance (MR) methods (FLASH, TrueFISP and pulse-wave). Since manual outlining of the aortic wall area is a very time-consuming process and depends on an operator's variability, an algorithm for the automatic segmentation of the artery wall from MR images through the entire heart cycle is presented. The reliable detection of the artery cross-sectional area over the whole heart cycle was possible with a relative error of about 1%. Optimizing the temporal resolution to 60 ms we obtained a relative error in compliance of about 7% from TrueFISP (1.0 x 1.0 x 10 mm3, signal-to-noise ratio (SNR) > 12) and FLASH (0.7 x 0.7 x 10 mm3, SNR > 12) measurements in volunteers. Pulse-wave measurements yielded an error of more than 9%. In a study of ten volunteers, a compliance between C = 3 x 10-5 Pa-1 and C = 8 x 10-5 Pa-1 was determined, depending on age. The results of the TrueFISP and the pulse-wave measurements agreed very well with one another (confidence interval of 1 x 10-5 Pa-1) while the results of the FLASH method more clearly deviated from the TrueFISP and pulse-wave (confidence interval of more than 2 x 10-5 Pa-1)