TY - JOUR
T1 - Determining pulse wave velocity using MRI
T2 - a comparison and repeatability of results using seven transit time algorithms
AU - Butlin, M.
AU - Hickson, S.
AU - Graves, M. J.
AU - McEniery, C. M.
AU - Avolio, A. P.
AU - Wilkinson, I. B.
PY - 2008
Y1 - 2008
N2 - Aim: MRI provides a non-invasive method for assessing segmental aortic pulse wave velocity (aPWV). However, the best mathematical algorithm for transit time calculation using MRI flow waves is unclear. Methods: 7 different algorithms were applied to aortic flow waveforms measured by MRI (10 subjects, 36±7 years, 4 male). Two measurements were recorded in each subject on different days for repeatability analysis. PWV was calculated between 5 sites along the aorta. Outlier PWV results were classed as a ‘‘failed’’ measurement and the success rate calculated. Bland- Altman plots were constructed for each algorithm, and repeatability calculated. Agreement between different methods was calculated using repeated measures analysis. Results: The method of intersecting lines of fit during late diastole and early systole had the highest success rate followed by the Fourier analysis phase- slope method (99%; 98% respectively). Repeatability of measurement was highest using the phase-slope method followed by the method of intersecting lines (standard deviation 1.9; 2.2 m/s respectively). Methods of deviation of a systolic line of fit, maximum of second derivative, intersecting lines of fit, and the corner detection algorithm had the highest agreement, corrected for repeatability (corrected standard deviation range 1.8-1.9 m/s). Conclusions: Whilst agreement between several PWV algorithms was high, no one algorithm was better in all categories. The intersection of lines of fit method was most robust. The phase-slope method showed the greatest repeatability. These findings are important in aPWV measurement, and for reliable and accurate PWV measurement in general.
AB - Aim: MRI provides a non-invasive method for assessing segmental aortic pulse wave velocity (aPWV). However, the best mathematical algorithm for transit time calculation using MRI flow waves is unclear. Methods: 7 different algorithms were applied to aortic flow waveforms measured by MRI (10 subjects, 36±7 years, 4 male). Two measurements were recorded in each subject on different days for repeatability analysis. PWV was calculated between 5 sites along the aorta. Outlier PWV results were classed as a ‘‘failed’’ measurement and the success rate calculated. Bland- Altman plots were constructed for each algorithm, and repeatability calculated. Agreement between different methods was calculated using repeated measures analysis. Results: The method of intersecting lines of fit during late diastole and early systole had the highest success rate followed by the Fourier analysis phase- slope method (99%; 98% respectively). Repeatability of measurement was highest using the phase-slope method followed by the method of intersecting lines (standard deviation 1.9; 2.2 m/s respectively). Methods of deviation of a systolic line of fit, maximum of second derivative, intersecting lines of fit, and the corner detection algorithm had the highest agreement, corrected for repeatability (corrected standard deviation range 1.8-1.9 m/s). Conclusions: Whilst agreement between several PWV algorithms was high, no one algorithm was better in all categories. The intersection of lines of fit method was most robust. The phase-slope method showed the greatest repeatability. These findings are important in aPWV measurement, and for reliable and accurate PWV measurement in general.
U2 - 10.1016/j.artres.2008.08.338
DO - 10.1016/j.artres.2008.08.338
M3 - Meeting abstract
SN - 1872-9312
VL - 2
SP - 99
JO - Artery Research
JF - Artery Research
IS - 3
M1 - P1.31
ER -