Session SA1.5

Influence of the Temporal Resolution on the Quantification of Displacement Fields in Cardiac MR Tagged Images

J García-Barnés, D Gil, A Bajo, MJ Ledesma-Carbayo, C Santa-Marta*

Universitat Autónoma de Barcelona
Barcelona, Spain

It is difficult to acquire tagged cardiac MR images with a high temporal resolution using clinical MR scanners. Our aim is to assess the influence of the temporal resolution of a series of images when it is used to quantify myocardial motion parameters based on displacement fields. Our images are acquired in a Philips Intera 1.5 T scanner (Philips Medical Systems, The Netherlands), with a SPAMM prepared Turbo Field Echo sequence. This sequence has been prepared by the authors, it is synchronized with both the electrocardiogram and the respiration, and the images are acquired while the patient is breathing freely. The temporal resolution is 13 phases for 80 bpm and the acquisition time is 1’12’’. To double this temporal resolution a second series of images is acquired, but introducing a time shift between the first frame and the triggering signal. The time shift corresponds to half of the spacing time between two consecutive frames. We interleave these two acquisitions S1 and S2, and obtain a new series that doubles the temporal resolution of the individual ones, halving the temporal gap between frames. The tissue deformation (computed with the Harmonic Phase Flow, HPF, in our case) extracted for each sequence, S1 and S2, provides two vector fields, Vi (for i=1, 2) sampled at the cardiac phases each frame was acquired. The vector field connecting such phases, IV, is obtained by applying HPF to the interleaved series To assess the influence of the temporal resolution, we calculate the displacement field for S1, S2 and the interleaved images with HFP and compare the results with the displacement obtained tracking manually-placed points on the myocardium. The quadratic error provides our quality measure for quantifying the influence of temporal resolution: it is 27% smaller for the interleaved sequence than for S1 or S2 when processed independently. We can conclude that if we are to use the image series for quantification, the results will improve in more than 20% if the temporal resolution doubles.

(Abstract Control Number: 272)