“…Measuring myocardial deformations from tagged MR sequences relies on estimating a dense displacement field consistent with the motion of the structured tagging pattern (Kerwin et al, 2001;Amini and Prince, 2001;McVeigh and Ö ztü rk, 2001). Classical approaches comprise: (i) differential optical flow-based methods coupled with intensity correction schemes (Prince and McVeigh, 1992;Denney, 1994;Denney and Prince, 1994;Gupta and Prince, 1995;Gupta et al, 1997;Dougherty et al, 1999); (ii) phase-based optical flow methods such as HARP (Osman et al, 1999;Osman and Prince, 2000;Amini and Prince, 2001); (iii) segmentation-based approaches combining tag segmentation, sparse motion estimation along tags, and dense motion interpolation over the image domain (Kumar and Goldgof, 1994;Guttman et al, 1994;Denney and Prince, 1995;Kraitchman et al, 1995;Young et al, 1995;Park et al, 1996;Declerck, 1997;Guttman et al, 1997;Amini et al, 1998;Denney, 1999;Huang et al, 1999;Kerwin and Prince, 1999;Clarysse et al, 2000;Haber et al, 2000;Ö ztürk and McVeigh, 2000;). However, their clinical applicability is questioned by intrinsic limitations: differential optical flow fails for large displacements due to the local nature of constraint equations; phase-based optical flow is limited to 1-D tags and proves to be sensitive to artifacts; tag segmentation often requires supervision to deal with tag contrast attenuation over time due to T1 demagnetization (tag fading out), and tag destruction along myocardial boundaries induced by blood flows within the cardiac chambers.…”