We present a new method for detailed tracking of cardiac motion based on MR-tagging imaging, multi-scale optical flow, and HARP-like image filtering.In earlier work, we showed that the results obtained with our method correlate very well with Phase Contrast MRI. In this paper we combine the intrinsic spatiotemporal evidence from perpendicularly encoded tagging sequences so as to obtain a dense, unambiguous motion field. We compare our results to those obtained using a different (provably incorrect) strategy advocated in the literature, in which horizontal and vertical components of the motion field are extracted separately from the motion pattern of vertically, respectively horizontally initialized tags. Fromthe extracted motion fields, rotation of the LV is calculated for three subjects, in three slices using the generalised Stokes’ theorem.
|Title of host publication||Analysis of Functional Medical Images (MICCAI 2008 Workshop, New York, USA, September 10, 2008)|
|Editors||G. Hamarneh, R. Abugharbieh|
|Publication status||Published - 2008|