Comments on "Retrospective motion correction in digital subtraction angiography: a review"

Although Close and Whiting found the paper by Meijering el al. (ibid., vol. 18, p. 2-21, 1999) to be both thorough and informative, Close and Whiting feel compelled to correct what they feel to be an inaccurate characterization of their work. Meijering et al. refer to Close and Whitings' 1995 paper ("Motion compensated signal and background estimation from coronary angiograms," in Medical Imaging 1995: image processing, M. H. Loew, Ed., Proc. SPIE, vol. 2434, p. 185-94, 1995) as being "for the purpose of motion analysis" and, more specifically, "for the analysis of cardiac or vascular motion (vessel tracking)." Although Close and Whiting do perform vessel tracking, the purpose of their work is to obtain an accurate densitometric representation of the projected vessel lumen. Indeed, the title of their 1995 paper, "Motion-compensated signal and background estimation from coronary angiograms" clearly refers to densitometric estimation rather than analysis of motion. The basis of Close and Whitings' approach is the observation that although the anatomical motion is inherently three-dimensional (3-D) and nonuniform, the motion of a short vessel segment can be approximated well by a single rigid motion. Therefore, Close and Whiting expect that by decomposing an angiographic image sequence into moving layers, one can separate the densitometric profile of the vessel from background structures which have different motion.