Evaluation of a vessel-tracking-based technique for dynamic targeting in human liver

The purpose of this study was to evaluate a novel vessel-tracking-based technique for tracking of human liver. The novelty of the proposed technique is that it measures the translation and deformation of a local tissue region based on the displacements of a set of vessels of interest instead of the entire organ. The position of the target point was estimated from the relative positions of the center-of-masses of the vessels, assuming that the topological relationship between the target point and center-of-masses is unchanged during breathing. To reduce inaccuracy due to the delay between vessel image acquisition and sonication, the near-future target position was predicted based on the vessel displacements in the images extracted from an image library acquired before the tracking stage. Experiments on healthy volunteers demonstrated that regardless of the respiratory condition, appropriate combinations of three center-of-masses from the vessels situated around the target-tissue position yielded an estimation error of less than 2 mm, which was significantly smaller than that obtained when using a single center-of-mass trio. The effect of the tracking delay was successfully compensated, with a prediction error of less than 3 mm, by using over four images selected from the image library.