Computation of liver deformations for minimally invasive surgery

In minimally invasive surgery (MIS), surgeons perform surgical planning by preoperative CT images which contain the spatial information of tumors and blood vessels in the targeted operation area. However, the spatial information will be compromised once the operation begins and the surgeon can only reply on his/her anatomical knowledge and judgment. In most cases, surgeons will work in an open-loop situation, without really knowing the anatomical information of the area under operations. In this paper, finite-element methods are used to compute the deformations of non-rigid tissues, such as liver, from the original tissue model constructed from the CT images, and the spatial information of the deformed tissue surface which is provided by endoscope. In our simulations, the mean error between the ground truth and the computed position is 0.35mm, which is quite satisfactory for practical usages. 94.12% of the grid points in liver and 98.94% of the grid points in vessels have errors under 1mm.