[2D/3D liver image matching based on curvature signature]

Evaluating the volume of liver is very important to diagnose and cure some liver diseases. We have proposed a method in which ultrasound liver image and 3D standard liver model are combined to estimate the liver volume. The key to estimate liver volume lies in the accuracy of image registration between ultrasound liver curve and corresponding 3D liver model. Because the ultrasound liver curve and 3D model outline did not have the same scale, we used scale invariant signature based on the integral of curvature (SISIC) to carry out the curve matching. This method is not affected by the different scales between the two curves, and the results indicated that the proposed method implements a good performance.