Evaluation of Liver Shape Approximation and Characterization

A computational framework is presented for 3-D liver shape approximation and characterization in order to determine the accuracy of shape reconstruction via Spherical Harmonics expansion. Spherical Harmonics is a powerful mathematical tool for expanding the shape. But in medical domain, livers have very variation geometry, in shape, size, and volume. In this regards, we evaluated and optimized the Spherical Harmonics to create 3-D parametric surface of the liver from Computed Tomography (CT) imaging system which may useful for Shape modeling, surface representation, physical measurement of objects and mathematical model. We select randomly 5 livers from more 100 dataset. Mean Hausdorff distance Errors and Dice similarity coefficient DSC by liver volumes was measured. The experimental results showed that the best order of Spherical Harmonics Expansion for livers is between 12~17.