How to fast and accurately reconstruct complex surfaces of organs has become increasingly important while the interslice distance is much larger than the interpixel distance. First

this paper presents a shape-keeping and adaptive feature points sampling algorithm

which can reduce storage and computational requirments in the post-processing procedure. Second

an algorithm for reconstructing surfaces from CT cross-sections based on shape feature point matching is presented

which takes into account four properties of tomographic images

i. e. intensity

gradient magnitude

gradient direction

and position disparity of the feature points to improve the accduracy of the reconstructed surface. The experimental results show the effectiveness of the above algorithms.