In this paper, a new anatomic contour extraction algorithm based on the contourlet transform and pulse-coupled neural network (PCNN) was presented. The contourlet transform is applied on the vertebra CT image firstly, using its multiscale and multidirectional characteristics to get the coefficients which can effectively capture the contours of an image and depict the contours sparsely. Then PCNN is utilized to extract contours from the contourlet approximation image. Finally, the contours of image are reconstructed by inverse contourlet transform. Results: The proposed algorithm is tested on vertebra CT images, and we get some encouraging results, which are better than those based on wavelet. Conclusions: In this paper, we extract the contours of vertebra correctly, which effectively represent multiscale and multidirectional singularities of an image. This proposed method is very significant for the clinical applications and medical researches. © (2012) Trans Tech Publications.