In this paper, we proposed a mathod for lumbar vertebra CT image segmentation based on the contourlet transform and artificial neural networks(ANNs). The proposed method consists of three portions. In the first part, contourlet transform is used to decompose the CT image to obtain the contourlet coefficients. In the second part, the self-organizing competitive artificial neural network is employed to optimize and extract the low frequency coefficients coefficients of contourlet transformation, reduce the number of coefficients greatly. The last part, the optimized coefficients are inverse contourlet transformed with the original coefficients,the segmented image is reconstructed. The experimental results show the accuracy of human lumbar vertebra CT image segmentation based on the proposed method is encouraged.