Medical Image Segmentation Based on Wavelet Transformation and IGGVF

Medical images often have low contract and SNR and adoption traditional image segmentation algorithms usually can not get satisfying results. In this paper, we propose a new algorithm based on wavelet transformation and the improved GGVF (IGGVF) for their segmentation. Firstly, wavelet transformation is carried out on the original medical image to get multi-scale reconstructed approximate images. Next a new initial setting method is employed for gaining the initial contour then it is deformed according to the IGGVF snake model to attain the ultimately rough contour in the largest reconstructed image. Afterwards, this contour is considered as the initial contour and continues to be deformed in smaller scale reconstructed image. Good experimental performance on medical image reveals that it is more robust to noise and can segment medical images very accurately.