Improved Active Contour Model through Automatic Initialisation : Liver Segmentation

Liver segmentation from Computed Tomography (CT) images is a crucial step in computer aided diagnosis. In this paper, an efficient liver segmentation method based on a new active contour initialization is proposed. First, with Rayleigh distribution, the background noisy in CT image is estimated and removed. Then, Gaussian mixture model (GMM) is employed to compute the optimal threshold based on different values of component Gaussian distributions. To increase the segmentation accuracy, edge detection and morphological filter are employed to remove the minor regions. Finally, the obtained edge of the processed CT image is given as an initial edge map to the Gradient Vector Flow (GVF) model. The proposed method was evaluated on a public clinical database of 34 different people with normal liver and 51 people with abnormal liver with tumors.