Automatic liver segmentation method based on improved region growing algorithm

It is difficult to select seed points for the region growing algorithm and the problem of irregular liver shape. This paper proposes an automatic liver segmentation method with the centroid as the seed point of the region growing algorithm. The adaptive median filtering in order to noise reduction and binarization are performed on the image, and the largest connected region of the CT image is locked as the initial contour of the liver. The centroid of the largest connected domain was obtained as the seed point location for regional growth, and the image was segmented using the dual-threshold regional growth method. The experimental results show, This method replaces the traditional method of manually selecting seed points, and solves the problem of manually selecting seed points for the region growing algorithm; At the same time, the use of double-threshold segmentation improves the accuracy of liver region segmentation, makes segmentation more accurate, and the edges and texture parts are smoother and sharper.