A method for segmenting an image comprising: determining an initial estimation of a boundary between at least two components in a region of an image to be segmented; providing image data from the region of the image to be segmented, the image data representing gray level values of a plurality of image elements of the components; modelling the image data on a mixture of at least two statistical distributions, each statistical distribution having more than one parameter, and each component being associated with certain weights of the statistical distributions in the mixture; estimating the parameters of the statistical distributions in the mixture; for each component, estimating the weights of the statistical distributions in the mixture based on the estimated parameters and the image data of each image element; and optimizing the initial estimation of the boundary between the components based on the estimated parameters and estimated weights.