SYSTEM AND METHOD FOR INTENSITY INHOMOGENEOUS IMAGE SEGMENTATION

摘要

The present disclosure relates to a computerized method for segmenting an intensity inhomogeneous image. The method comprises: identifying a region of interest (ROI) to be segmented in the image; determining a polynomial representation of the intensity of the ROI using a set of polynomials; determining a sparse shape representation of the ROI based on a sparse combination of a set of training shapes associated with the ROI; generating a segmentation model from the polynomial representation and the sparse shape representation; and minimizing an energy functional of the segmentation model for segmenting the ROI in the image. The segmentation model may include contour deformation, wherein the ROI in the polynomial representation is deformed according to the sparse shape representation.