Multistage approach for automatic spleen segmentation in MRI sequences

Abstract

Most of the known methods of segmentation of the abdominal organs are not automated for the whole series of images or are semi-automatic and require additional intervention by the user. This is typical for cases where the difference in intensity of the grey level between the subject and the background is small. This paper presents a multistage approach for spleen segmentation from MRI-sequences. It is based on segmentation methods such as active contours without edges and k-mean clustering. The proposed approach consists of some basic stages. The first stage is pre-processing, based on image enhancement and morphological operation. Two atlas models are created, which are used in the initial image to define the initial contour at which the segmentation begins. The proposed approach allows extracting the spleen in the different depth images, which has a variable form and unstable position. The conducted experiments are showing the robustness of the proposed approach. The obtained results demonstrate the effectiveness of the approach for application in screening diagnostics.