Organ Segmentation Using Atlas Registration and Fuzzy Connectedness

In this paper, a framework combining the atlas registration and the fuzzy connectedness for the automatic segmentation of abdominal organs is proposed. First, a pre-labeled atlas is registered onto the subject to establish the proper correspondence between the atlas and subject. Second, we utilize the fuzzy connectedness framework to segment organs of interest from the subject image. Taking advantage of the correspondences between atlas and subject, common interactive or empirical specification of fuzzy connectedness parameters are replaced by automatic and subject-oriented specification. The performance of the proposed method has been qualitatively validated on CT and MRI images with manual segmentation as ground truth. Results have shown the versatility of this method in segmenting images of different modalities. This automatic method is expected to find wide applications in 3D visualization, radiation treatments planning, and medical database construction