A Novel Multiresolution Fuzzy Segmentation Method on MR Image

Multiresolution-based magneticresonance (MR) image segmentation has attracted attention for itsability to capture rich information across scales compared with theconventional segmentation methods. In this paper, a newscale-space-based segmentation model is presented, where both theintra-scale andinter-scale properties are considered and formulated as two fuzzy energyfunctions. Meanwhile, a control parameter is introduced to adjust thecontribution of the similarity character across scales and theclustering character within the scale. By minimizing the combinedinter/intra energy function, the multiresolution fuzzy segmentationalgorithm is derived. Then the coarse to fine leading segmentation isperformed automatically and iteratively on a set of multiresolutionimages. The validity of the proposed algorithm is demonstrated by thetest image and pathological MR images. Experiments show that by thisapproach the segmentation results, especially in the tumor areadelineation, are more precise than those of the conventional fuzzysegmentation methods.