Automatic Brain Tumor Segmentation from MRI Images using Superpixels based Split and Merge Algorithm

Brain tumor segmentation, an essential but challenging task, has long attracted much attention from the medical imaging community. Recently, successful applications of sparse coding and dictionary learning has emerged in various vision problems including image segmentation. In this paper, a superpixel-based framework for automated brain tumor segmentation is introduced. A reformulation of the processes that composed the split and merge technique is proposed. A recursive split process is done first, then, after the merge process, a segmentation of the image is obtained. This is possible because the merging is implemented like a growth process, so the grouping has been eliminated. The use of a complete quad tree helps to achieve the reformulation. The determination of the region uniformity in each process is carried out with a different predicate, because of the different nature of the two processes. Experiments on a blocks world image and an industrial parts image are included to show the result of applying the algorithm.