A brain tumor segmentation framework based on outlier detection*1

Dense deformation field estimation for atlas-based segmentation of pathological MR brain images

“…However, most of the state-of-the-art methods need to segment at least the lesion (excepting [7]). While this could be a limitation in some complex cases with infiltrating tumors or presence of edema, there exist automated methods that allow an accurate segmentation of a large range of lesions [16][17][18] (see [19] for a recent review of brain tumor segmentation algorithms). Voxel-based methods naturally avoid the problem of presegmenting functionally important brain structures (excepting the lesion) since they directly work on voxel intensities.…”

Detection and Segmentation of Pathological Structures by the Extended Graph-Shifts Algorithm

“…However, we note that a direct comparison is difficult due to different data, manual raters, and others. The Jaccard scores for Clark et al [4] are about 70%, for Prastawa et al [7] are 80% (both on very limited datasets with seven and three patients respectively), and Corso et al [5] is 85% (on training data with of five cases) The extended graph-shifts algorithm is also the fastest among these taking about a minute to perform the segmentation on each of these scans (preprocessing takes about five minutes). We show some examples in figure 5.…”

“…From the computer vision perspective, the task is to label regions of an image into pathological and non-pathological components. This is a special case of the well-known image segmentation problem which has a large literature in computer vision [1,2,3] and medical imaging [4,5,6,7,8,9,10] In previous work [11], we developed a hierarchical algorithm called graph-shifts which we applied to the task of segmenting sub-cortical structures formulated as energy function minimization. The algorithm does energy minimization by iteratively transforming the hierarchical graph representation.…”

“…Fletcher-Heath et al [6] take a fuzzy clustering approach to the segmentation followed by 3D connected components to build the tumor shape. Prastawa et al [7] present a detection/segmentation algorithm based on learning voxel-intensity distributions for normal brain matter and detecting outlier voxels, which are considered tumor.…”

Quick detection of brain tumors and edemas: A bounding box method using symmetry

“…Region growing [19] based tumor detection techniques suffer high time complexity. Statistical pattern recognition based methods [13], [21], [23] fall short, partly because large deformations occur in the intracranial tissues due to the growth of the tumor and edema. These methods detect abnormal regions using a registered brain atlas as a model for healthy brains.…”