Level-set evolution with region competition: automatic 3-D segmentation of brain tumors

Abstract

Abstract-This paper discusses the development of a new method for the automatic segmentation of anatomical structures from volumetric medical images. Driving application is the segmentation of 3-D tumor structures from magnetic resonance images (MRI), which is known to be a very challenging segmentation problem due to the variability of tumor geometry and intensity patterns. Level set evolution combining global smoothness with the flexibility of topology changes offers significant advantages over conventional statistical classification followed by mathematical morphology. Level set evolution with constant propagation needs to be initialized either completely inside or outside and can leak through weak or missing boundary parts. Replacing the constant propagation term by a signed local statistical force overcomes these limitations and results in a region competition method that converges to a stable solution.Applied to MR images presenting tumors, probabilities for background and tumor regions are calculated from a preand post-contrast difference image and mixture-modelling fit of the histogram. The whole image is used for initialization of the level set evolution to segment the blobby-shaped tumor boundaries. Preliminary results on five cases presenting different tumors with significant shape and intensity variability demonstrate that the new method might become a powerful and efficient tool for the clinic. Validity is demonstrated by comparison with manual expert segmentation.

Automatic brain tumor segmentation by subject specific modification of atlas priors1

“…The first set of manual segmentations is chosen as a gold standard for the purpose of evaluating the method. As another measure of performance, we also compared the manual segmentation results with the segmentations generated by a semi-automatic segmentation tool that uses levelset evolution (19). The semi-automatic segmentations are generated from the T1 contrast difference image with the user specifying the rough estimate of the tumor location by placing bubbles in the 3D volume.…”

“…It is therefore necessary to impose a shape constraint on the tumor prior. To remove the thin, sharp features of the blood vessels we apply a region competition level-set evolution method (19). This constraint forces the structure within the tumor prior probabilities to be relatively smooth and blobby ( Figure 6).…”

Effects of Healthy Aging Measured By Intracranial Compartment Volumes Using a Designed MR Brain Database

“…The lateral ventricular CSF was segmented using 3D snakes with region competition priors [15]. The snake was initialized near the ventricles and evolved based on the probability map for CSF, provided as input to the program.…”

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.…”