Fuzzy region integration approach for subcortical structure segmentation

An automated method for segmentation of subcortical structures in MR head scans is presented. Features extracted from the images, namely intensities of the structures and relative distances and directions, are processed in the fuzzy framework. To overcome the imprecision and uncertainty, the features are fuzzified and fused together to maximize detection accuracy. Experiments on an ensemble of MR head scans show promise of the current approach.