A New Region Growing Medical Image Segmentation Algorithm Based on Interval Type-2 Fuzzy Sets

摘要

Segmentation of regions of interest plays important role in computer aided brain medical image diagnosis. Fuzzy techniques are widely used for this purpose as they can handle with imprecise or vague image information. The major achievement of this research is the introduction of a new region growing segmentation technique for which any fuzzy control model is possible to be applied. This enables the combination of physician knowledge, easily represented by fuzzy rules and therefore generalized, with the main concept of fuzzy control. Our approach is capable to segment cerebrospinal fluid in the cortical and subcortical areas of the brain. The study was performed by applying the segmentation method proposed on a dataset of 228 computed tomography scans of patients with diagnosed Alzheimer disease.