Region merging in medical image segmentation and interpretation

A u t o m a t e d segmentation and interpretation of medical images is o f t en a difficult task due to the complexi ty of the image data. Using cu r ren t reg ion growing segmentation techniques, t h e image is almost always oversegmented or undersegmented because the region growing processes are mostly based on region homogeneity properties. Seman t i c region growing approaches often s t a r t w i t h an oversegmented image and use U priori knowledge to merge regions in objects. We describe here a region merging method that is useful i n o u r s eman t i c genetic image segmenta t ion and in t e rp re t a t ion method.