The Research And Realization On Brain Extraction Of T1 Weighted MR Volumetric Data

At present, research on human brain has become one of the scientific hotspots because scientists can widely study human brain without human damages owing to the medical imaging's improvement. There are many applications (for example brain visualization, brain diagnose etc) related to brain imaging which either require, or benefit from, the ability to accurately segment brain from brain image which includes lots of non-brain tissue, such as skull, scalp, eyeballs. Therefore, brain extraction is the key of the human brain research, and the foundation of the following brain research.For the request of fast and precision in brain extraction, a strategy is put forward in this paper that brain is extracted twice automated.First brain extraction method is based on deformable surface models according to global understanding of the image after compared with other medical image segmentation methods. Firstly the difference of intensity of different tissue is enlarged through intensity histogram processing according to brain anatomic and imaging knowledge, and the distribution of brain tissue in intensity histogram. Secondly the brain surface is modeled by a surface tessellation using connected triangles and the surface model is allowed to slowly deform, one vertex at a time, following forces that keep the surface well-spaced and smooth, whilst attempting to move towards the brain's edge. Thirdly the vertex is stopped by triangle self-intersecting. Finally brain extraction is finished by region growing.Second brain extraction method is based on mathematical morphology according to semi-global understanding of the image, with the object of voxel. For the character of T1 weighted MR volumetric data, a 3D structure element ----6 neighbors is applied. Firstly volumetric data is made into different regions. Then different regions are eroded into different degree. Finally mathematical morphology achieves a more precision brain extraction. At the same time mathematical morphology is made use of in 3D image segmentation.Brian extraction is accomplished. Then brain extraction methods are evaluated and the parameters are estimated by the evaluation law of images segmentation. It is proved that brain extraction based on deformable surface models has precision more than 92%, and brain extraction of mathematical morphology has precision more than 94%,after 7 sets of T1 weighted MR volumetric data are experimented for brain extraction.