3D Geometry Feature Bilateral Smoothing For Segmenttation Of Brain Model

With the rapid developing of the economy and medical standard, biomedical images technology has become an important research and clinical development direction. Therefore, biomedical images stored in digital form are growing quickly. The human brain is one of the most important parts of human, and the cortical surface includes a lot of complex and visual information contents, just because of the complexity, manual methods have been unable to make accurate, objective and rapid analysis and processing, so there needs to be an effective and automatic method and system to analyze and process biomedical images. The human cortical surface reconstructed from three-dimensional brain magnetic resonance images is one object of our study. This paper mainly focuses on the sulci and gyri segmentation and the sulcal ravines extraction of cerebral cortex surface.The main contributions of this paper are summarized as follows:1. Based on the idea of bilateral filtering, we proposed a bilateral smoothing method for the principal curvatures and the principal directions on the cortical surface. Firstly, estimating the principal curvatures and principal directions on the surface, however, due to the complexity of the surface structure of the cerebral cortex, there are large distortions on the estimated principal curvatures and directions. So, we need to smooth these geometry features, the main purpose is to make distributions of the principal curvatures and directions of the surface seem more reliable. Experimental results show that the proposed bilateral method can effectively smooth these geometry features.2. Based on the smoothed principal curvatures of the cortical surface, we segment sulcal regions and gyral regions of the cortical surface. Here we successfully applied the graph-cut image segmentation method to the three-dimensional surface of the cerebral cortex. Experimental results show that the graph-cut method is very effective and fast to the sulcal and gyral regions segmentation of three-dimensional cortical surface.3. This paper presented a hill climbing method based on the maximum principal curvatures to segment the sulcal basins on the cortical surface. According to distribution patterns of the maximum principal curvatures changing from small to large, from negative to positive, this paper proposed a maximum principal curvature hill climbing method for the sulcal basins segmentation of the cortical surface. Experimental results proved the validity and accuracy of this method.4. A sucal ravines extraction method based on the maximum principal curvatures and the maximum principal directions of the cortical surface is proposed. This method is simple, fast and effective, and it can extract a complete sulcal ravine for each sulcal region. These extracted sulcal ravines successfully reflect the topology of the cortical surface. Experimental results show that this method for the extraction of sulcal ravines is very effective and stability.