Translator Disclaimer
1 July 1991 Morphological segmentation and 3-D rendering of the brain in magnetic resonance imaging
Author Affiliations +
The authors have developed a morphological technique, based on geodesic dilation using fast propagation of regional maxima, for segmenting the skin, bone, and dura from 3-D MR studies of the head, exposing the outer surface of the brain for 3-D rendering. The proposed algorithm for segmentation belongs to the class of connectivity segmentation techniques and uses morphological gray scale reconstruction and the distance function to discriminate connections by their width. By following only the connections wider than a critical dimension, the connectivity does not extend outside of the brain through nerves and other small paths connecting the brain to other tissues. On an IBM 6000 RISC workstation, the entire segmentation process takes less than three seconds per slice. With the volume segmented, we render the brain using a modification of the rendering process proposed by Michael Bomans and Karl-Heinz Hohn for visualizing poorly defined surfaces such as the sulci of the brain. The modification uses a depth map transformation that permits replacing the compute-intensive 3-D closing with the rolling ball applied in 2-D. This method also eliminates the need to maintain two separate volumes for the rendering process.
© (1991) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
William H. Connor and Pedro J. Diaz "Morphological segmentation and 3-D rendering of the brain in magnetic resonance imaging", Proc. SPIE 1568, Image Algebra and Morphological Image Processing II, (1 July 1991);

Back to Top