Example of Extracted Extremal Lines and Points

In Fig. 4, we can see an example of the lines extracted automatically (with a manual choice of the isointensity threshold) in a CT image of a dry skull. Some of the 550 crest lines may be recognized as anatomical landmarks, such as the orbits or the inferior border of the mandible. The lines are colored by the sign of the e2 extremality. Thus, extremal points are located at the color changes along the lines. There are around 3000 such extremal points.

In an MR image, the surface of the brain is not very well defined by an isointensity of the image. A presegmentation step is usually needed to isolate the brain from the rest of the data. This can be done with a combination of mathematical morphological operators, filtering, and the search for connected parts or with an automatic "surface edge" extractor, such as the zero-crossing of the image Laplacian. In Fig. 5, we used a segmentation of the surface of the brain and extracted the crest lines on this surface. Lines in red (with a positive largest curvature) roughly correspond to sulci, whereas blue lines (with a negative largest curvature) could be interpreted as gyri.

FIGURE 4 (Left) An axial slice of a 3D CT image of a dry skull in Plexiglas. (Middle and right) The crest lines extracted on this image. The isointensity was manually chosen to delimit the skull. Original 3D image courtesy of GE-CGR, Buc, France. See also Plate 68.
FIGURE 5 (Left) A sagittal slice of a 3D MR image. (Middle and right) 3D views of the extremal lines extracted superimposed on the surface of the brain. Original MR images and segmentation courtesy of Prof. R. Kikinis, Brigham and Women's Hospital, Boston. See also Plate 69.
0 0

Post a comment