Graphics: Visual and Interactive Computing
Contour Connection
|
|
The Euclidean distance field can give an alternative approach to the calculation of an object from contour slices. We enclose the slices within a volume data set, and then decide whether each voxel is inside or outside the contours within each slice. We also calculate the distance from every voxel to the closest point on the set of contours. In theory isosurfacing the data set for the level 0 set of points should give you a surface which when intersected by the original planes will give the original contours. As can be seen from the images on the right, those surfaces are quite good representations for a possible representation of the surface that could have been described by those contours. This work was as a result of the requirement to reconstruct a torso and lungs from an MRI scan which only had 17 slices (torso) and 12 slices (lungs) available. The bottom two images show the resulting surfaces produced for those contours. The top images show how the method treats some of the 'difficult' branching cases - 1 to 2, 1 to 2 to 3 (note the hole), 3 to 5, 1 to 4, and S to a circle, and a 1 to 2 branch with a jagged interface. This method has since been used to create tetrahedral models from MRI scans for numerical simulation of blood flow. Main Reference
|
