| |
Flat Maps
22
tion (nodes and edges) about the connected representation of the gray matter. This representa-
tion is derived from the classification file.
The program used for segmenting is called mrGray. This program also allows checking of
the white matter surface topology before growing gray matter [Wandell, et al., 2000 #38]. In
addition, it can be used to display anatomical data and overlay, for example color-coded re-
gions of interest, in three-space. The 3D model can be rotated and various settings such as
lighting can be set.
3.7.4 Flattening
After a suitable representation of the gray matter has been calculated in three-space, this rep-
resentation can be flattened by using a flattening algorithm. Different algorithms can perform
this task. The program that is currently used at our lab is called mrFlatMesh. It creates ana-
tomical (cortical) flat maps. The result is a flat two-dimensional image. There are various
measures that the program can try to preserve, e.g., angles, areas, or interpoint distances. In
[Teo, et al., 1997 #8], the flattening algorithm calculates a flattened representation so that the
distances between pairs of points in the gray matter in three-space are as similar as possible to
their distances in the two-dimensional representation. The distances in three-space are calcu-
lated along paths within the gray matter.
[Wandell, et al., 2000 #38] describe a method that preserves neighborhood relationships;
no large twists in the flattened representation occur. The nodes of the created mesh in 3-space
can be placed in the plane in a single step by performing a single efficient linear calculation.
This makes the flattening algorithm much faster than an iterative approach. After the nodes
have been initially placed within the plane, their positions are iteratively adjusted so that dis-
tances between nodes in three-space match corresponding distances in the flat map as well as
possible (Figure 8). At this stage, the mesh topology is not changed anymore.
|  |
|
| |
|
|
