| |
160
7 Results
7.1 Overview
The visual areas identified by mrFindBorders are comparable to those segmented by an ex-
perienced human expert. It has been shown by using both artificial and measured data that the
algorithm is accurate and robust. Especially for area V1, which is often the only area that re-
searchers need to identify, the results are usually very good. However, it must be acknowl-
edged that it is not possible to deal automatically with all data sets as well as a human expert
at this time. This is especially true for data with high levels of noise or irregular pattern that
cannot be found in the atlas image.
Since a general approach for fitting an atlas to a measured fMRI image has been devel-
oped, mrFindBorders may become capable of identifying other functional brain areas in the
future, especially if the quality of flat maps keeps improving. Beyond identifying visual areas,
the deformed atlases can be used to create iso-eccentricity and iso-angle lines, which is impor-
tant for studies of cortical magnification.
Some results are given in this chapter. At first, an artificially created reference image was
used to test the algorithm. Then, two measured phase-wedge images were presented to the al-
gorithm.
|  |
|
| |
|
|
