| |
Summary
190
deformed by using estimated displacement fields, an energy function evaluates the quality of
the deformation. A good deformation is expected to yield a good fit between atlas and corre-
sponding measured image. By minimizing the energy, the fit can be improved. An optimiza-
tion technique does this. Once the atlas matches the measured image well enough, the visual
areas in the measured image can be inferred from the atlas annotations. The user interface of
mrFindBorders allows easily evaluating and modifying the results. Besides the identification
task, other results produced by mrFindBorders can be applied in the study of cortical magnifi-
cation. For example, it is possible to match images based on a ring visual stimulus to create
iso-eccentricity lines.
The program has been tested with various data sets. In many cases, the results were very
similar to those created by an experienced human expert. However, there are still many im-
provements possible and the tool can be extended without difficulty. For example, once the
quality of flat maps and/or the calculation of visual field sign maps have been improved, vis-
ual field sign maps can easily be used in the registration process. It is also straightforwardly
possible to plug in other registration algorithms. A fast elastic matching algorithm based on
linear elasticity theory has already proofed to work in principle.
|  |
|
| |
|
|
