Spatial registration of digital brain atlases based on fuzzy set theory

We present a semiautomatic method based on fuzzy set theory for adjusting a computerized brain atlas to magnetic resonance images (MRIs) of the human cerebral cortex. The atlas was registered to three-dimensional MRI data sets of 10 healthy volunteers. After a global matching using the external contour of the brain, several local procedures were performed regarding selected primary furrows and cytoarchitectonic areas. The final transformation matrix was calculated with respect to these anatomical structures and to their local matrices. Evaluation revealed an increase in accuracy as expressed by a reduction of the visible mismatch with respect to the registration of cortical and subcortical brain structures.