Automatic detection and labelling of the human cortical folds in magnetic resonance data sets

Abstract

The folding of the cortical surface of the human brain varies dramatically from person to person. However, the folding pattern is not arbitrary. The cortical folds (also called “sulci”) often serve as landmarks for referencing brain locations, and the most pronounced sulci have names that are well established in the neuroanatomical literature. In this paper, we will present a method that both automatically detects and attributes neuroanatomical names to these folds using image analysis methods applied to magnetic resonance data of human brains. More precisely, we subdivide each fold into a number of substructures which we call sulcal basins, and attach labels to these basins. These sulcal basins form a complete parcellation of the cortical surface.

The algorithm reported here is important in the context of human brain mapping. Human brain mapping aims at establishing correspondences between brain function and brain anatomy. One of the most intriguing problems in this field is the high inter-personal variability of human neuroanatomy which makes studies across many subjects very difficult. Most previous attempts at solving this problem are based on various methods of image registration where MR data sets of different subjects are warped until they overlap. We believe that in the process of warping too much of the individual anatomy is destroyed so that relevant information is lost. The approach presented in this paper allows inter-personal comparisons without having to resort to image warping. Our concept of sulcal basins allows to establish a complete parcellation of the cortical surface into separate regions. These regions are neuroanatomically meaningful and can be identified from MR data sets across many subjects. At the same time, the parcellation is detailed enough to be useful for brain mapping purposes.