On navigating the human cerebral cortex: response to 'in praise of tedious anatomy'

Abstract

Individual variability of the human cerebral cortex is a source of both fascination and frustration. The fascination arises because variability in cortical structure and function may account for many aspects of our unique personalities and cognitive capabilities. For neuroimagers, the frustration arises because variability presents serious obstacles when attempting to assign particular functional activation patterns to specific cortical areas. Devlin and Poldrack cogently summarize many of the key issues, and they make useful suggestions for linking function to anatomy using a standardized stereotaxic space. This commentary provides a broader perspective on the nature of individual variability that has implications for the choice of strategies used to compensate for variability. It also includes information about the actual differences between various registration strategies and introduces a new strategy for converting neuroimaging data to a standard stereotaxic space.

Figure 1

Quantitative maps of the differences between stereotaxic spaces and registration processes. A, B. Lateral and medial view of the distance between corresponding points on average fiducial surfaces in MNI space (FLIRT algorithm, MNI152 template) and T & T space (AFNI implementation of piecewise linear algorithm), displayed on the T & T average fiducial surface. Each average fiducial surface is based on the shapes of 12 normal individuals hemispheres (Van Essen, 2005). C, D. Lateral and medial views of an analogous difference map for the FLIRT and SPM2 average fiducial surfaces, both using the MNI152 template. A more comprehensive set of comparisons is available in the SumsDB database and can be viewed online using WebCaret (see http://sumsdb.wustl.edu/sums/directory.do?id=6604123&dir_name=COMPARE_STEREOTAXIC_SPACES

Figure 2A,B

Data from 312 studies (8,559 foci) mapped to the PALS-B12 atlas. Using a one-step mapping option in Caret, each focus was projected to the appropriate PALS-B12 atlas surface. Since each focus preserves its original distance from the atlas surface, their new coordinates reflect an accurate transformation from many starting spaces into a single output space. Only foci within 10 mm from the atlas cerebral surface are displayed in this figure; this excludes a number of published activations centered in the cerebellum. C. Foci from a set of tasks related to imitation and action observation overlaid on a probabilistic map of area 44 (Amunts et al., 1999) registered by the linear method. Many of the data were obtained from the Functional Imaging Net Database (FIND; Hamilton and Van Horn, 2006) and BREDE (Nielsen, 2003) databases. Data are accessible at http://sumsdb.wustl.edu/sums/directory.do?id=6529195&dir_name=STEREOTAXIC_FOCI