Cortical Flattening Applied to High-Resolution ¹⁸F-FDG PET

Abstract

Group studies using PET and other types of neuroimaging require some means to achieve congruence of brain structures across subjects, such that scans from individuals varying in brain shape and gyral anatomy can be analyzed together. Volume registration methods are the most widely used approach to achieve this congruence. They are fast and typically require little manual interaction, but, unfortunately, it is difficult to achieve a good match between cortical areas in volume space, especially where folding patterns vary across subjects. Cortical flattening is a recent, alternative strategy: Its key features are explicit definition of cortex, such that white matter or cerebrospinal fluid compartments are largely excluded from the analysis volume, and subsequent registration of the cortical sheet in its natural, 2-dimensional topology. This type of registration has been demonstrated to provide better matching of congruent cortical structures than volume methods and, thus, offers a potentially more robust way of analyzing PET data. Methods: Here, we explore the applicability of cortical flattening of coregistered MRI to 18 F-FDG PET on the HRRT system (high-resolution research tomograph), the highest-resolution whole-head scanner available to date. Results: We report average values and SD of cortical metabolism in a pilot study of the dominant hemisphere in 9 control subjects and provide estimates of group sizes necessary for studies using this technique. Conclusion: We conclude that cortical flattening with subsequent surface registration is a feasible and promising strategy for group studies on the HRRT, providing the highest fidelity maps of human cortical glucose consumption to date.