Uncovering the locus coeruleus: Comparison of localization methods for functional analysis.

Functional neuroimaging of small brainstem structures in humans is gaining interest due to their potential importance in aging and many clinical conditions. Researchers have used different methods to measure activity in the locus coeruleus (LC), the main noradrenergic nucleus in the brain. However, the extent to which these different LC localization methods yield similar results is unclear. In the present article, we compared four different approaches to estimate localization of the LC in a large sample (N = 98): 1) a probabilistic map from a previous study, 2) masks segmented from neuromelanin-sensitive scans, both manually and semi-automatically, 3) components from a masked-independent components analysis of the functional data, and 4) a mask from pupil regression of the functional data. The four methods have all been used previously in the imaging community to localize the LC in vivo in humans. We report several measures of similarity between the LC masks obtained from the different methods. In addition, we compare functional connectivity maps obtained from the different masks. We conclude that sample-specific masks appear more suitable than masks obtained from an independent sample, that masks based on structural versus functional methods may capture different portions of LC, and that, at the group level, the creation of a "consensus" mask using more than one approach may give a better estimate of LC localization.