Poster F34, Tuesday, March 28, 8:00 – 10:00 am, Pacific Concourse
The Reliability of Brain State Properties
Derek M. Smith1, Yiran Zhao1, Behnaz Yousefi1, Shella D. Keilholz2, Eric H. Schumacher1; 1Georgia Institute of Technology, 2Emory University
Research employing dynamic functional connectivity methods supports the hypothesis that the composition of functional brain networks is dynamic and changes over time and information processing demands. However, before metrics of dynamic functional connectivity can be used to differentiate between individual abilities and task situations, the reliability of these measures must be better understood. In order to gain insights into the consistency of network dynamics, patterns of co-activation were obtained using K-means clustering of the images from the resting state scans of 100 unrelated Human Connectome Project subjects. The clustering procedure was applied separately to scans from two separate sessions. The primary goals were to assess: 1) the reliability of brain states across time; 2) the consistency of their frequency of occurrence; 3) and the consistency of their dynamics. Cluster centroids were similar between the two sessions. Frequency correlations between the first and second day reached values in the r = .5 range for most brain states. The general pattern of state transitions was similar between the two days as well. An increase in the frequency of occurrence across the course of the scan was found for two states on both days. In addition, associations between state properties and latent trait intelligence were explored. These findings highlight characteristics of brain states that are constant across time and demonstrated that dynamic connectivity measures can benefit differential research.
Topic Area: EXECUTIVE PROCESSES: Other