Shella Keilholz1, Alessio Medda2, Lukas Hoffmann3, Matthew E. Magnuson1, Garth Thompson1, Wen-Ju Pan1
1Biomedical Engineering, Emory/Georgia Tech, Atlanta, GA, United States; 2Georgia Tech Research Institute, Atlanta, GA, United States; 3Neuroscience Program, Emory University, Atlanta, GA, United States
While functional connectivity has typically been calculated over the length of an entire scan, interest has been growing in dynamic analysis methods that can detect changes in connectivity on much shorter time scales. Dynamic connectivity can be examined using sliding window correlation, but the properties of the dynamics depend on the window length, making a data-driven approach more attractive. We have developed an algorithm based on wavelet decomposition that clusters voxels into groups with similar temporal and spectral properties. The resulting clusters agree well with anatomy in the rat and the wavelet decomposition features exhibit sensitivity to network dynamics.