Single Sideband Modulation As A Tool To Improve Functional Connectivity Estimation

Time-resolved functional network connectivity (trFNC) provides a useful tool for representing functional magnetic resonance imaging (fMRI) data with functional networks that change with time. Partly due to its simplicity, sliding window Pearson correlation (SWPC) is the most widely-used method for trFNC estimation. In SWPC, the window size should be selected long enough to avoid spurious estimates of connectivity values, and short enough to capture meaningful fast variations in connectivity estimates. To solve this issue, we propose a method inspired by single sideband (SSB) modulation that allows us to select small window sizes for SWPC without filtering out important low-frequency activity information. We use simulation to show the improvement offered by the proposed method. Additionally, we use fMRI data to show that SSB-SWPC estimates have reduced spurious variation compared with typical SWPC estimators.