Temporal Stability of Resting State fMRI Data Analysis by Independent Components Method

This paper is devoted to the analysis of the temporal stability of the independent components obtained by analyzing data of resting state functional Magnetic Resonance Imaging (fMRI). We analyzed 25 datasets of healthy volunteers, consisting of 1000 time samples each. The fMRI data recording time was 33.3 min for each volunteer. This approach made it possible to divide the experimental session into several time ranges to assess the temporal stability of the results obtained with the independent component analysis (ICA). During the analysis, the property of additivity of independent components was discovered: the dynamics of the independent components obtained in the analysis of individual time ranges have a high level of Pearson correlation (at least 0.9) with the dynamics of the independent components obtained in the analysis of the full experimental session, i.e., the result of ICA is robust to the choice of window size when analyzing a representative data sample. It was also shown that the time series of independent components, which topology corresponds to resting state networks, have a correlation with the global signal at the level of 0.4–0.5.