Stability of low-frequency fluctuation amplitudes in prolonged resting-state fMRI.

The (fractional) amplitudes of low-frequency fluctuations (f)ALFF are popular measures for the magnitude of low-frequency oscillations in resting-state fMRI (R-fMRI) data. Both measures can be directly derived from the spectral power of R-fMRI time courses. Numerous studies suggest that ALFF and fALFF might be used as biomarkers for a variety of diseases including schizophrenia, major depressive disorder, and obsessive-compulsive disorder. However, the temporal stability of (f)ALFF values, which is of great importance for the application of (f)ALFF both as a biomarker and scaling parameter, has not been studied in detail yet. Here, we quantify the temporal stability, robustness and reproducibility of both ALFF and fALFF maps obtained from R-fMRI data by performing statistical analyses over 55 minute resting-state scans which included a period of NaCl infusion. We also examine the differences of using either raw or standardised (f)ALFF maps. Our analyses show that no significant changes of (f)ALFF values over the 55minute period occur for both raw and standardised (f)ALFF maps. In addition, we demonstrate that raw (f)ALFF maps across subjects are correlated with head motion as quantified via frame-wise displacement, whereas no such correlation is present in standardised (f)ALFF maps. In conclusion, the results of our study show that both ALFF and fALFF qualify as potential biomarkers due to their high temporal stability.