Figure 1
Schematic of MEG data preprocessing and analyses. (A–D) Standardized resting-state MEG data preprocessing procedures. (E) Visual selection and removal of time segments with remaining artefacts (red box), drowsiness or eyes-open. (F) Subsequent analyses of clean source-space MEG time series were restricted to the parieto-occipital cortex (Supplementary Table S1) and the left and right hippocampi. (G) Power spectra (1–48 Hz) were obtained by means of a Fast Fourier Transform. (H) Long-range temporal correlations in the extended alpha band (6–13 Hz) were estimated using the DFA algorithm. To obtain the DFA exponent, the algorithm fits a slope, α, to the mean fluctuations of the signal profile (y-axis) across log-spaced time windows (x-axis). (I) The fE/I algorithm correlates the average amplitude alpha band oscillations (x-axis) to the amplitude-normalized fluctuation functions (nF(t), y-axis) calculated in 5-s windows. (J) The aperiodic exponent of the power spectrum in the gamma frequencies (30–48 Hz) was obtained using the FOOOF algorithm. (K) Hypotheses for E–I (im)balance in AD patients. (L–N) Hypothesized DFA exponents, fE/I values and aperiodic exponents in healthy (black), excitation-dominated (red) and inhibition-dominated (blue) states. MEG Magnetoencephalography, MRI magnetic resonance imaging, tSSS temporal extension of signal space separation, ROIs regions of interest, AAL-atlas automated anatomical labeling atlas, L + R left and right hemisphere, DFA detrended fluctuation analyses, fE/I functional excitation/inhibition, HE healthy elderly controls, SCD subjective cognitive decline, MCI mild cognitive impairment, AD dementia due to Alzheimer’s disease.
