Computational protocol: Circadian regulation of slow waves in human sleep: Topographical aspects

Similar protocols

Protocol publication

[…] After visual scoring, visual identification of EEG artefacts, and appropriate band-pass filtering (see below) of artefact free NREM sleep stages 2, 3 and 4 EEGs from 12 derivations, we applied algorithms based on the NumPy, SciPy and Matplotlib libraries for scientific computing () to quantify parameters of SWs such as incidence, amplitude, period, slope in accordance with previous studies (A–C). Channels were analysed independently.All EEG artefacts (muscle activity/sweating) for each individual EEG channel were visually identified by an experienced scorer and annotated on a three second basis using the EEG browser software Vitascore version 1.5 (Temec Instruments B.V., Kerkrade, The Netherlands). Thereafter all EEG channels were exported for further qEEG analyses. Segments annotated as artefacts were not used in the subsequent SW detection (for a flowchart of the analyses see ).All EEG artefacts (muscle activity/sweating) for each individual EEG channel were visually identified by an experienced scorer and annotated on a three second basis using the EEG browser software Vitascore version 1.5 (Temec Instruments B.V., Kerkrade, The Netherlands). Thereafter all EEG channels were exported for further qEEG analyses. Segments annotated as artefacts were not used in the subsequent SW detection (for a flowchart of the analyses see Inline Supplementary Fig. S1). Inline Supplementary Fig. S1 Fig. S1 Fig. S1 Inline Supplementary Fig. S1 can be found online at http://dx.doi.org/10.1016/j.neuroimage.2015.05.012.The EEG time series were decimated by a factor of two (keeping every second value) and band-pass filtered in the 0.5–4 Hz interval by a FIR window filter using a Blackman–Harris window of 2001 points (~ 15.6 s)(). The large width of the window, unlike second type Chebyshev filters, has the benefit of producing practically no side lobes in the frequency response of the filter while phase information remains intact (). The applied filter was chosen from a number of candidates based on the visual inspection of the filtered signal and its power spectrum.The main rationale for decimating data prior to filtering was the high computing demand due to the large taper window of the applied FIR filter. Since decimation affects only the highest frequencies (64–128 Hz) while those of interest are in the lowest region of the spectrum (0.5–4 Hz), the results should not be influenced by the procedure. Unless stated otherwise, all quantities introduced below are defined in the context of the filtered signal. Half-waves, i.e. segments enclosed between subsequent zero crossings (points A and B in C), with maxima (minima) values exceeding (below) + 5 μV (− 5 μV), below (exceeding) + 100 μV (− 100 μV), with frequencies in the 0.5–4 Hz interval, not overlapping with segments labelled as artefacts and not belonging to sleep stages other than NREM stages 2, 3 and 4 were retained and a number of properties were extracted following the procedure described previously (). The peak of the positive (negative) half-wave is defined as its highest (lowest) point or maximum (minimum) (P). We refer to the segment between the start (A) of the half-wave and its peak location (O) as the initial phase (AO), while the remaining segment (OB) as the final phase. Although we quantified the duration, the frequency of a half-wave can be defined as the inverse of its period, i.e., twice the duration of the half-wave. The mean initial slope is calculated as the half-wave maximum (minimum) divided by the duration of the initial phase (PO/OA). Similarly, the mean final slope refers to the ratio of the maximum (minimum) over the duration of the final phase (PO/BO). The maximum initial slope is defined as the largest absolute value of the signal's first derivate in the initial phase (at I). The maximum final slope is computed similarly from the final phase (at F). Further to these primary SW parameters, we also calculated the duration for the entire SW between the subsequent zero crossings (AB), the overall mean slope and maximum slope as averaged between the initial and final segment of the SW half-waves. The parameters of all individually detected SWs were collapsed into 20 min consecutive intervals between lights out and lights on (28 intervals in 9 h 20 min sleep). […]

Pipeline specifications

Software tools Numpy, matplotlib
Application Miscellaneous
Diseases Neurodegenerative Diseases