2.3 Band-power analysis

A two second sliding window with 50% overlap was used to extract absolute spectral power in five different frequency bands (delta: 1-4Hz, theta: 4–8 Hz, alpha: 8–13 Hz, beta: 13–30 Hz, and gamma: 30-70Hz) across each channel. The mean absolute powers (and the standard deviations) in each sliding window were averaged across windows for each frequency band, channel and patient. This gives five features per channel across three different states (before, during, and after) for each patient. This is visually summarised in Fig 1A–1C. The spatial average time series (average over all the channels) in broadband (1-70Hz) has also been plotted in Fig 2 across three states to visualise the effect of carbogen over time. The data was smoothed using a moving median filter (n = 31), to minimize the fluctuation of the signal for visualisation purpose, however, all the statistical calculations were carried out on original non-smoothed signal (S1 Fig).

(A, D, G, I, K) Effect sizes (Cohen’s d) of the band power change of “before-during” and (B, E, H, J, L) “before-after” across five frequency bands. (C, F) Time series of the average band-power in broadband (1-70Hz) for Patient 3 and Patient 5 respectively. Positive effect size denotes decrement in absolute band power. Topographical maps are ordered as delta, theta, alpha, beta and gamma. The frames surrounding topographical maps signify difference in the respective states they are representing. For instance, Fig 2C has yellow, red, and blue colours over laid across before, during, and after epochs respectively and topographical maps in Fig 2A has “yellow-red” colour frame representing the effect size between “before-during” states.