Characterization of Infant EEG Dynamics During Anesthesia

Preprocessing was carried out with Natus NeuroWorks (Natus Medical Inc.) and in-built MATLAB code (MathWorks Inc., Natick, MA). Ear electrodes (M1 and M2) were excluded from the final analysis due to poor surface-to-skin contact for the majority of infants.
EEG signals were re-montaged to a nearest neighbor Laplacian reference using distances along the scalp surface to weight neighboring electrode contributions. We applied an anti-aliasing filter of 80 Hz and down-sampled the EEG data to 256 Hz.
EEG dynamics were analyzed at three distinct periods of (1) MOSSA, (2) awake (pre-anesthesia), and (3) emergence from general anesthesia. (1) MOSSA analysis: in the first part of this study, we were interested in characterizing EEG features during MOSSA. For each infant, a 10-min EEG segment during a period of maintenance of general anesthesia adequate for surgery and where end-tidal sevoflurane concentration was maintained at a constant concentration (+0.1%) was identified. Within this time segment, a 5-min epoch was selected from ‘artifact-free’ EEG, where motion or electrocautery artifacts were not present in the EEG. Channels with noise or artifacts were excluded from the analysis by visual inspection. EEG data analyzed during MOSSA involved epochs with median time after propofol or nitrous oxide administration of 77 min (IQR 32–93 min). (2) Awake state to MOSSA analysis: in the second part of this study, we were interested in characterizing the relationship with EEG features during the awake state, and comparing them to MOSSA. For each infant, the EEG and video recording starting before any anesthetics were administered was identified. Each video was reviewed and the sleep state of the infant identified using behavioral and vocal markers. The awake state was defined as having one of three features (1) eyes open, (2) body movement, and/or (3) crying. For each corresponding EEG segment, we extracted an 11-s ‘artifact-free’ epoch by visual inspection. If there was no ‘artifact-free’ data available then we did not include that epoch in the analysis. (3) Emergence from general anesthesia analysis: in the third part of this study, we were interested in characterizing relationship with EEG features, end-tidal sevoflurane concentration and onset of body movement. For each infant, an EEG segment starting from 10 min before sevoflurane gas was turned off, and finishing when end-tidal sevoflurane concentration was 0% was identified. For each EEG segment, we extracted a 30-s ‘artifact-free’ epoch by visual inspection at 10 end-tidal sevoflurane concentration ranges: (1) 0 to <0.3%; (2) 0.3 to <0.6%; (3) 0.6 to <0.9%; (4) 0.9 to <1.2%; (5) 1.2 to <1.5%; (6) 1.5 to <1.8%; (7) 1.8 to <2.1%; (8) 2.1 to <2.4%; (9) 2.4 to <2.7%; and (10) 2.7 to <3.0%. A 15- or 20-s epoch was used in three occasions. If there was no ‘artifact-free’ data available in a specific concentration range then we did not include that epoch in the analysis.