Publication Date




Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Biomedical Engineering (Engineering)

Date of Defense


First Committee Member

Jorge E. Bohórquez

Second Committee Member

Özcan Özdamar

Third Committee Member

Richard R. McNeer II

Fourth Committee Member

Suhrud M. Rajguru

Fifth Committee Member

Roman Dudaryk


Background: There is a need to expand the research on intraoperative awareness to better understand the patient’s neurologic responses and effectively predict and prevent unintended awareness. Previous studies in altered levels of consciousness in anesthesia and sleep have only focused on either electroencephalogram (EEG) or evoked potentials (EP) analysis. Additionally, EP studies only focused on either transient or steady-state response. Purpose: To investigate the inherent patterns associated with altered levels of consciousness in both anesthesia and sleep models as extracted from the simultaneous acquisition of both multi-rate auditory EPs and EEG spectral descriptors. Methods: Anesthesia recordings were acquired from five patients during elective surgery while sleep recordings were acquired from fourteen healthy volunteers. For anesthesia, a multi-rate stimulation paradigm composed of (a) 5 Hz isochronic (b) 20, 30, and 40 Hz short, low-jittered, and (c) silence regions were used. The sleep study stimulus consisted of (a) 5 Hz isochronic, (b) 30 Hz low-jittered, and (c) silence regions. In both anesthesia and sleep, the click stimulus was 200 μsec in duration, and it was delivered to the right ear at 70 dB nHL and 60 dB nHL, respectively. Physiological data such as spontaneous respiration rate, expired agent concentration, bispectral index, heart rate, and core body temperature, were collected for all anesthesia patients. In anesthesia, EEG was continuously acquired from Fz-M2 channel, while in sleep it was acquired from Fz-A2 and Cz-A2 with additional channels for scoring. These EEG signals were analyzed using the full set of spectral descriptors. Trend analysis for EEG attributes and moving average for EP attributes were used for the anesthesia study while for the sleep study these attributes were investigated by contrasting between the sleep stages. Two independent investigators assigned the sleep stages to EEG segments with a manual computer-aided sleep scoring software, which was specifically developed for the sleep study. To extract both high-rate transient and steady-state responses the continuous loop averaging deconvolution (CLAD) method was used. The 5 Hz EPs responses were processed with broadband filtering, starting at 1Hz. Findings and Conclusions: Spectral descriptors, extracted from EEG, showed that delta power ratio and median frequency showed significant separation between levels in both anesthesia and sleep. In the separation of anesthesia levels, the aforementioned features exhibited better performance than both spectral entropy (SE) and bispectral index (BIS). Channel Fz-A2 compared to Cz-A2 showed better discrimination between sleep stages. EPs elicited by low rate (5 Hz) stimulation, when acquired using low frequency preserving filters such as filters with broad frequency bands, showed greater sensitivity in separating anesthesia levels and in differentiating sleep stages. Future works should include an exhaustive study of delta power ratio and median frequency for EEG and low rate responses that consider low frequencies for EP.


anesthesia; BIS; sleep; EEG; monitoring; evoked potential