Objective response detection technique in frequency-domain for reflecting changes in MLAEP

Abstract

The mid-latency auditory evoked potential (MLAEP) occurs within the interval 10–100 ms post-stimulus and exhibits morphological changes due to coma, general anaesthesia and neurological and psychiatric diseases. This work evaluates an objective response detection technique in the frequency domain, the magnitude squared coherence ( κ ˆ 2 ( f ) ) , in reflecting MLAEPs changes in amplitude and latencies. Both simulation and EEG data were used in this investigation. The EEG during auditory stimulation in 10 volunteers under propofol-induced anaesthesia was acquired. Based on the null hypothesis of non-existing response, critical values have been derived in order to detect the auditory response. The alternative hypothesis was used to infer about maximal-response frequencies as well as to derive confidence intervals for the estimated values of κ ˆ 2 ( f ) during anaesthesia. Changes in both simulated and real MLAEPs could be tracked by κ ˆ 2 ( f ) , even for low SNR. High degree of cortical activation occurred predominantly between 30 and 90 Hz. This finding could be related to the sensory-originated phase-locked oscillation in the gamma band (30–70 Hz). Hence, for applications where changes on MLAEPs should be assessed, the use of κ ˆ 2 ( f ) as an objective statistical response detection technique is worth to be applied.