OBJECTIVE: It was hypothesized that somato- sensory evoked potentials can be achieved faster by selective averaging during periods of low spontaneous electroen- cephalographic (EEG) activity. We analyzed the components of EEG that decrease the signal-to-noise ratio of somatosensory evoked potential (SEP) recordings during propofol anesthesia. METHODS: Patient EEGs were recorded with a high sampling frequency during deep anesthesia, when EEGs were in burst suppression. EEGs were segmented visually into bursts, spindles, suppressions, and artifacts. Tibial somatosensory evoked potentials (tSEPs) were averaged offline separately for burst, suppression, and spindle segments using a signal bandwidth of 30-200 Hz. Averages achieved with 2, 4, 8, 16, 64, 128, and 256 responses were compared both visually, and by calculating the signal-to-noise ratios. RESULTS: During bursts and spindles, the noise levels were similar and significantly higher than during suppressions. Four to eight times more responses had to be averaged during bursts and spindles than during suppressions in order to achieve a similar response quality. Averaging selectively during suppressions can therefore yield reliable tSEPs in approximately one-fifth of the time required during bursts. CONCLUSION: The major source of EEG noise in tSEP recordings is the mixed frequency activity of the slow waves of bursts that occur during propofol anesthesia. Spindles also have frequency components that increase noise levels, but these are less important, as the number of spindles is fewer. The fastest way to obtain reliable tSEPs is by averaging selectively during suppressions.
OBJECTIVE: It was hypothesized that somato- sensory evoked potentials can be achieved faster by selective averaging during periods of low spontaneous electroen- cephalographic (EEG) activity. We analyzed the components of EEG that decrease the signal-to-noise ratio of somatosensory evoked potential (SEP) recordings during propofol anesthesia. METHODS:Patient EEGs were recorded with a high sampling frequency during deep anesthesia, when EEGs were in burst suppression. EEGs were segmented visually into bursts, spindles, suppressions, and artifacts. Tibial somatosensory evoked potentials (tSEPs) were averaged offline separately for burst, suppression, and spindle segments using a signal bandwidth of 30-200 Hz. Averages achieved with 2, 4, 8, 16, 64, 128, and 256 responses were compared both visually, and by calculating the signal-to-noise ratios. RESULTS: During bursts and spindles, the noise levels were similar and significantly higher than during suppressions. Four to eight times more responses had to be averaged during bursts and spindles than during suppressions in order to achieve a similar response quality. Averaging selectively during suppressions can therefore yield reliable tSEPs in approximately one-fifth of the time required during bursts. CONCLUSION: The major source of EEG noise in tSEP recordings is the mixed frequency activity of the slow waves of bursts that occur during propofol anesthesia. Spindles also have frequency components that increase noise levels, but these are less important, as the number of spindles is fewer. The fastest way to obtain reliable tSEPs is by averaging selectively during suppressions.
Authors: Mika Särkelä; Seppo Mustola; Tapio Seppänen; Miika Koskinen; Pasi Lepola; Kalervo Suominen; Tatu Juvonen; Heli Tolvanen-Laakso; Ville Jäntti Journal: J Clin Monit Comput Date: 2002-02 Impact factor: 2.502