Biomedical Engineering Reference
In-Depth Information
QUAZI suppression index was designed to detect burst suppression in the presence
of a wandering baseline voltage. QUAZI incorporates slow wave (<1.0-Hz) infor-
mation derived from the frequency domain to detect burst activity superimposed on
these slow waves that would “fool” the original BSR algorithm by exceeding the
voltage criteria for electrical “silence.”
The waveform data in the current epoch is prepared for conversion to the fre-
quency domain by a Blackman window function, as illustrated in Figure 9.6(a).
Then the FFT and the bispectrum of the current EEG epoch are calculated. The
resulting spectrum and bispectrum are smoothed using a running average against
those calculated in the prior minute, then the frequency domain-based
subparameters “SynchFastSlow” and “BetaRatio” are computed. The BetaRatio
subparameter is the log ratio of power in two empirically derived frequency bands:
log[(P30, 47 Hz)/(P11, 20 Hz)]. The SynchFastSlow subparameter is the contribu-
tion from high-order (bispectral) analysis. SynchFastSlow is defined as another log
ratio. Here the log of the ratio of the sum of all bispectra peaks in the area from 0.5
to 47 Hz over the sum of the bispectrum in the area from 40 to 47 Hz. The resulting
BIS is defined as a proprietary combination of these qEEG subparameters.
Each of the component subparameters was chosen to have a specific range of
anesthetic effect where they perform best; that is, the SynchFastSlow (HOS) parame-
ter is well correlated with behavioral responses during moderate sedation or light
anesthesia. The combination algorithm that determines BIS therefore weights the
Beta Ratio (FFT) most heavily when the EEG has the characteristics of light seda-
tion. The SynchFastSlow (bispectral component) predominates during the phenom-
ena of EEG activation (excitement phase) and during surgical levels of hypnosis, and
the BSR and QUAZI detect very deep anesthesia.
The subparameters are combined using a nonlinear function whose coefficients
were determined by the iterative data collection and tuning process. Two key fea-
tures of the Aspect BIS multivariate model are, first, that it accounts for the nonlin-
ear stages of EEG activity by allowing different subparameters to dominate the
resulting BIS as the EEG changes its character with increasing anesthesia. Second,
the model framework is extensible, so new subparameters can be added to improve
performance, if needed, in the presence of new anesthetic regimes. The combination
of the four subparameters produces a single number, BIS, which decreases
monotonically with decreasing level of consciousness (hypnosis). As described ear-
lier, computation of a bispectral parameter (SynchFastSlow) requires averaging sev-
eral epochs; therefore, the BIS value reported on the front panel of the monitor
represents an average value derived from the previous 60 seconds of usable data.
9.5.4 Bispectral Index: Clinical Results
BIS has been empirically demonstrated to correlate with behavioral measures of
sedation and light anesthesia [62-67] due to a variety of anesthetics including
isoflurane [67, 68] sevoflurane [69, 70] and desflurane [71] vapors, and propofol
[67] and midazolam [67], which are parenteral anesthetics of different classes. The
BIS parameter is not sensitive to the effects of ketamine because that agent's domi-
nant effect on the EEG is in the theta-band range. Although it is difficult to test mem-
ory formation in pediatric patients, it appears, based on dose-response, that the BIS
Search WWH ::
Custom Search