Biomedical Engineering Reference
In-Depth Information
Rampil and F. J. Sasse, unpublished results, 1977]. SEF95 is the frequency below
which 95% of the power in the spectrum resides. Clearly, either approach to SEF
calculation provides a monitor that is only sensitive to changes in the width of the
spectral distribution (there is always some energy in the low-frequency range).
Approximately 260 peer-reviewed papers describe the use of SEF. The field of
pharmacodynamics (analysis of the time-varying effects of drugs on physiology) of
anesthetics and opioids benefited enormously from access to the relatively sensitive,
specific and real-time SEF. Many of the algorithms driving open-loop anesthetic
infusion systems use population kinetic data derived using SEF.
Similar to MPF, few of the existing published trials examine the utility of SEF in
reducing drug dosing while ensuring clinically adequate anesthesia. In our hands,
neither the SEF nor the F95 seems to predict probability of movement response to
painful stimulus or verbal command in volunteers [35] at least in part due to the
biphasic characteristic of its dose response curve. While the SEF is quite sensitive to
anesthetic effect, there is also substantial variation across patients and across drugs.
Therefore, a specific numeric value for SEF that indicates adequate anesthetic effect
in one patient may be not be adequate in the same patient using a different drug. A
rapid decline in SEF (>50% decrease sustained below prior baseline within <30 sec-
onds) in a patient being monitored has, however, been reliably correlated with the
onset of cerebral ischemia during carotid artery surgery [3, 54-57].
Many commonly used general anesthetics produce burst suppression EEG pat-
terns without slowing the waves present during the remaining bursts, thus pure SEF
of the epoch would not reflect the additional anesthetic-induced depression. Com-
bining the SEF with the BSR parameter to form the burst-compensated SEF (BcSEF)
creates a parameter that appears to smoothly track changes in the EEG due to either
slowing or suppression from isoflurane or desflurane [43, 58]:
BSR
⎛
⎜
⎞
⎟
BcSEF
=
SEF
1
−
100
Spectral qEEG parameters such as MPF or SEF compress into a single variable
the 60 or more spectral power estimates that constitute the typical EEG spectrum. As
Levy pointed out [59], a single feature may not be sensitive to all possible changes in
spectral distribution. Frequency domain-based qEEG parameters, like their time
domain-based relatives, are generally averaged over time prior to display. The
author uses nonlinear smoothing when computing SEF that strongly filters small
variations, but passes large changes with little filtering. This approach, also known as
a tracking filter, diminishes noise, but briskly displays major changes, such as those
that can occur secondary to ischemia or following a bolus injection of anesthetics.
Figure 9.6(a) illustrates a sample of an anesthetized human EEG as it is trans-
formed by common analytical processes. The original EEG waveform is
x
(
i
) (after
analog antialias filtering with a bandpass of 0.3 to 30 Hz) digitized at 256 Hz for 16
seconds. The tracing below
x
(
i
) demonstrates the effect of windowing on the origi-
nal signal. Windowing is a technique employed to reduce distortion from epoch end
artifacts in subsequent frequency-domain processing. A window consists of a set of
digital values with the same number of members as the data epoch. In this case, a
Blackman window was employed:
Search WWH ::
Custom Search