Biomedical Engineering Reference
In-Depth Information
dialysis. BIS may be accurate with suspended cells, but
unfortunately it is impossible to estimate the extracel-
lular electric volume of tissues because of anisotropy and
limited low frequency data. It is common practice not to
measure below 1 kHz, and low frequency dispersions are
therefore neglected. The measuring current is usually
around 0.5 mA, higher current levels are difficult to use
because the threshold of perception is reached at the
lowest frequencies (cf. Section 4.1.17).
The third
method is the prediction of ECWand TBW
with low (1-5 kHz) and high (100-500 kHz) frequency
impedance data (dual- or multifrequency BIA), with the
ICW calculated by difference. Volume calibration is
obtained with regression equations as in single-frequency
BIA. Like BIS, multifrequency BIA relies on the wrong
assumption of tissue isotropy with low frequency current
only flowing around cells. If the hydration of the fat free
mass is not fixed at 73% (e.g. in hemodialysis, in patients
with edema or heart failure) or when body weight is
meaningless for body compartments (e.g. in ascites,
pregnancy and severe obesitiy); then BIA, multifre-
quency BIA and BIS prediction equations should not be
used.
A more recent method is to use the complex imped-
ance vector at 50 kHz in a probabilistic Wessel diagram.
Such vector BIA (or BIVA) is based upon patterns of the
impedance vector relating body impedance to body
hy-
dration
(Piccoli et al., 1994). BIVA is a single frequency
BIA that follows a black-box approach considering
Z
as
a random output of a stochastic system (current flow
through anisotropic tissues). The method consistently
applies to whole body or segmental measurements nor-
malized by the conductor length (height (
H
) for whole
body, body segment length for segmental). BIVA only
needs to take care of the measurement error (in the order
of 3% at 50 kHz) and of the biological variability of
subjects in any clinical condition. The intersubject vari-
ability of
Z
is represented with the bivariate normal
distribution (i.e. with elliptical probability regions in the
Wessel plane) which are confidence (95%) and tolerance
ellipses (50%, 75%, 95%) for mean and individual vectors
respectively (cf.
Fig. 4.1-12
). The intersubject variability
of the impedance vector is represented with the bivariate
normal distribution (i.e. with elliptical sex-specific
probability regions (50%, 75%, and 95% tolerance ellip-
ses) in the Wessel plane. Vector components are nor-
malized by the subject's height (
R/H,
and
X/H,
in Ohm/m).
Upper and lower poles of the 75% tolerance ellipses
represent bioelectrical thresholds for dehydration and
fluid overload, respectively. Vector components can also
be transformed into dimensionless z-scores which allow
comparisons of vector position between different ana-
lyzers (Piccoli et al., 2002). Clinical information on
hydration is obtained through patterns of vector distri-
bution with respect to the healthy population of the same
R/H, Ohm/m
Figure 4.1-12 Z probability graph. Vector position and migration
in the Wessel plane are interpreted and ranked according to
directions: (a) Vector displacement parallel to the major axis of an
ellipse is associated with a progressive change in soft tissue
hydration (short term changes: hours, days). (b) A vector lying on
the left or right side of the major axis of an ellipse is associated
with more or less cell mass respectively (long term changes:
weeks, months).
race, sex, and class of BMI (body mass index) and age (cf.
Fig. 4.1-12
). From clinical validation studies in adults,
vectors falling out of the 75% tolerance ellipse indicate
abnormal tissue impedance. Vector position is inter-
preted and ranked following two directions in the Wessel
plane, as depicted in
Fig. 4.1-12
. The basic pattern has
been recently validated with deuterium dilution (Lukaski
et al., 2007).
4.1.11 Implanted active thoracic
devices
Bioimpedance is especially attractive as a potentially
useful transducing mechanism in implantable devices
such as pacemakers (including pacemakers with cardiac
resynchronization therapy (CRT) and internal cardi-
overter defibrillators (ICD's)) for several reasons: First,
the device circuitry and electrode vector configurations to
perform such measurements is relatively simple and al-
ready exists in many of the current implantable devices
and lead configurations. However, sampling resolution is
limited inmany of these circuits preventing sufficient data
for detecting changes in cardiac impedance waveforms.
Therefore, low resolution impedance measurements that
provide information on fluid status or respiratory im-
pedance trends as a function oftime (day(s)/month(s))
