Download TETRA-POLAR BIOELECTRICAL IMPEDANCE ANALYZER

TETRA-POLAR BIOELECTRICAL IMPEDANCE ANALYZER
Daniel Santoso1, Yuh-Show Tsai 2
Department of Electronics and Computer Engineering, Satya Wacana Christian University
52 – 60 Diponegoro Rd., Salatiga, Indonesia 50711
2
Department of Biomedical Engineering, Chung Yuan Christian University
200 Chung Pei Rd., Chung-Li, Taiwan R.O.C. 32023
email : [email protected], [email protected]
1
ABSTRACT
The paper introduces a tetra-polar
bioelectrical impedance analyzer design using
microcontroller which can provide information
regarding body composition based on bioelectrical
impedance analysis (BIA). The primary objective of
this development is to provide a non-invasive, valid,
and reliable technique of estimating human body
composition. In this research, body composition is
divided into body fat (BF), body water (TBW), and
body muscle (BM) which is displayed in percentage
using character liquid crystal display (LCD). The
analyzer uses Fricke’s electrical circuit to model the
behavior of biological tissues in vivo. From current
stimulus perspective, the analyzer is categorized
multi-frequency BIA means the circuit uses
different frequencies to evaluate body composition.
The method of injecting current to body and
measure the voltage drop is tetra-polar method. A
prototype has been developed and the performance
has been evaluated. The system was tested on the
human body repeatedly over certain period and the
gathered data set is analyzed. The data analysis
indicates that BF, TBW, and BM estimation has
maximum error 2.82%, 0.65%, and 0.88%
respectively.
Keywords: BIA, body fat, body water, body
muscle, microcontroller.
1
INTRODUCTION
Assessment of human body composition is
an important factor in determining the nutritional
status of an individual and of populations. Although
a variety of body composition methods is available,
the majority are limited to the research or clinical
laboratory. It
includes
densitometry
[1],
determination of total body water (TBW) [2],
potassium (TBK) [3], nitrogen [4], and calcium [5],
tomography [6], and electrical conductivity [7].
Other techniques that have been developed for field
use, such as skinfold thickness [8] and ultrasound
[9], are less reliable predictors of human body
composition. Thus, there is a need for a technique
that provides reliable and valid estimates of human
body composition. The method should be noninvasive and suitable for use outside of the
laboratory. Measurement of whole body
bioelectrical impedance is a approach that may meet
this need.
The method for determining body impedance
is based upon the condition of applied current in the
organism. In biological structures, application of a
constant alternating current results in an impedance
to the spread of the current that is frequency
dependent. The living organism consists of intraand extracellular fluids that behave as electrical
conductors, and cell membranes that act as
electrical capacitors. These capacitors are regarded
as imperfect reactive element. At low frequencies,
the current passes mainly through the extracellular
fluid while at higher frequencies; it passes through
extra- and intracellular fluids [10,11]. In this
manner, body fluids and electrolytes are responsible
for electrical conduction and cell membranes are
involved in capacitance.
The pioneering work of relating electrical
impedance measurements to biological function has
been conducted since 1871. Thomasset [10,11]
conducted the original studies using electrical
impedance measurements as an index of total body
water (TBW), using two subcutaneously inserted
needles. Hoffer et al.[12] and Nyboer [13] first
introduced the four-surface electrode bioelectrical
impedance
analysis
(BIA)
technique.
A
disadvantage of surface electrodes is that a high
current (800 mA) and high voltage must be utilized
to decrease the instability of injected current related
to cutaneous impedance (10 000 Ω/cm2) [14].
By the 1970s the foundations of BIA were
established, including those that underpinned the
relationships between the impedance and the body
water content of the body. A variety of single
frequency
BIA
analyzers
then
became
II-21
ISSN 2085-1944
II-22
The 6th International Conference on Information & Communication Technology and Systems
commercially available, and by the 1990s, the
market included several multi-frequency analyzers.
The use of BIA as a bedside method has increased
because the equipment is portable and safe, the
procedure is simple and noninvasive, and the results
are reproducible and rapidly obtained.
The purpose of this research is to develop a
tetra-polar bioelectrical impedance analyzer for
assessing total body water (TBW), body fat (BF),
and body muscle (BM), to determine the reliability
of impedance measures, and investigate the validity
of this measurements by comparison with
commercially available instrument.
2
PRINCIPLES, DESIGN, AND
IMPLEMENTATION
In this chapter, the fundamental principles of
bioelectrical impedance analysis are reviewed first
in order to enable better understanding upon design
and implementation section that described
subsequently.
2.1
Principles of Bioelectrical
Impedance Analysis
The resistance (R) of a length of
homogeneous conductive material of uniform cross
sectional area is proportional to its length (L) and
inversely proportional to its cross sectional area
(A), as illustrated in figure below.
Figure 1. Cylinder model for the relationship between impedance
and geometry
where ρ is the resistivity of the conducting materials
and V equals AL. Although the body is not a
uniform cylinder and its conductivity is not
constant, an empirical relationship can be
established between the impedance quotient
(length2/R) and the volume of water, which contains
electrolytes that conduct the electrical current
through the body. In practice, it is easier to measure
height than the conductive length, which is usually
from wrist to ankle. Therefore, the empirical
relationship is between lean body mass (typically
73% water) and height2/R.
Due to the inherent field in-homogeneity in
the body, the term height2/R describes an equivalent
cylinder, which must be matched to the real
geometry by an appropriate coefficient. This
coefficient depends on various factors, among them
also the anatomy of the segments under
investigation. Therefore, errors occur when there
are alterations in resistivity of the conductive
material, variations in the ratio height to conductive
length, and variations in the shape of the body and
body segments (body segments behave as if they
are in series with each other, with shorter and
thicker segments contributing less to the total R).
Another complexity is that the body offers
two types of R to an electrical current: capacitive R
(reactance), and resistive R (simply called
resistance). The capacitance arises from cell
membranes, and the R from extra- and intracellular
fluid. Impedance is the term used to describe the
combination of the two. Several electrical circuits
have been used to describe the behavior of
biological tissues in vivo [15]. One of them
involves arranging R and capacitance in series,
another in parallel, whilst others are more complex.
A circuit that is commonly used to represent
biological tissues in vivo is one in which the R of
extracellular fluid is arranged in parallel to the
second arm of the circuit, which consists of
capacitance and R of intracellular fluid in series, as
illustrated below.
Hence resistance:
R = ρL/A = ρL2/V
(1)
and volume:
V = ρL2/R
(2)
ISSN 2085-1944
Tetra-Polar Bioelectrical Impedance Analyzer-Daniel Santoso
Figure 2. The human body consists of resistance and capacitance
connected in parallel or in series
R and capacitance can all be measured over
a range of frequencies (most single-frequency BIA
analyzers operate at 50 kHz). At zero (or low)
frequency, the current does not penetrate the cell
membrane, which acts as an insulator, and therefore
the current passes through the extracellular fluid,
which is responsible for the measured R of the body
R0. At infinite frequency (or very high frequency)
the capacitor behaves as a perfect (or near perfect)
capacitor, and therefore the total body R (R∞)
reflects the combined of both intracellular and
extracellular fluid. Since practical constraints and
the occurrence of multiple dispersions prevent the
use of a direct current (zero frequency) or very high
frequency AC currents, the R values at the ideal
measurement frequencies are predicted using a
Cole–Cole plot [16] (negative reactance versus R
plot), with R0 theoretically representing the R of the
extracellular fluid (intracellular water) and R 
representing the R of intra- and extracellular fluid
(TBW), as illustrated in Figure 3.
Figure 3. Diagram of the graphical derivation of the phase angle;
its relationship with resistance (R), reactance (Xc), impedance
(Z) and the frequency of the applied current
The cylinder model in Figure 1 which is
representation of human body, can be partitioned
into several compartments as illustrated below.
II-23
Figure 4. Diagram of fat-free mass (FFM), total body water
(TBW), intra-cellular water (ICW), extra-cellular water (ECW),
and body cell mass (BCM)
According to the Figure 4, human body can
be divided into two major partitions, fat mass (FM)
and fat-free mass (FFM). FFM is everything that is
not body fat. It is mainly composed of body water
and small portion of protein and mineral. This
portion will be used to derive body water and body
muscle percentages. Whereas FFM is everything
that is not body fat, there is no consensus on the
physiological meaning of measures of body cell
mass (BCM). The BCM is the protein rich
compartment, which is affected in catabolic states,
and estimating the size is difficult because it is a
complex compartment, comprising all non-adipose
cells as well as the aqueous compartment of
adipocytes.
BIA measurements must be standardized in
order to obtain reproducible results. Reported mean
coefficients of variation for within-day R
measurements are  1–2%; daily or weekly intraindividual variability is slightly larger ranging from
 2% to 3.5% [17 - 20]. Day-to-day coefficients of
variation increases for frequencies lower than 50
kHz [21]. Overall reproducibility/precision is 2.7–
4.0% [19]. Prediction errors were estimated to be 3–
8% for TBW and 3.5–6% for FFM, respectively
[22, 23].
Subjects must be measured (recall values are
not acceptable) for height and weight at the time of
the BIA measurement. Standardized conditions with
regard to body position, previous exercise, dietary
intake and skin temperature must be respected. No
interference with pacemaker or defibrillators is
anticipated [24]. Although there are no known
incidents reported as a result of BIA measurements,
the possibility cannot be eliminated that the induced
ISSN 2085-1944
The 6th International Conference on Information & Communication Technology and Systems
II-24
field of current during the measurement could alter
the pacemaker or defibrillator activity.
2.2
Design and Implementation of
Bioelectrical Impedance Analyzer
using a separate set of electrodes will be that of the
deep tissues, since negligible current is drawn
through the electrode. The tetra-polar arrangement
is illustrated in Figure 6.
i
LV
The analyzer design involves mixed
circuitry, both analog and digital. Primarily, the
analog circuit is used to perform signal conditioning
upon bio-potential signal picked up by electrodes.
The digital counterpart is used to administer signal
injection to human body through electrodes and
process the acquired signal from analog circuit into
human readable form with the help of analog to
digital converter (ADC). The block diagram of
bioelectrical impedance analyzer is illustrated in
Figure 5. The heart of the system is microcontroller
MPC82G516A from Megawin Technology.
Active low pass filter
5 Khz
Wave-shape
converter 1
#
LV
PA3
#
PC4
50 Khz
i
#
#
v
PC5
RH
PA7
GND
PC2
PA0
INT0
SEL
R302
INT1
R301
1
A
2
3
1
RV
AC to DC
converter
B
RV
Electrode pads
LH
RH
RICW
RC
RECW
RC
LH
v
V
RV
i
XC
RC
Z
v
V
RC
RH
Figure 6. Four electrode pads and equivalent circuit of body
impedance
A circuit model called Fricke’s circuit
[Figure 2] is used to represents body impedance.
There are four additional resistors included to the
model due to skin resistance. The electrode pads are
made from particular conductive material deposited
on glass surface.
Wave-shape
converter 2
VR2
LH
Z
LV
#
ADC
2
3
Differential amplifier
multiplexer
Figure 5. Biolectrical impedance analyzer block diagram
Although the block diagram looks
complicated, the working mechanism is quite
straightforward. As it is reflected from the name,
the sensing target for BIA estimation is body
impedance. Like the impedance in the electrical
circuit, the common measurement technique is to
inject a certain amount of alternating current to
target impedance and measure the voltage drop due
to impedance. The impedance is then simply
voltage per current. This technique is also
applicable to BIA estimation, with slight
modification.
In this research, the method of injecting
current to body and measure the voltage drop is
known as tetra-polar method. Current is injected
through the electrode pads under the front of each
foot and the voltage measured by the electrodes
under the heel. Since the aim in BIA is to measure
the impedance of the deep tissues, the tetra-polar
method is preferred as the voltage drop measured
From current stimulus perspective, this BIA
circuit is categorized multi-frequency BIA means
the circuit uses different frequencies to evaluate
FFM, TBW, ECW, and ICW. In this research, the
lower frequency is set at 5 kHz while the higher
frequency is set at 50 kHz. The microcontroller is
responsible for these signals generation. Timer 2
and timer 2 interrupt is used to address this task.
The timer is set to auto-reload mode and
appropriate reload value is assigned to timer
register. Each time timer 2 overflows; the routine in
timer 2 ISR alternates the state of PA3. Therefore,
as long as timer 2 runs, continuous square wave
signal generated and outputted to PA3.
The signal gating is governed by PC4 and
PC5. When PC4 at logic “high”, 5 kHz signal is
delivered to the system and PC5 does likewise with
50 kHz signal. The requirement for BIA circuit is
sine wave signal; therefore the square wave signal
should be converted using converter 1 and converter
2. These blocks have similar construction, only
differ in component values. Before injected to target
impedance, this signal is amplified and filtered
using active low-pass filter.
Each complete impedance measurement
takes three consecutive steps. This sequence is
performed by multiplexer, which is controlled by
PA7 and PC2. The better understanding about this
sequence can be obtained by simplifying BIA block
diagram, as illustrated in Figure 7.
ISSN 2085-1944
Tetra-Polar Bioelectrical Impedance Analyzer-Daniel Santoso
5/50 kHz
5/50 kHz
V
5/50 kHz
A
3.7
vo
vo
R301
ZL
ZL
i
R302
R301
B
ZL
2.6
R302
C
Step 2
INT0
D
i
R302
C
Step 1
vo
R301
i
II-25
C
Step 3
0.8
Figure 7. Simplified BIA block diagram and measurement steps
t
0
R301 and R302 serve as reference
resistances; the value is 300 Ω and 600 Ω
respectively. The sequence is quite straightforward,
and also applies both to 5 kHz and 50 kHz
impedance measurement. Firstly, the multiplexer’s
inputs are directed to measure voltage drop between
two reference resistances (vAC). Secondly, the
multiplexer’s inputs are then directed to measure
voltage drop between single reference resistance
(vBC). Finally, to obtain the target impedance value,
the multiplexer’s inputs are directed to measure
voltage drop between target impedance (vDC). It is
required to measure impedance at 50 kHz and
impedance at 5 kHz. Therefore, the complete
sequence to estimate body composition takes 6
steps.
The output of multiplexer is connected to the
input of differential amplifier to be amplified with
certain gain. Up to this stage the signal is still
alternating current (ac). Therefore a circuit to get
direct current (dc) equivalent is inserted in between
differential amplifier and ADC. In addition, this
circuit also adjusts the dc output to a certain level
that is within the range of ADC input.
The ADC circuit doesn’t refer to embedded
ADC inside Megawin MPC82G516A. Instead, it
refers to external mixed circuit serves as dual-slope
integrating ADC – like. The primary function of
this circuit is to translate a certain level of DC input
voltage to a pulse-width modulated square wave
signal. Clearly, in this ADC the change of input
voltage will modulate the signal’s duty cycle. The
duty cycle is then fed to microcontroller and
interpreted using particular algorithm. By doing this
way, voltage information (volt) is conveyed to
microcontroller in time unit (second). The
waveform fed to microcontroller’s INT0 (interrupt
port 0) is illustrated in Figure 8.
t1
t2
T
Figure 8. Pulse-width modulated signal as representation of
voltage
For this application, t2 is used to obtain the
target impedance (ZL) value. The measurement
concept is illustrated in Figure 9.
t2(us)
(x2,y2)
t2b
(x3,y3)
t2c
t2a
0
(x1,y1)
vbc
vdc
vac
300
ZL
600

 
V

R(Ohm)
Figure 9. Impedance measurement based on t2 reading
At first step, microcontroller obtains t2a as
representation of vAC. Secondly; t2b is then obtained
as representation of vBC. It is commonly understood
that value of vAC is twice the value of vBC since the
same current flows through those resistances.
However, it is not relevant to t2a and t2b. Finally, at
third step, t2c which is representation of vDC is
obtained.
The first two steps are used to establish
reference line. The slope of this line must be
calculated using equation (3) to enable impedance
measurement.
m
y2  y1
x1  x2
(3)
The m in equation (3) can be also expressed using
equation (4).
m
ISSN 2085-1944
y3  y1
x3  x1
(4)
The 6th International Conference on Information & Communication Technology and Systems
II-26
By rearranging equation (4), the actual impedance
(x3) can be obtained.
x3  x1 
y3  y1
m
Z  ( X C  RICW ) || RECW
(5)
The impedance at 5 kHz and impedance at
50 kHz, along with weight, height, age, and gender
information will be used to estimate body
composition. The final result of this estimation is
percentage of body fat (BF), total body water
(TBW), and body muscle (BM) which is displayed
on character liquid crystal display (LCD). The
formulas are not shown due to proprietary right.
Impedance measurement at 50 kHz is used to
estimate body fat and total body water while
impedance measurement at 5 kHz is used to
estimate body muscle. Actually, impedance
measurement at 50 kHz means TBW estimation.
TBW itself consists of ECW and ICW. Information
about body muscle percentage is derived from ICW,
meanwhile the impedance at 5 kHz only represents
ECW. Therefore, ICW value must be obtained from
TBW – ECW.
3
with impedance Z. The value of Z can be calculated
using (6).
RESULT
The
design
considerations
and
implementation regarding bioelectrical impedance
analyzer has been elaborated in previous subchapter. A prototype of the analyzer has been
successfully developed and tested.
Basically there are two tests have been
conducted upon the prototype to determine its
reliability of impedance measures, and investigate
the validity of the measurements. The purpose of
the first test is to ensure that the analyzer can
measure impedance of electric circuit model of
human body properly. It is reasonable to assume
that if BIA section shows correct result in
measuring impedance of electric circuit, it will
remain correct for measuring impedance of human
body as well. After that, at the second test, the
analyzer applied to human body to investigate its
performance in the real-world application.
The circuit model for human body
impedance used in the first test is Fricke’s circuit.
This model is further improved by incorporating
skin resistance on each terminal to simulate real
application, as illustrated in Figure 6 previously.
The circuit model is basically a simple RC network
(6)
In the first test, RICW = 560 Ω, RECW = 560 Ω, C =
4.7 nF, RC = 820 Ω were used. The impedance both
at 5 kHz and 50 kHz were then measured. The
experiment was repeated 10 times to ensure the
repeatability. The BIA results (BF, TBW, and BM)
were obtained by plugging in Z5kHz and Z50kHz in
BIA formula inside the microcontroller. The other
parameters needed to compute BIA (gender, age,
height, and weight) were manually inputted. The
result is based on reading of LCD screen located on
front panel of the analyzer as illustrated in Figure 9.
Figure 9. LCD screen when measuring impedance in Fricke’s
circuit
The result of 10 consecutive experiments is
summarized in Table 1.
Table 1. Result of impedance test on Fricke’s circuit
Z5kHz(Ω)
550
550
550
548
550
548
550
549
550
548
Z50kHz(Ω)
380
385
382
381
381
383
387
386
380
387
BF(%)
8.0
8.3
8.1
8.0
8.0
8.1
8.4
8.3
8.0
8.4
TBW(%)
67.3
66.8
67.3
67.3
67.3
66.8
66.8
66.8
67.3
66.8
BM(%)
55.5
54.8
55.5
55.5
55.5
54.8
54.8
54.8
55.5
54.8
Based on hand calculation, Z5kHz and Z50kHz
equals to 554 ∟ -4o Ω and 376 ∟ -19o Ω. In
average, the test result indicates Z5kHz and Z50kHz
equals to 549 Ω and 383 Ω which is close to the
result from hand calculation. The analyzer only
measures the real part from the complex impedance
while the imaginary is omitted. Based on data
analysis of the first test, the impedance
measurement at 5 kHz has maximum error 0.1 %
while 50 kHz counterpart has 0.45%.
The second test can be regarded as simple
clinical test since it was conducted on human body
ISSN 2085-1944
Tetra-Polar Bioelectrical Impedance Analyzer-Daniel Santoso
over a period of time. This test was conducted in 10
consecutives days. Each day, the test was conducted
in two sessions, morning and evening, repeated for
10 times per session. It is assumed that during
repetition, the environmental factors will not change
drastically thus any variation on the result shall be
caused by the internal mechanism of the system. In
average, one session takes 15 minutes. On every
session, subject’s weight and height were measured
using traditional weight scale and ruler since recall
values are not acceptable in BIA. A commercial
product from SOEHNLE, namely Body Balance
was used as reference instrument. Totally, the
system has been tested at least for 200 times thus it
also provides reliability information of the analyzer.
According to test scheme, there are vast amount of
data set that can be analyzed to evaluate the
performance of the system. However, those data are
not shown in this paper because of the limited
space. From data analysis, it can be concluded that
BF, TBW, and BM estimation has maximum error
2.82%, 0.65%, and 0.88% respectively.
4
CONCLUSION AND FUTURE
DIRECTIVES
A prototype of tetra – polar bioelectrical
impedance analyzer has been designed and
implemented based on Megawin 82G5126A
microcontroller. The analyzer uses tetra-polar
method for injecting current to body and measure
the voltage drop. From current stimulus perspective,
the analyzer is categorized multi-frequency BIA
which uses 5 kHz as lower frequency and 50 kHz as
higher frequency. Using Fricke’s circuit model, the
impedance measurement at 5 kHz has maximum
error 0.1 % while 50 kHz counterpart has 0.45%. A
simple clinical test has been conducted as well on
human body repeatedly over a period of time. From
data analysis upon the tests, it can be concluded that
BF, TBW, and BM estimation has maximum error
2.82%, 0.65%, and 0.88% respectively.
Thus far, the analyzer has capability of
determining body composition based on
bioelectrical impedance analysis. The composition
which consists of body fat, body water, and body
muscle are displayed on LCD in terms of
percentage. This can further be improved by
implementing algorithm to compare acquired body
composition with a particular standard. Based on
the comparison and analysis, the analyzer is then
generates recommendation regarding diet and
exercise for user.
II-27
REFERENCES
[1] Brozek J, Grande F, Anderson JT, Keys A
(1963) Densitometric analysis of body
composition: revision of some quantitative
assumptions. Annals of New York Academy
of Sciences l10:113-40.
[2] Panaretto BA. (1968) Estimation of body
composition by the dilution of hydrogen
isotopes. In: Body Composition in Animals
and Man. Washington, DC: National
Academy of Sciences, National Research
Council, pp.200-17.
[3] Forbes GB, Hursh JB. (1963) Age and sex
trends in lean body mass calculated from 40K
measurements: with a note on the theoretical
basis for the procedure. Annals of New York
Academy of Sciences l10:255-63.
[4] Vartsky D, Ellis KJ, Cohn SH. (1979) In vivo
quantification of body nitrogen by neutron
capture prompt gamma-ray analysis. Journal
of Nuclear Medicine 20:l158-65.
[5] Cohn SH, Shukla KK, Dombrowski CS,
Fairchild RG. Design and calibration of a
“broad beam” 238Pu, Be neutron source for
total body neutron activation analysis
activation analysis. Journal of Nuclear
Medicine l3:487-92.
[7] Presta E, Wang J, Harrison 0G. Bjorntorp P.
Harker WH, Van Itallie TB. (1983)
Measurement of total body electrical
conductivity: a new method for estimation of
body composition. American Journal of
Clinical Nutrition 37: 735-9.
[8] Lohman SG. (1981) Skinfolds and body
density and their relation to body fatness: a
review. Human Biology 53:181-225.
[9] Haymes EM, Lundegren HM, Loomis JL,
Buskirk ER. (1976) Validity ofthe
ultrasonic technique as a method of
measuring subcutaneous adipose tissue.
Annals of Human Biology 3:245-5 1.
[10] Thomasset (1962) A. Bio-electrical
properties
of
tissue
impedance
measurements. Lyon Medical 207:l07-18.
[11] Thomasset A. (1963) Bio-electrical
properties of tissues. Lyon Medical
209:1325-52.
[12] Hoffer EC, Clifton KM, Simpson DC.
(1969)
Correlation
of
wholebody
impedance with total body volume. Journal
of Applied Physiology 27:531–4.
[13] Nyboer J. (1970) Electrical impedance
plethysmograph, 2nd ed. Springfield, IL:
CC Thomas.
ISSN 2085-1944
II-28
The 6th International Conference on Information & Communication Technology and Systems
[14] Boulier A, Fricker J, Thomasset A-L,
Apfelbaum M. (1990) Fat-free mass
estimation by the two-electrode impedance
method. American Journal of Clinical
Nutrition 52:581–5.
[15] Gudivaka R, Schoeller DA, Kushner RF,
Bolt MJ. (1999) Single- and multifrequency
models for bioelectrical impedance analysis
of body water compartments. Journal of
Applied Physiology 87: 1087–96.
[16] Cole KS. (1941) Dispersion and absorption
in dielectrics. I. Alternating current
characteristics. Journal of Chemical Physics
9:341–951.
[17] Kushner RF, Schoeller DA. (1986)
Estimation of total body water by
bioelectrical impedance analysis. American
Journal of Clinical Nutrition 44:417–24.
[18] Segal KR, Gutin B, Presta E, Wang J, Van
Itallie TB. (1985) Estimation of human
body composition by electrical impedance
methods: a comparative study. Journal of
Applied Physiology 58:1565–71.
[19] Lukaski HC, Johnson PE, Bolonchuk WW,
Lykken GL. (1985) Assessment of fat-free
mass using bioelectrical impedance
measurements of the human body.
[20]
[21]
[22]
[23]
[24]
ISSN 2085-1944
American Journal of Clinical Nutrition
41:810–7.
Ward LC, Byrne NM, Rutter K, et al.
(1997) Reliability of multiple frequency
bioelectrical impedance analysis: an
intermachine comparison. American Journal
Human Biology 9: 63–72.
Jansen DF, Korbijn CM, Deurenberg P.
(1992) Variability of body density and body
impedance at different frequencies.
European Journal Clinical Nutrition
46:865–71.
Sun SS, Chumlea WC, Heymsfield SB, et
al. (2003) Development of bioelectrical
impedance analysis prediction equations for
body composition with the use of a
multicomponent model for use in
epidemiologic surveys. American Journal of
Clinical Nutrition 77:331–40.
Kushner
RF.
(1992)
Bioelectrical
impedance analysis a review of principles
and applications. Journal of the American
College of Nutrition 11: 199–209.
Liang MT, Norris S. (1993) Effects of skin
blood flow and temperature on bioelectric
impedance after exercise. Medicine &
Science in Sports & Exercise 25:1231–9.