Lactate threshold by muscle electrical

impedance in professional rowers

Cite as: Rev. Sci. Instrum. 88, 045105 (2017); https://doi.org/10.1063/1.4979072
Submitted: 31 July 2016 . Accepted: 11 March 2017 . Published Online: 12 April 2017

B. Jotta, A. B. B. Coutinho, A. V. Pino, and M. N. Souza

ARTICLES YOU MAY BE INTERESTED IN

Detection of questionable occlusal carious lesions using an electrical bioimpedance

method with fractional electrical model
Review of Scientific Instruments 87, 084305 (2016); https://doi.org/10.1063/1.4961547

A novel multi-jet burner for hot flue gases of wide range of temperatures and compositions
for optical diagnostics of solid fuels gasification/combustion
Review of Scientific Instruments 88, 045104 (2017); https://doi.org/10.1063/1.4979638

Measurement of fluctuations of electrostatic force acting between a dielectric plate and an

electrostatic drive
Review of Scientific Instruments 88, 044701 (2017); https://doi.org/10.1063/1.4979707

Rev. Sci. Instrum. 88, 045105 (2017); https://doi.org/10.1063/1.4979072 88, 045105

© 2017 Author(s).
 REVIEW OF SCIENTIFIC INSTRUMENTS 88, 045105 (2017)

Lactate threshold by muscle electrical impedance

in professional rowers
B. Jotta, A. B. B. Coutinho, A. V. Pino, and M. N. Souza
Biomedical Engineering Program, Universidade Federal do Rio de Janeiro/COPPE,
Rio de Janeiro 21941-914, Brazil
(Received 31 July 2016; accepted 11 March 2017; published online 12 April 2017)
Lactate threshold (LT) is one of the physiological parameters usually used in rowing sport training
prescription because it indicates the transitions from aerobic to anaerobic metabolism. Assessment of
LT is classically based on a series of values of blood lactate concentrations obtained during progressive
exercise tests and thus has an invasive aspect. The feasibility of noninvasive LT estimative through
bioelectrical impedance spectroscopy (BIS) data collected in thigh muscles during rowing ergometer
exercise tests was investigated. Nineteen professional rowers, age 19 (mean) ± 4.8 (standard deviation)
yr, height 187.3 ± 6.6 cm, body mass 83 ± 7.7 kg, and training experience of 7 ± 4 yr, were evaluated
in a rowing ergometer progressive test with paired measures of blood lactate concentration and BIS in
thigh muscles. Bioelectrical impedance data were obtained by using a bipolar method of spectroscopy
based on the current response to a voltage step. An electrical model was used to interpret BIS data and
to derive parameters that were investigated to estimate LT noninvasively. From the serial blood lactate
measurements, LT was also determined through Dmax method (LTDmax ). The zero crossing of the
second derivative of kinetic of the capacitance electrode (Ce), one of the BIS parameters, was used to
estimate LT. The agreement between the LT estimates through BIS (LTBIS ) and through Dmax method
(LTDmax ) was evaluated using Bland-Altman plots, leading to a mean difference between the estimates
of just 0.07 W and a Pearson correlation coefficient r = 0.85. This result supports the utilization of
the proposed method based on BIS parameters for estimating noninvasively the lactate threshold in
rowing. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4979072]

I. INTRODUCTION assessment is invasive and involves a serial blood sampling,

Progressive exercise tests are ordinarily used to evaluate
physical endurance of athletes, and the lactate threshold (LT)
is one of the physiological parameters that can be obtained
during such tests and that is usually used in training prescrip-
tion.1 Lactate threshold, which can be defined as the load
of exercise associated to an abrupt increase in blood lactate
concentration, indicates the transitions from aerobic to anaero-
bic metabolism.2,3 Training intensities slightly above LT have
been considered a high stimulation of oxidative metabolism
of muscle cells and thus LT is the highest appropriate load for
endurance training.2,3
There are different methods to assess the LT and most of
them are based either on a fixed concentration of lactate values
or on lactate individual kinetic standards.4 Despite having been
developed on running, and later, on bicycle tests, LT assess-
ment is also applied in other sports, as for example rowing, to
determine individual loads in athletes training.4 One interest-
ing aspect is that the movement pattern or the ergometer used
in the approach adopted in the LT assessment has been asso-
ciated to different blood lactate concentrations interpreted as
the transition from aerobic to anaerobic metabolism.5 A pos-
sible explanation for the influence of body muscles proportion
used is that the musculatures of upper and lower limbs have
different percentages of muscle fiber type,6 thus resulting in a
different glycolytic metabolism.2
Despite its popularity for training prescription and the
movement pattern used, the classical procedure for the LT
which is considered uncomfortable for some athletes.7 Thus,
techniques related to muscle metabolic state as bioelectrical
impedance spectroscopy (BIS)7 have been used to estimate
the LT to eliminate such an invasive aspect.
Although LT assessment through noninvasive techniques
has been reported previously, e.g., electromyography (EMG)8
and near infrared spectroscopy (NIRS),9 to the best of our
knowledge there are no previous studies that investigated a
noninvasive method with BIS to estimate LT during movement
patterns with a higher proportion of the active body muscles,
like rowing or running. Such a LT estimation would be of
great value for professional rowers since it could eliminate the
invasive nature of serial blood sampling, increasing the number
of LT assessments during a given period of time, helping in
the physical conditioning of such athletes. Thus, the aim of the
present study was to investigate the feasibility of LT estimation
through the BIS parameter collected in thigh muscles during
rowing ergometer exercise tests. Results of LT obtained by the
traditional serial blood sampling method were compared to the
estimates of LT achieved through BIS parameters.
II. MATERIALS AND METHODS
A. Subjects and protocols
Nineteen professional rowers of a traditional rowing club
in Rio de Janeiro—Brazil, age 19 (mean) ± 4.8 (standard devi-
ation) yr, height 187.3 ± 6.6 cm, body mass 83 ± 7.7 kg and

0034-6748/2017/88(4)/045105/6/$30.00 88, 045105-1 Published by AIP Publishing.

045105-2 Jotta et al.
training experience of 7 ± 4 yr, participated in this study.
Five subjects are members of the Brazilian National Row-
ing Team and one of them competed in the 2012 Olympic
Games. The study was submitted and approved by the ethics
committee of the Clementino Fraga Filho University Hospi-
tal, at the Federal University of Rio de Janeiro, under number
1.343.260.
The preparation for the exercise test began with the
removal of the stratum corneum layer by rubbing with cotton
and saline solution for electrodes placement. This procedure
was performed on the anterior thigh in order to place the
electrodes for measuring the bioelectrical impedance. Two
metal brass electrodes (12 cm2 area) with electrolytic gel
were used to minimize the effect of electrode impedance and
improve the accuracy of BIS measures in consensus with
recent recommendations.10 These electrodes were placed at
20% of the distance between the upper edge of the patella
and the anterior superior iliac spine, displaced 5 cm laterally
to the lateral edge of the patella. The distance between the
electrodes was 9 cm. Both electrodes and wires near were
fixed on the thigh with an elastic cotton tape (Kinesiotaping,
USA).
Volunteers were submitted to a fixed incremental work-
load exercise test on a rowing ergometer (Concept, USA)
with 5-6 stages of 5 min each and a 2-min interval between
them. The initial work load was set to 100-110 W and
the incremental load was 25-35 W. Both loads were deter-
mined based on the individual performance during train-
ings. Capillary blood samples to measure lactate concen-
trations were collected on the fingertips and paired with
BIS acquisition during the intervals between each stage.
BIS data were collected with the lower limb in a standard
position (keeping—the extension of the knee). The test was
stopped when the athlete could not complete the intended
workload.
B. Instrumentation and data acquisition
Despite many biological tissues generate their own elec-
tricity (endogenic bioelectricity), as the biopotentials associ-
ated to heart or neurons, bioelectrical impedance expresses
the passive electrical properties of the biological tissues. To be
measured, bioelectrical impedance uses exogenic electricity,
i.e., externally applied electricity. In other words, bioelectrical
impedance reflects the ability of a biological tissue to impede
(or oppose) the electric current flow.11
Due to its constitution, biological tissue can be regarded as
a conductor, a dielectric, or a combination of both.12,13 For low
frequencies of the exogenic electricity (<1 kHz), most of the
biological tissues can be seen as electrolytic conductors. For
high frequencies (>1 MHz), the dielectric properties dominate
the opposition offered by the biological tissue to the exogenic
electricity. In the middle range, both behaviors will determine
the type of opposition the biological tissue will present.13
(Re + Ri ) Rb Cm Ce + (Ri + Ro ) Rb Co Ce + [Ri Ro + (Ri + Ro ) Re ] Cm Co + (Ri + Ro ) Re Co Ce + Re Ri Cm Ce
C=
{[Ri Ro + (Ri + Ro ) Re ] Rb + Re Ri Ro } Cm Co Ce
Rev. Sci. Instrum. 88, 045105 (2017)
FIG. 1. Electrical model used to interpret the current response to a step voltage
excitation.
Such oppositions to the electric current flow are intrinsi-
cally associated to the electrochemical phenomena that take
place in the biological tissue.13,14 For this reason, bioelectrical
impedance represents a noninvasive experimental technique
that allows the investigation of many biological/physiological
behaviors of the live organisms.11
Bioelectrical impedance data were obtained by using a
bipolar method of spectroscopy based on the current response
to a voltage step (BISTEP), already used in a previous study.6
This technique can be considered equivalent to a multifre-
quency analysis from 1 kHz to 500 kHz because of the spectral
properties of the excitation signal, reducing the number of
interrogating signals applied to the biological system relative
to the classical sinusoidal sweep method.
The electrical model depicted in Figure 2 was consid-
ered to model the bioelectrical impedance data acquisition. It
is composed of a five-element model of the muscle9 and a
simple model for the electrode/tissue interface. The parame-
ters considered were as follows: electrode capacitance (Ce),
electrode resistance (Rb), extracellular resistance (Re), mem-
brane capacitance (Cm), intracellular resistance (Ri), organelle
resistance (Ro), and organelle capacitance (Co).
It is possible to demonstrate that in the Laplace domain,
where s = σ + jω (rd/s) and ω = 2πf (f in Hz), the current
response to a voltage step of magnitude V d can be written as
follows:
s2 + As + B
I(s) = Ip , (1)
s3 + Cs2 + Ds + E
where the constants below are dependent on the electrical
parameter of the model depicted on Figure 1 as
Vd
Ip =  Re Ri Ro
, (2)
Rb + Ri Ro +(Ri +Ro )Re
[(Re + Ri ) Cm + (Ri + Ro ) Co ]
A= , (3)
[(Ri Ro ) + (Ri + Ro ) Re ] Cm Co
1
B= , (4)
[Ri Ro + (Ri + Ro ) Re ] Cm Co
, (5)
045105-3 Jotta et al.
(Re + Ri ) Cm + (Ri + Ro ) Co + (Re + Rb ) Ce
D= . (6)
{[Ri Ro + (Ri + Ro ) Ri ] Rb + Re Ri Ro } Cm Co Ce
1
E= (7)
{[Ri Ro + (Ri + Ro ) Ri ] Rb + Re Ri Ro }Cm Co Ce
Equation (1) can be rewritten in a partial fraction as
" #
k1 k2 k3
I (s) = Ip + + , (8)
(s − p1) (s − p2) (s − p3)
where p1 , p2 , and p3 are the roots of the denominator of
Equation (1), i.e., the poles of I(s), and k 1 , k 2 , and k 3 are
the poles residuals.
Inverse Laplace transform of Equation (8) can lead to the
current response in time domain (Equation (9)). It is worth-
while to mention that the poles and the poles residuals are
calculated numerically,
f g
i (t) = Ip k1 ep1t + k2 ep2t + k2 ep3t . (9)
Experimental current response of the biological tissue was
compared to the theoretical counterpart (Equation (9)) in order
to estimate the electrical model parameters by an error min-
imization process between the experimental and theoretical
data. A program was developed in MATLAB 7 (Mathworks,
USA) to perform such electrical parameter estimates and used
the Matlab fminsearch function and an error function designed
to this purpose that considered the mean square error between
the experimental and theoretical responses. Electrode resis-
tance (Rb) was considered constant during the estimative
process.
Blood lactate concentration assessments were accom-
plished by an Accutrend automated lactate analyzer (Roche
Diagnostics, Switzerland), which has been considered in
agreement with gold standard report methods.15 Such lac-
tate concentration assessments were performed in the intervals
between the exercise stages, being paired with the acquisition
of bioelectrical impedance data. Blood samples were retrieved
from the fingers of the subjects. The first and second blood
drops were discarded and only the third drop was used in the
blood lactate concentration assessment. The Accutrend auto-
mated lactate analyzer uses one drop of the capillary blood
for each analysis. A blood drop is applied to a test strip with
standardized volume capacity. Then the test strip is inserted
into the lactate analyzer.
Estimates of the LT for each subject were determined
initially from the individual kinetics of the measured lactate
concentration through the Dmax method (LTDmax ), ordinarily
preferred in rowing due to its objectivity and reliability.16 The
Dmax method is characterized by identifying the power cor-
responding to the LT by computing the longest distance from
the curve describing the kinetics of the lactate concentration
and a straight line connecting the first and the last points of
such kinetics.16
Since each athlete presented a different value of LTDmax ,
in order to allow a better initial exploration of the kinetic asso-
ciated to the BIS parameters, a graphical normalization of the
exercise load was performed by dividing the workload values
by the value of the LTDmax . Therefore, after such normalization
a unit value corresponded to the LTDmax for all the volunteers.
Moreover, because each volunteer presented a different value
Rev. Sci. Instrum. 88, 045105 (2017)
for basal bioimpedance, we investigated only relative changes
of the electrical parameters calculated as unity minus the ratio
between the parameter for a given workload and the initial
value of such parameter.
Kinetics of all the BIS parameters in relation to the nor-
malized progressive workload were explored, by observing
original data as well as by the first and second derivative,
in order to identify the best parameter to be used in the LT
estimation by BIS.
An exploratory statistical analysis of LT estimates by
the invasive method of serial blood sampling and the Dmax
method (LTDmax ) and by the BIS method (LTBIS ) was per-
formed by medians and their respective quartiles. The agree-
ment between the two estimates was evaluated by a Bland
Altman analysis with limits of agreement set to two standard
deviations. The Pearson correlation coefficient (r) was calcu-
lated between LTDmax and LTBIS . Routines were developed in
MATLAB 7 (Mathworks, USA) for such data processing and
statistical analysis. We realized a t-test for the values of LTDmax
and LTBIS using a significance level of 0.05.
III. RESULTS
After an exploratory analysis of all parameters related
to the electrical model (Figure 1) and considering the nor-
malization procedure above-mentioned, two parameters (Ri,
and Ce) were selected for further analysis. These parameters
showed the most systematic behaviors in the normalized plots
for most of the subjects in this sample. For example, Ri param-
eter showed a decreasing kinetics in relation to the progression
of the workload (Figure 2(a)), while the parameter Ce showed
a remarkable increasing behavior (Figure 2(b)).
Among the two initially selected parameters, Ce showed
the most consistent kinetics for all the volunteers. Thus, such
parameter was elected the best one for the evaluation of the
changes in the slopes along the workload. On this issue,
Figure 3 shows the second derivative of the curve describ-
ing the kinetics associated to Ce, where one can see that most
of the time the zero crossing close to the normalized lactate
threshold (represented by the unity).
Based on the above-mentioned findings, the criterion
established for estimating the lactate threshold by BIS (LTBIS )
was searching the workload associated to the zero crossing
of the second derivative of the curve describing the kinetics
associated to the Ce parameter. Results of such an estimate
and the ones associated to the Dmax method (LTDmax ) were
compared by boxplots (Figure 4(b)) and by a Bland-Altman
plot (Figure 4(a)) comparing the two lactate threshold esti-
mation methods (LTDmax and LTBIS ), which showed a good
agreement between them (mean difference of 0.07 W) with
just two outliers for which probably the kinetics of the Ce
parameter is a little bit different from most of the volunteers.
However, even for these two volunteers, the LT estimate per-
formed by bioimpedance could be considered as a reasonable
workload for endurance training. The Pearson correlation coef-
ficient between the LTDmax and LTBIS estimates was r = 0.85
(p < 0.05) and the mean square error for all the 19 volunteers
was 17 W. The t-test for mean pointed to the absence of
045105-4 Jotta et al.
statistically significant difference (p = 0.98) between the
estimates LTDmax and LTBIS .
IV. DISCUSSION
An increasing pattern in relation to the progression of
exercise intensity was observed in the Ce parameter, which is
associated to the capacitive effect of the electrode-electrolyte
interface.13 In BIS applications where the changes on ionic
FIG. 3. Second derivative of Ce kinetics of rowers (9) during a progressive
workload test.
Rev. Sci. Instrum. 88, 045105 (2017)
FIG. 2. Example of the kinetics associated to the relative
change of electrical parameters Ri (a) and Ce (b) of rowers
(9 subjects) during a progressive workload test.
concentrations of the extracellular medium are considered
negligible during the course of the experiment, such elec-
trode capacitance can be considered constant. However, if the
experimental condition causes significant changes on ionic
concentration of the electrolytic medium, one can expect the
changes of such capacitance due to Gouy Chapman model
(Equation (10))13
Cárea= ε0 εs , (10)
λd
where ε s is the dielectric constant of the electrolytic solution;
ε 0 permittivity of free space, and λ d wavelength Debye,13
which can be expressed by
s
ε 0 ε s kT
λd = , (11)
2z2 q2 c
where k = Boltzmann constant; T = absolute temperature; z
= valence of the ion considered; q = electric charge, and c =
molar ion concentration.
As can be seen in Equations (10) and (11), electrode capac-
itance will vary in proportion to the square root of the ionic
concentration. Therefore, it can be hypothesized that increases
in the concentration of free ions considered relevant for con-
ductivity in extracellular compartments evoked a rise of charge
in the blood during the progress of exercise intensity and
elevated the capacitance electrode.
Changes in the concentration of the ions present in the
intra and extracellular compartments can explain the behav-
ior of some resistances and capacitances of the model used.
Thus, one can expect that changes in the slope of such ionic
changes may result on the slopes of the kinetics associated to
045105-5 Jotta et al.
the BIS parameters, and that this characteristic would allow the
development of a method for LT detection. An example is the
fact that the kinetics of blood [K+ ] presents a marked change
inflection regarding the progression of exercise intensity
related to LT occurrence. In this sense, some studies17,18
Rev. Sci. Instrum. 88, 045105 (2017)
FIG. 4. (a) Bland-Altman graphic comparing the esti-
mates LTDmax and LTBIS . (b) Boxplot of LTDmax and
LTBIS .
Dmax method to the lactate concentration data. The advan-
tage of the proposed method is its noninvasive aspect, which
can reduce problems associated to asepsis, contamination,
and reluctance of some athletes because of religious issues
associated to blood sampling. These aspects would allow the
have reported similar kinetics of [K+ ] and [La ] and inflec-
tions of the first almost coincident with the occurrence of the
ventilatory threshold (VT).
In this context, the findings of this study confirm the pos-
sibility of estimating the LT by BIS parameters evaluated on a
smaller proportion of the active muscles during exercise. These
findings are consolidated through the linear correlation coef-
ficient and the insignificance of the mean difference between
LTDmax and LTBIS obtained. These results gain particular rele-
vance if considering the application of the method proposed as
a reference for prescribing exercises for lower limbs and tests
performed during the sport season. Another relevant aspect is
the elimination of the invasive characteristic related to blood
measure through the evaluation of a minor proportion of mus-
cle in activity. In addition, the use of LTBIS in lower limbs
could help in assessing aspects of muscle synergism by giving
information about the metabolic state of the leg muscles. It
must be mentioned that as an indirect method the LT estimates
by bioelectrical impedance can be influenced if the subject is
in prolong fasting or exceeded in the consumption of food or
beverages before the LT assessment.
V. SUMMARY
The findings of this study confirm the feasibility of esti-
mating the LT by means of the BIS parameters associated to
a smaller proportion of the active muscles during exercise.
The results showed that such estimates are equivalent to the
ones obtained by a classical and well accepted method in the
sport field, i.e., serial blood sampling and application of the
application of the physiological assessment of athletes more
frequently, helping in the prescription of exercises and increas-
ing the athletic performance. Future studies using bioelectrical
impedance to estimate LT during running could expand the use
of this technique to the other sports.
1 L. V. Billat, “Use of blood lactate measurements for prediction of exercise
performance and for control of training,” Sports Med. 22, 157–175 (1996).
2 G. A. Brooks, “Anaerobic threshold: Review of the concept and directions
for future research,” Med. Sci. Sports Exercise 17, 22–34 (1985).
3 O. Faude, W. Kindermann, and T. Meyer, “Lactate threshold concepts,”
Sports Med. 39, 469–490 (2009).
4 Y. Li et al., “Maximal lactate steady state in kayaking,” Int. J. Sports Med.
35, 939–942 (2014).
5 R. Beneke, R. Hutler, and R. M. Leithauser, “Maximal lactate-steady-state
independent of performance,” Med. Sci. Sports Exercise 32, 1135–1139
(2000).
6 R. Dahmane et al., “Spatial fiber type distribution in normal human mus-
cle: Histochemical and tensiomyographical evaluation,” J Biomech. 38,
2451–2459 (2005).
7 R. L. Alvarenga and M. N. Souza, “Lactate thresholds by bioelectrical
impedance spectroscopy,” J. Exercise Physiol. 12, 22–33 (2009).
8 A. Lucı́a, O. Sánchez, A. Carvajal, and J. L. Chicharro, “Analysis of
the aerobic-anaerobic transition in elite cyclists during incremental exer-
cise with the use of electromyography,” Br. J. Sports Med. 33, 178–185
(1999).
9 B. Grassi, V. Quaresima, C. Marconi, M. Ferrari, and P. Cerretelli,
“Blood lactate accumulation and muscle deoxygenation during incremental
exercise,” J. Appl. Physiol. 87, 348–355 (1999).
10 E. M. Bartels, E. K. Sørensen, and A. P. Harrison, “Multi-frequency
bioimpedance in human muscle assessment,” Physiol. Rep. 3, e12354
(2015).
11 O. G. Martinsen and S. Grimnes, Bioimpedance and Bioelectricity Basics
(Academic, London, 2011), p. 1.
12 A. De Lorenzo, A. Andreoli, J. Matthie, and P. Withers, “Predicting body
cell mass with bioimpedance by using theoretical methods: A technological
review,” J. Appl. Physiol. 82, 1542–1558 (1997).
045105-6 Jotta et al.
13 W. Kuang and S. O. Nelson, “Low-frequency dielectric properties of bio-
logical tissues: A review with some new insights,” Trans. ASAE 41, 173
(1998).
14 C. A. Shiffman, R. Aaron, and S. B. Rutkove, “Electrical impedance of
muscle during isometric contraction,” Physiol. Meas. 24, 213 (2003).
15 C. Baldari et al., “Accuracy, reliability, linearity of Accutrend and Lactate
Pro versus EBIO plus analyzer,” Eur. J. Appl. Physiol. 107, 105–111 (2009).
Rev. Sci. Instrum. 88, 045105 (2017)
16 A. Erdogan et al., “Non-invasive indices for the estimation of the anaerobic
threshold of oarsmen,” J. Int. Med. Res. 38, 901–915 (2010).
17 P. Mcloughlin et al., “Exercise-induced changes in plasma potas-
sium and the ventilatory threshold in man,” J. Physiol. 479, 139–147
(1994).
18 S. Darabi et al., “Ventilation, potassium and lactate during incremental
exercise in men athletes,” Res. J. Biol. Sci. 4, 427–429 (2009).