Vol. 85, Issue 3, 1123-1134, September 1998
Effects of lung volume on diaphragm EMG signal strength during
voluntary contractions
Jennifer
Beck1,2,
Christer
Sinderby3,4,
Lars
Lindström5, and
Alex
Grassino1,2
1 Department of Physiology,
McGill University, Montreal, Quebec H3G 1Y6;
2 Pavillon Notre Dame, Centre
Hospitalier de l'Université de Montréal, Montreal, Quebec
H2L 4M1; 3 Guy-Bernier
Research Center, Maisonneuve-Rosemont Hospital, University of Montreal,
Montreal, Quebec, Canada H1T 2M4;
4 Institute for Clinical
Neuroscience, University of Gothenburg, and
5 Department of Medical
Informatics, Sahlgrens University Hospital, Gothenburg S-41345, Sweden
 |
ABSTRACT |
The use of
esophageal recordings of the diaphragm electromyogram (EMG) signal
strength to evaluate diaphragm activation during voluntary contractions
in humans has recently been criticized because of a possible artifact
created by changes in lung volume. Therefore, the first aim of this
study was to evaluate whether there is an artifactual influence of lung
volume on the strength of the diaphragm EMG during voluntary
contractions. The second aim was to measure the required changes in
activation for changes in lung volume at a given tension, i.e., the
volume-activation relationship of the diaphragm. Healthy subjects
(n = 6) performed contractions of the
diaphragm at different transdiaphragmatic pressure (Pdi) targets (range
20-160 cmH2O) while
maintaining chest wall configuration constant at different lung
volumes. The diaphragm EMG was recorded with a multiple-array
esophageal electrode, with control of signal contamination and
electrode positioning. The effects of lung volume on the EMG were
studied by comparing the crural diaphragm EMG root mean square (RMS),
an index of crural diaphragm activation, with an index of global
diaphragm activation obtained by normalizing Pdi to the maximum Pdi at
the given muscle length
(Pdi/Pdimax@L) at the
different lung volumes. We observed a direct relationship between RMS
and Pdi/Pdimax@L independent of diaphragm length. The volume-activation relationship of
the diaphragm was equally affected by changes in lung volume as the
volume-Pdi relationship (60% change from functional residual capacity
to total lung capacity). We conclude that the RMS of the diaphragm EMG
is not artifactually influenced by lung volume and can be used as a
reliable index of diaphragm activation. The volume-activation
relationship can be used to infer changes in the length-tension
relationship of the diaphragm at submaximal activation/contraction
levels.
diaphragm activation; chest wall configuration; length-tension
relationship; esophageal electromyogram electrode
| |
INTRODUCTION |
THE USE OF ESOPHAGEAL recordings of the diaphragm
electromyogram (EMG) signal strength to evaluate diaphragm activation
during voluntary contractions in humans has recently been criticized because of a possible artifact created by changes in lung volume (8,
13). In the context of the present study, diaphragm activation refers
to the combined processes of diaphragm motor unit recruitment and motor
unit firing rate and is used as a global term, with no emphasis placed
on differentiating between the two processes. The concept of "EMG
signal strength" refers to any measure used to quantify the power,
area, or amplitude of the EMG signal. In the present study we later use
the root mean square (RMS) to quantify the diaphragm EMG.
The rationale for inferring diaphragm activation from the strength of
the diaphragm EMG signal is that the signal constitutes a spatial and
temporal summation of action potentials from the recruited diaphragm
motor units and their firing rate (see APPENDIX A). However, it has been shown that the amplitude of
the electrically evoked diaphragm compound muscle action potential (CMAP) obtained with an esophageal electrode during constant neural drive (supramaximal stimulation) is altered with changes in lung volume
(11, 23). We previously showed that changes in lung volume influence
the power spectrum center frequency (CF) of the diaphragm CMAPs, but
not of the voluntary EMG signal (2). This suggests that one should not
infer changes in the voluntary EMG signal strength with lung volume on
the basis of the behavior of the CMAP signal. To our knowledge, the
influence of lung volume on the diaphragm EMG signal strength obtained
during voluntary contractions has not been studied.
The aim of this study was to evaluate whether there is an artifactual
influence of lung volume on the strength (RMS) of the diaphragm
interference pattern EMG during voluntary contractions. In the event
that there is no artifactual influence of lung volume on the diaphragm
EMG signal strength, the second aim of this study was to evaluate the
required changes in voluntary activation for changes in lung volume at
a given transdiaphragmatic pressure (Pdi), i.e., the volume-activation
relationship of the diaphragm.
| |
METHODS |
Approach for evaluating "artifactual" influence of lung
volume on diaphragm EMG signal strength.
In the present study the suggested artifactual influence of lung volume
on the diaphragm EMG signal strength was tested by comparing the RMS of
the diaphragm EMG with another control index of diaphragm activation at
various lung volumes. The control index of diaphragm activation was
obtained by normalizing Pdi to the maximum Pdi
(Pdimax) at the diaphragm length
(L) under consideration (Pdi/Pdimax@L). The
rationale for the relationship between
Pdi/Pdimax@L and
diaphragm activation is provided in APPENDIX
B. The relationship between the integrated EMG and
relative force (normalized at the given length) has previously been
demonstrated in the biceps (33). Furthermore, relationships between
some parameter of quantified EMG and isometric tension (no
length-tension influences) have been demonstrated in the diaphragm (14)
and other skeletal muscles (19). The relationship between respiratory
neural and mechanical outputs has been discussed extensively by Younes
and Riddle (35).
The crural diaphragm EMG measured with an esophageal electrode
represents the activation of a sample of the crural diaphragm, whereas
Pdi/Pdimax@L
represents global diaphragm activation. Consequently, the presence of a
relationship between RMS and
Pdi/Pdimax@L would
also suggest homogeneous activation of the costal and crural diaphragm.
Where appropriate, the distinction between crural diaphragm activation
(RMS) and global diaphragm activation
(Pdi/Pdimax@L) will be
made. (According to APPENDIX B,
Pdi/Pdimax@L is actually an index of relative global diaphragm activation, but
for didactic reasons we simply refer to global diaphragm activation.)
Rationale for volume-activation relationship of the diaphragm.
The classic concept of the length-tension relationship of skeletal
muscle is based on measurements of tension developed at various lengths
during constant activation of the muscle. The analogy of the
length-tension relationship for the human diaphragm is the Pdi
developed at different lengths (lung volume/chest wall configuration)
for maximal evoked or voluntary activation of the diaphragm, i.e., the
volume-Pdi relationship. In the present study, we wanted to measure the
changes in diaphragm activation at different lengths, during constant
Pdi, i.e., the volume-activation relationship.
For a given increase in lung volume, we expect a relative increase in
activation (to achieve a given Pdi), which should be of a magnitude
similar to the relative decrease in Pdi (for a constant activation).
With accurate measurements of activation, it should be possible to
obtain both of these relationships at submaximal activation/tension
levels. In the present study, the volume-Pdi relationship was obtained
by asking subjects to perform combined Müller-expulsive
Pdimax maneuvers at different lung volumes. The volume-activation curve was obtained by asking subjects to
perform submaximal contractions of the diaphragm while maintaining a
constant target Pdi at different lung volumes.
As reported in RESULTS, it was not
possible to reproduce the exact target Pdi values, and there was a
small variability in the actually achieved Pdi
(Pdiact) for the given target
Pdi levels with changes in lung volume. To avoid the influence of this
variability in Pdiact on the RMS,
we expressed Pdiact in relation to
the RMS (Pdiact/RMS) and used this
ratio as the "activation" component in the volume-activation
relationship. Because the variability in
Pdiact was small, this ratio would
be minimally influenced by the nonlinear relationship between RMS and
Pdi. RMS is the denominator in the expression
Pdiact/RMS, which expresses the reciprocal of the changes in diaphragm activation.
Subjects.
Six healthy men, all familiar with respiratory maneuvers, agreed to
participate in the study. Their age, height, and weight are presented
in Table 1. The experimental setup is shown
in Fig. 1.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 1.
Experimental setup. Left: esophageal
electrode consisted of 7 sequential pairs of electrodes, with a 10-mm
interelectrode distance. Middle:
Konno-Mead diagram (y-axis, rib cage
displacement; x-axis, abdominal
displacement) showing 4 chest wall configurations: functional residual
capacity on relaxation curve (FRC), one-third of inspiratory capacity
( IC), two-thirds of inspiratory capacity ( IC), and
total lung capacity (TLC). A feedback of target diaphragmatic pressure
(Pdi) was provided to subjects on an oscilloscope.
Right: on-line display of diaphragm
electromyogram (EMG) showing raw signals from all 7 electrode pairs,
center frequency (CF) values, and root-mean-square (RMS) values
available to investigator throughout experiment (allowed monitoring of
electrode positioning and RMS values during test). Pga, gastric
pressure.
|
|
Signal acquisition.
Diaphragm EMG signals were obtained via a multiple-array esophageal
electrode consisting of eight stainless steel rings (2 mm wide, 2 mm
diameter), placed 10 mm apart, creating an array of seven sequential
differential bipolar electrode pairs, mounted on silicone tubing (2 mm
diameter). A schematic representation of the electrode is presented in
Fig. 1, left. The most caudal pair of
rings was referred to as "electrode pair
1" and the most cephalad pair of rings as
"electrode pair 7." A two-lead
differential electrocardiogram (ECG) was obtained from electrodes
(model FC24, Graphic Controls) placed on the sternum, vertically and 10 cm apart, for later diaphragm EMG sample selection.
A Teflon tube was placed inside the silicone tubing (0.75 mm diameter),
and a 5-cm-long, 1.5-cm-diameter latex balloon was mounted ~5 cm
below the most distal EMG ring to allow for measurements of gastric
pressure (Pga). Esophageal pressure (Pes) was measured via a separate
catheter (1 mm diameter). The two balloon catheters were connected to a
differential pressure transducer (Validyne ±250
cmH2O) to yield Pdi, which was
displayed to the subject on a storage oscilloscope (model 1604, Gould;
Fig. 1, middle). The Pga catheter
was also connected to a separate differential pressure transducer and
referenced to atmosphere. Pdi and Pga were recorded on an eight-channel
strip chart recorder (model 35-V7808-12, Gould) and on magnetic tape
(model 4000A, Vetter). The signals were later acquired (model DT 2801A,
Data Translation) at a sampling frequency of 100 Hz (12-bit
resolution).
Diaphragm EMG signals from electrode pairs
1-7 and the ECG were amplified (model INA102,
Burr-Brown) and high-pass filtered at 10 Hz with an antialiasing filter
at 1,000 Hz (model D70L8L 8-pole Bessel filter, Frequency Devices).
Diaphragm EMG signals were acquired and digitized by an
analog-to-digital converter (model 2821, Data Translation), with 12-bit
resolution, at a sampling frequency of 2,000 Hz, and stored on hard
disk for off-line analysis. EMG signals from all seven electrode pairs
were displayed to the investigator on a computer monitor.
Lung volume was assessed throughout the experiment by the method of
Konno and Mead (16) (Fig. 1,
middle). Two respiratory inductive
plethysmography bands (Respitrace, Ambulatory Monitoring) were used to
evaluate rib cage (RC) and abdominal (AB) displacement. The RC band was
placed around the upper portion of the thorax, vertically centered over
the nipples; the upper edge of the AB band was placed around the
abdomen at the level of the umbilicus. The RC signals were amplified
and displayed on the vertical axis, and the AB signals on the
horizontal axis of a storage oscilloscope (model 5103N, Tektronix; Fig.
1, middle). The RC and AB signals were recorded on an eight-channel strip chart recorder (model 35-V7808-12, Gould).
Experimental protocol.
Subjects were studied while seated in an upright chair, facing the two
storage oscilloscopes: one for the lung volume feedback (Konno-Mead
diagram) and the other for target Pdi. Respitrace bands were positioned
on the subjects and secured in place by a surgical bandage placed over
the thorax. Subjects were trained (1 day before the experiment) to
perform isovolume maneuvers and relaxation curves with visual feedback
from the oscilloscope. Once they were familiar with these maneuvers,
subjects practiced reaching defined points on the Konno-Mead diagram
(Fig. 1, middle).
On the day of the experiment, the esophageal electrode was passed
through the nose, swallowed, and positioned at the level of the
gastroesophageal junction by feedback from an on-line display of the
diaphragm EMG signals from all seven electrode pairs on the computer
monitor. Once the diaphragm was located at the center of the electrode
array, the electrode was fixed externally at the nose. The Pes catheter
was then also passed through the nose and inflated, its position was
confirmed by the occlusion test (1), and then it was fixed at the nose.
After the catheters and the Respitrace bands were positioned, subjects
were asked to perform a relaxation maneuver from total lung capacity
(TLC) to functional residual capacity (FRC) and a series of isovolume maneuvers at FRC, 30% of inspiratory capacity (one-third of IC), 60%
of inspiratory capacity (two-thirds of IC), and TLC (Fig. 1,
middle). To ensure that posture and
the position of the Respitrace bands remained constant, these maneuvers
were repeated throughout the experiment.
In the present study, four lung volumes were evaluated along the
relaxation curve: FRC, one-third of IC, two-thirds of IC, and TLC (Fig.
1, middle). We are aware that, at a
given lung volume, different chest wall configurations and, hence,
different diaphragm lengths can be obtained, but in this study we
restricted the configurations to the relaxation curve of the Konno-Mead
diagram, and hence the changes in chest wall configuration are referred
to as changes in lung volume. With this setup, we could assume that
diaphragm shortening occurs from FRC to TLC (12).
Maximal voluntary Pdi maneuvers were performed at four different
diaphragm lengths, FRC, one-third of IC, two-thirds of IC, and TLC
(combined Müller-expulsive maneuvers), randomly throughout the
experiment, with rest periods between each attempt. The highest of
three attempts was considered to be maximal. The
Pdimax maneuver performed at any
given length was referred to as
Pdimax@L.
Subjects were asked to reach one of the four predetermined points on
the Konno-Mead diagram (marked on the oscilloscope), and, while keeping
the beam of the scope at the target point (i.e., maintaining the same
chest wall configuration), subjects performed a voluntary, static
contraction of the diaphragm while actively maintaining the target Pdi
(range 20-160 cmH2O). All
target Pdi swings were referenced to the resting Pdi at FRC. Each
contraction lasted 5-10 s and was repeated five to eight times for
each lung volume. A rest period was allowed between contractions
(1-5 min). We did not instruct the subjects on how to generate the
target Pdi levels; i.e., we did not control the relative contribution of Pga and Pes to Pdi. Although the target Pdi was fixed, the actual
Pdi generated may not necessarily have been the target Pdi. For
example, the target Pdi may have been 20 cmH2O for the four lung volumes,
but the actual Pdi generated by the subject may have been 18-22
cmH2O. Throughout this study, the
Pdi that was to be maintained by the subjects during the contractions
is referred to as the target Pdi. The Pdi achieved by the subject is
referred to as Pdiact.
Signal analysis.
Diaphragm EMG signals were automatically processed with computer
algorithms that eliminate the ECG, control for signal contamination (30), and neutralize signal filtering due to changes in bipolar electrode positioning with respect to the diaphragm by implementation of the double-subtraction technique (29).
Briefly, EMG segments are selected from all seven electrode pairs
between successive QRS complexes of the ECGs (R-R interval 50-75%) (30). With an array of bipolar electrodes (interelectrode distance 10 mm), the electrode pair closest to the center of the electrically active region of the diaphragm
(EARdi center) can be
determined by cross correlating the signals from every second pair of
electrodes (e.g., pair 1 vs.
pair 3, pair
2 vs. pair 4) (3).
EARdi center lies between
the two most negatively correlated electrode pairs. Once the
EARdi center is determined,
the signal from the electrode pair 10 mm caudal to
EARdi center is subtracted from the signal from the electrode pair 10 mm cephalad to
EARdi center. This algorithm
yields a new signal, the double-subtracted signal, that is minimized in
bipolar electrode filtering and enhanced in signal-to-noise (SN) ratio
(29).
From the double-subtracted signal, the direct-current (DC) level and
trends were removed by linear regression, and the first and last zero
crossings of the segment were then determined. To fit the segments for
a fast Fourier transform of 1,024 points, the tails of the diaphragm
EMG sample were zero padded from the outermost zero crossings to the
boundaries of the segment. The time-domain EMG segment was converted to
the frequency domain by fast Fourier transform, and the power spectrums
were calculated. The power spectrums were then evaluated by four signal
contamination indexes: the signal-to-motion artifact (SM) ratio, the SN
ratio, the drop in power of the spectrum (DP) ratio, and a spectral
deformation (
) ratio (30). SM, SN, and DP are expressed in decibels,
whereas is expressed in relative units. Only signals fulfilling the levels of SM 
12 dB, SN 15 dB, DP 30 dB, and 
1.4 are included in the analysis.
The RMS of the diaphragm EMG was calculated as RMS = (M0/p)1/2,
where p is the number of points in the
signal, zero padding excluded, and
M0 is the
spectral moment of order zero, and spectral moments (M) of order
n are obtained by
where
i is the index over which the power
density-frequency product is summed, i = 0 is the DC component, and
imax is the index
associated with the highest frequency in the spectrum.
M0 is inversely
proportional to propagation velocity of the action potentials (20), and
therefore the RMS is predicted to increase when propagation velocity
decreases, if activation remains constant. In the present study, the
RMS values were corrected for changes in propagation velocity that may
have occurred (especially at the higher contraction levels) by
normalizing the RMS values to a correction factor involving the power
spectrum CF, where CF = M1/M0
and M of order
n are obtained as described above. The corrected RMS = uncorrected RMS × (CF/CFrest)1/2.
Hereafter, the corrected RMS will simply be referred to as RMS. For
every lung volume and target Pdi that we studied, a mean (corrected) RMS value was calculated. Also, means ± SD were calculated for the
Pdiact swings.
| |
RESULTS |
Outcome of experimental protocol.
After a practice session, all subjects were able to coordinate the
respiratory muscles to achieve the target Pdi levels while maintaining
chest wall configuration constant. The outcome of the experimental
protocol is demonstrated for a representative subject in Fig.
2 and shows the response of the RMS with
lung volume for the various target Pdi values. Also, for a given lung volume, RMS values increased to achieve higher target Pdi values. In
all six subjects the Pdiact (for a
given target Pdi) varied little (Table 2)
at the four lung volumes (range 2.4 ± 1.3 cmH2O). The mean coefficient of
variation for the Pdiact for all
subjects was 3.1 ± 2.5%.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 2.
Outcome of experimental protocol in 1 subject. Response of RMS (left
y-axis) with lung volume
(x-axis) for various target Pdi values
in 1 representative subject. Each family of filled symbols (dashed
lines) represents target Pdi at which contractions were performed: 20 cmH2O ( ), 40 cmH2O ( ), 60 cmH2O ( ), 80 cmH2O ( ), and 100 cmH2O ( ).  , Reduction in
maximal Pdi values at a given length
(Pdimax@L) obtained in
this subject with increasing lung volume (right
y-axis, thick solid line). Data show
that, to generate a given target Pdi, RMS increases in inverse
proportion to Pdimax@L
with increasing lung volume (right
y-axis).
|
|
Influence of lung volume on Pdimax@L.
In all subjects, values of
Pdimax@L decreased
with increasing lung volume (Table 3). The
maneuvers were performed while subjects maintained chest wall
configuration constant, as observed on the Konno-Mead diagram display.
At TLC, Pdimax@L values dropped on average by 60% from FRC. The relationship between Pdimax@L and lung
volume for the group of six subjects was fairly linear.
Lung volume does not influence the relationship between RMS and
Pdi/Pdimax@L.
When the two indexes of activation were compared, RMS was directly
related to
Pdi/Pdimax@L, as
depicted for the six individual subjects in Fig.
3. These results also demonstrate that
there is no artifactual influence of lung volume on the diaphragm EMG
during voluntary contractions. In some subjects, outlier points were
observed (e.g., subjects 2, 3, and
6) and represent the RMS values
obtained at TLC, when the subjects were contracting the diaphragm at
the 20-cmH2O target Pdi.
Disregarding these outlier values, a single best curve fit was drawn
through the data points (from the origin) to visually emphasize the
relationship between RMS and
Pdi/Pdimax@L.
The decision to plot a curve and not a straight line through the data
points was based on the following.
1) Linear regression analysis of the
data indicated y-intercepts that were
different from zero (Table 4), and the relationship was expected to go through the origin, where there should
be no electrical activity when the diaphragm is not activated. 2) APPENDIX
A predicts that the RMS varies with the square root of activation, and hence the RMS is expected to
increase very little at increasing activation levels.
3) Comparison of the coefficients of
determination
(R2) obtained
with first- and second-order polynomial equations (Table 4) suggested a
slight but nonsignificant (P = 0.083) improvement in
R2 from a linear
curve fit to a second-order polynomial;
R2 increased on
average by 3.00 ± 3.41% (range 0-9%). The nonnormalized values of the RMS varied among subjects (note different scales for the
y-axes in Fig. 3) and are most likely
dependent on the anatomic differences among individuals (e.g., radial
muscle-to-electrode distance).
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 3.
There is no artifactual influence of lung volume on diaphragm EMG RMS.
Comparison of RMS (y-axis) and Pdi normalized
to Pdimax@L
(Pdi/Pdimax@L)
(x-axis) for subjects
1-6 (A-F)
shows no artifactual influence of lung volume on diaphragm EMG, as well
as a direct relationship between crural diaphragm activation and global
diaphragm activation, up to moderate activation levels. Arrows,
distinct outlier points.
|
|
Crural diaphragm activation in response to diaphragm shortening.
Figure 4A
demonstrates in subject 5 the
curvilinearity of the relationship between the Pdi and the RMS at
different lung volumes. The best curve fit obtained for each lung
volume has been drawn for visual clarity. Figure
4B demonstrates the relationship
between Pdiact/RMS and lung
volume for the various target Pdi values. At a given lung volume
Pdiact/RMS showed a large
variability, as presented for the six subjects in Table
5. (All data points, including the outlier
values in Fig. 3, are included in this analysis.) Figure
4C shows the relationship between
Pdiact and RMS with changes in
lung volume when the data were normalized to the FRC value, with
correction for differences in the intercepts. The slopes of the
reduction in Pdiact/RMS with
increasing lung volume were similar for the different target Pdi
values.
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 4.
Crural diaphragm activation in response to diaphragm shortening.
A: in subject
5, relationship between Pdi actually achieved
(Pdiact) and RMS at different
lung volumes in 1 subject. Pdiact/RMS relationship at
different muscle lengths is curvilinear. Best curve fit obtained for
each lung volume has been drawn for visual clarity.
B:
Pdiact/RMS with changes in lung
volume for different target Pdi values.
C: relative change in
Pdiact/RMS for different target
Pdi values plotted vs. lung volume.
|
|
In subject 5 the relative reduction in
Pdiact/RMS was similar to the
relative change in
Pdimax@L with
increasing lung volume (Fig. 4C).
Figure 5 shows the group mean relative change in Pdiact/RMS obtained
for all target Pdi values and
Pdimax@L at the four
different lung volumes, according to the analysis performed in Fig.
4C. The three outlier values from Fig.
3 (indicated by arrows, TLC, 20 cmH2O) are included in this
analysis, which reduced
Pdiact/RMS at TLC, and in part
explain the slight deviation away from the line of identity (dashed
line). For this group of six subjects, we found a proportional
relationship between the volume-activation and volume-Pdi relationship.
The data suggest that, with an increase in lung volume equal to 33% of
IC, there is a 20% reduction in
Pdimax@L (the
volume-Pdi relationship) and a 20% decrease in
Pdiact/RMS.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 5.
Identity plot of relative changes in
Pdiact/RMS and
Pdimax@L. Plot
describes relationship between
Pdimax@L and
Pdiact/RMS at different lung
volumes (group mean data). n, No. of subjects.
|
|
| |
DISCUSSION |
The results of this study demonstrate that
1) there is no artifactual effect of
lung volume on the diaphragm EMG signal strength during voluntary
contractions, 2) crural diaphragm
activation, as measured by the RMS of the diaphragm EMG, is related to
global diaphragm activation, inferred from
Pdi/Pdimax@L, and
3) the volume-activation
relationship can be used to infer changes in the length-tension
relationship of the diaphragm at submaximal diaphragm
activation/contraction levels.
Critique of Pdi/Pdimax@L as an index of
global diaphragm activation.
In the present study, global diaphragm activation was inferred by the
relative Pdi, which was calculated by normalizing the Pdi to the
maximum Pdi obtained at the given diaphragm length (Pdi/Pdimax@L).
Reliability of this index of activation was dependent on the target
contractions and
Pdimax@L maneuvers
being performed at the same muscle length. By controlling the
maintenance of chest wall configuration during the target contractions
and the Pdimax@L
maneuvers, it was assumed that changes in diaphragm length were kept to
a minimum during these maneuvers (16). Feedback from the Konno-Mead
diagram also ensured that the contractions were always repeated at the same target lung volume without displacement of the chest wall.
Another limitation of using the
Pdimax@L for
normalization is that the maneuver is voluntary and depends on the subject's motivation and experience. In the present study, the
Pdimax@L values at FRC
(from a combined Müller-expulsive maneuver performed with
feedback) were well within the range that has been reported in
healthy subjects (18). With respect to the higher lung volumes, we
observed a relative reduction in
Pdimax@L similar to
that obtained by others, whether elicited voluntarily or by electrical
or magnetic phrenic nerve stimulation (6, 15, 17, 32). Previous
investigators used the twitch interpolation technique to show that
subjects who are experienced in performing Pdimax maneuvers are able to
voluntarily and maximally recruit their diaphragm at FRC (5) and at
lung volumes above FRC (22).
Although there are limitations to using
Pdi/Pdimax@L as an
index of global diaphragm activation, we believed it was the most
suitable for comparison to the RMS of the diaphragm EMG. Other proposed
measures of activation, such as the single-twitch interpolation
technique, provide information about maximal diaphragm motor unit
recruitment, and not diaphragm motor unit firing rate, and hence this
technique provides only one part of the information related to
diaphragm activation. Another method that has been used to evaluate
diaphragm activation has been to look at the firing rate of single
motor unit action potentials recorded with needle electrodes (8). For
evaluation of diaphragm activation in the present study, this technique
would not have been suitable, because the motor unit action potentials
would not have been discernible at higher contraction levels, and
again, motor unit firing rate provides only part of the information
related to activation. Therefore, for the purpose of the evaluation of
the EMG as a measure of activation at different lung volumes, we chose
Pdi/Pdimax@L as the
most appropriate index of global diaphragm activation.
Critique of diaphragm EMG RMS as an index of crural diaphragm
activation.
Besides the possible influence of lung volume on the voluntary EMG
signal, which was ruled out in the present study, several other factors
can artifactually influence the RMS of the diaphragm EMG power spectrum
(for isometric, constant-force contractions). Most importantly, the EMG
signal strength is dependent on accurate methodology for acquisition
and analysis, as well as electrode configuration and positioning. The
RMS is also affected by factors that influence signal quality, such as
cardiac activity, esophageal peristalsis, external noise, electrode
motion artifacts, and aliasing. These issues have previously been
discussed in detail, and methods are now available to overcome the
artifactual influences (3, 4, 29, 30).
Another factor that can affect the RMS value is a reduction in the
conduction velocity (CV) of the muscle fiber action potentials, which
could occur during sustained forceful contractions or with reductions
in temperature. The spectral moment of order zero, used in calculations
of the RMS, is inversely proportional to CV (20), and hence reductions
in CV will result in increases in the RMS values that are unrelated to
changes in muscle activation. In the present study, a fatigue
correction factor involving the CF was calculated for each lung volume
and each target contraction level (we assumed that changes in CF
paralleled changes in CV). The use of CF in the correction of the RMS
(for changes in CV) requires that CF not itself be artifactually
influenced by lung volume, as was demonstrated by Beck et al. (2, 4),
or other artifacts such as signal quality (30) or electrode positioning (3, 4). A lack of correction for changes in CV would result in
overestimations of the RMS value.
Theoretically, a limitation of the RMS as an index of crural diaphragm
activation is that the RMS is predicted to increase less than
activation at very high firing rates and/or very high numbers
of recruited motor units (see APPENDIX
A). With respect to the experimental data, we found in
most subjects a fairly linear relationship between the RMS and the
global diaphragm activation up to ~75% of maximum activation levels,
indicating that underestimation of the RMS to reflect activation may
occur above this level.
One other limitation of the use of the EMG signal strength is that
because of anatomic and physiological differences, nonnormalized RMS
values cannot be used to compare activation levels among different subjects. Various techniques have been presented to obtain a reference value for normalizing the RMS, such as magnitudes of increase from
resting or supine levels (9, 10), percentage of value obtained during
an inspiration to TLC (14) or a maximal Pdi maneuver (34), or simply
normalization of the RMS to the highest value obtained at any time.
None of these investigators, however, provided any rationale or
evaluation to justify their reference value. It has been suggested that
a simple inspiration to TLC is suitable to obtain a maximum voluntary
RMS value for normalization purposes (unpublished observations). One
must keep in mind that because of the limited increase in RMS at high
activation levels, the maximum RMS may actually be an underestimation
of the true maximal activation, providing overestimations in the
relative RMS.
By using the crural diaphragm EMG measured with an esophageal
electrode, signals are obtained from an area of unknown size located in
the region of the crural diaphragm. Hence, the crural diaphragm EMG is
only a sample of the entire diaphragmatic motor unit population. The
findings of the present study, however, show that similar activation
levels obtained with various forces at different lengths provide
similar changes in crural diaphragm EMG and
Pdi/Pdimax@L. This
indicates that a sample measurement of activated motor units in the
crural diaphragm has the potential to estimate changes in activation of
the whole diaphragm.
We are aware that our results were obtained under strict experimental
conditions, i.e., during isometric contractions of the diaphragm
performed at four defined lung volumes along the relaxation curve.
However, subjects had the freedom to choose the Pga and Pes
contributions to Pdi, and our results of a direct relationship between
the RMS and
Pdi/Pdimax@L were the
same for subject 3, who mainly
performed the maneuvers by changing Pga, and for
subject 2, who mainly changed Pes. We
observed that, at an extreme chest wall configuration such as TLC,
three subjects showed outlier values in the
RMS-Pdi/Pdimax@L relationship when the target Pdi was 20 cmH2O, suggesting deviations in
the relationship between crural diaphragm activation and global diaphragm activation at this extreme diaphragm length; however, the
data points related to TLC when the target Pdi was 40
cmH2O behaved as expected.
In conclusion, the use of the crural diaphragm EMG RMS value to infer
global diaphragm activation can be considered to be valid at
physiological activation levels and at nonextreme chest wall
configurations. Our findings are in agreement with previous investigators' reports on homogeneity between costal and crural diaphragm activation during breathing (7, 21, 24, 27, 28).
Effects of lung volume on interference pattern EMG signal strength.
The close relation between the diaphragm EMG and global diaphragm
activation contradicts previous statements that the voluntary diaphragm
EMG is systematically affected by changes in lung volume (8, 13).
Because of the design of the study, where a range of activation levels
was evaluated at different lung volumes, it would not have been
possible to obtain a relationship between the
Pdi/Pdimax@L and the
EMG RMS, two indexes of activation, if there had been a systematic
effect of lung volume on the RMS values. The findings of the present
study are also in agreement with a previous study, where we could not
demonstrate any lung volume-related changes in the frequency content of
the voluntary EMG (2).
Previous allegations (8, 13) about the inaccuracy of using diaphragm
EMG to assess diaphragm activation have been based on findings that
diaphragm CMAP amplitude is affected by lung volume (11, 23). The
diaphragm CMAP represents the summated synchronized electrical activity
generated by all motor units after a supramaximal stimulus of the
phrenic nerve(s). The voluntary signal represents the summated
electrical activity generated by asynchronously firing crural diaphragm
motor units. Because of the fundamental differences in signal
characteristics of the CMAP and the voluntary signal, in this study and
in a previous study, we have clearly demonstrated that the behavior of
the synchronized CMAP signal cannot be used to infer the behavior of
the voluntary EMG signal. With respect to changes in lung volume, we
previously (2) discussed the possible reasons for the differences in
behavior of the two signals (in terms of their frequency content). The results of the present study did not provide any additional information describing the differences between the two types of signals, but we
have confirmed that the voluntary EMG and CMAP signals behave differently with respect to changes in lung volume. We therefore conclude that the voluntary diaphragm EMG signal, when acquired and
analyzed with appropriate methodology, is adequate for evaluation of
diaphragm activation, under conditions where diaphragm length changes
are expected to occur or in patients with chronic obstructive pulmonary
disease who are hyperinflated because of expiratory flow limitation.
Volume-activation and length-tension relationships.
We have demonstrated that, given the conditions of the present study
and within a physiological range of activation levels, RMS ~ Pdi/Pdimax@L.
Therefore, the volume-activation relationship, which is the activation
required to generate a given Pdi, can be expressed as
1) RMS/Pdi, where RMS is the index
used to infer activation and Pdi is the targeted Pdi, or
2)
(Pdi/Pdimax@L)/Pdi, where Pdi/Pdimax@L is
the index used to infer activation.
(Pdi/Pdimax@L)/Pdi can
be simplified to
Pdimax@L. Therefore,
RMS/Pdi ~ Pdimax@L. With changes in lung volume, RMS/Pdi is a representation of the activation needed to generate a given Pdi, and
Pdimax@L is the
pressure response for maximum voluntary activation. With increasing
lung volume, the RMS increases for a given Pdi, whereas Pdimax@L is reduced,
resulting in a reciprocal relationship between the volume-activation
and volume-Pdi relationship. In summary, measurements of activation
required to generate a given Pdi or measurements of the Pdi generated
for a fixed activation level can provide information about the
length-tension relationship of the diaphragm at submaximal or maximal
levels of activation/contraction.
Implications of a curvilinear relationship between RMS and Pdi.
In the present study (Fig. 4A), as
well as in previous studies (14), it has been demonstrated that the
relationship between RMS and Pdi is curvilinear. The implication of a
nonlinear relationship is that
Pdiact/RMS could vary
severalfold at a given lung volume (Table 5) by varying the target Pdi.
The example in Fig. 4B shows that a
change in target Pdi from 20 to 120 cmH2O produces the same change in
Pdiact/RMS obtained with maximal
diaphragm shortening. By keeping Pdi or RMS constant (as constant as
possible), the influence of this curvilinearity on the Pdi/RMS will
be reduced. This approach is demonstrated in Figs.
4C and 5, where we show that the
relative change in the ratio for a given Pdi can provide information
about changes in the volume-activation relationship, which in turn can
be used to infer relative changes in the length-tension relationship.
Because of the curvilinearity, one should be cautious in using the
ratio of RMS to Pdi or Pdi to RMS as a measurement of diaphragm
efficiency or effectiveness.
Conclusion.
The findings of the present study justify theoretically and
experimentally the use of the crural diaphragm EMG RMS to continuously monitor diaphragm activation in humans. We have, in particular, demonstrated that the RMS value of the voluntary diaphragm EMG is not
artifactually influenced by changes in lung volume. The diaphragm EMG
is different from other methods to evaluate diaphragm activation; it
provides information about the combined processes of firing rate and
motor unit recruitment. The volume-activation relationship (diaphragm
activation required to generate a given force with increasing lung
volume) can be used to infer the length-tension relationship of the
diaphragm (diaphragm force response to a given level of activation at
different lengths) at submaximal activation/contraction levels.
| |
APPENDIX A |
Theoretical Description of the Relationship Between Activation and EMG
Signal Strength
Myoelectric properties.
Consider a muscle (or the part of a muscle that contributes to the
observed myoelectric signal) containing a number
(M) of active motor units each with
a repetition rate equal to
1/TR that, for
simplicity, is assumed to have the same value in the mean for all the
units. We introduce the concept of the intensity (I) of the muscle's
electrical activity
|
(A1)
|
The
contributions from one motor unit are assumed to be uncorrelated with
the contributions from the other units. The individual motor unit
contributions are characterized by the action potential duration
(TD) and a
strength measure, which will emerge from the calculations below.
Statistical preliminaries.
We introduce the amplitude density function
fz(z),
which describes the differential probability of finding a certain value
(in our case, the voltage of the signal)
z of a random variable
z. The density function is related to the distribution function
Fz(z),
which is the integral over
fz(z)
|
(A2)
|
If
fz(z)
is normalized (the area under the density curve attains unity), the
function
Fz(z) can also be expressed as the probability
(P) of finding the variable z below the level
z
|
(A3)
|
The
density function is useful, since the expected value of the random
variable z is
|
(A4)
|
which
is also known as the DC value. Furthermore, the expected value of the
square of the variable z is
|
(A5)
|
which
also is the square of the RMS value. The variance of the variable
z is related to the mentioned
quantities as follows
|
(A6)
|
The
above expressions, with statistical basis, explain why the RMS is
frequently used to describe signal properties. It is interesting to
compare the characteristics of the RMS value with another measure of
signal strength, the full-wave-rectified and averaged value (FRA). The
FRA value can be obtained from the amplitude density function as
follows
|
(A7)
|
This
represents a rectifier characteristic that is linear but with opposite
signs for positive and negative values of the input signal. Many signal
strength-measuring devices are actually measuring the FRA value but are
calibrated to show the RMS value. This can only be true for one
particular signal waveform; mostly, it is assumed that the signal has a
sinusoidal shape.
Signal summation.
To further clarify the properties of the density function, consider an
action potential of duration
TD that occurs
once in the observation interval of length
T0 (Fig.
6). The value of
fz(z) is proportional to the relative time the signal spends at a certain level, and thus the area under the peak of
fz(z) at zero value of z is proportional to
T0 
TD. We can thus
write the expression for the density function as
|
(A8)
|
where
we have split the density function into one part describing the action
potential (p) as such and one part describing the silent
interval outside the potential [voltage level zero, modeled by
the Dirac delta function,
(z)].
View larger version (10K):
[in this window]
[in a new window]
|
Fig. 6.
Relation between signal as a function of time and amplitude density
function. Left: action potential in
time domain, observed in observation interval
(T0).
Right: amplitude density function
fz(z)
of signal voltage (z). au, Arbitrary
units.
|
|
The signal summation of randomly distributed contributions can be
performed as follows. A fundamental theorem concerning the sum of two
random variables
|
(A9)
|
states
that the amplitude density function of the sum equals the convolution
of the densities of the contributing signals, which are assumed to be
independent (25)
|
(A10)
|
The
sum of two motor unit action potentials, denoted
x and
y, of the same shape and duration
(TD) randomly
occurring in the observation interval
T0 is obtained by
combining Eqs. A8 and A10. If the observation interval is
long in comparison with the duration of the potentials, the likelihood
of obtaining overlap between the potentials is low. Also, the main
contribution to the convolution of
fx(x)
and
fy(y)
comes from the (x) and
(y) functions. With these assumptions the
combination of Eqs. A8 and A10 can be
approximated
|
(A11)
|
which is further simplified to
|
(A12)
|
Generalization
to N potentials, still with the
assumption that they occupy only a small fraction of the observation
interval, results in the approximation
|
(A13)
|
Increasing
the number of sources so that they start to overlap, we find that the
density function will lose the peak
[(z)] at zero level
and begin to diffuse into a more spread-out shape. In the case of a
very large number of randomly distributed sources, the shape will
become Gaussian (with the interesting property that the convolution
between 2 Gaussian functions yields another Gaussian function). The
summation in this case can be illustrated by two Gaussian
distributions, denoted x and
y, with zero mean and variances
2x and
2y, respectively. Thus, for
the sum
![<IT>f</IT><SUB><B>z</B></SUB>(<IT>z</IT>) = <LIM><OP>∫</OP><LL>−∞</LL><UL>+∞</UL></LIM> <FR><NU>exp [−<IT>x</IT><SUP>2</SUP>/2&sfgr;<SUP>2</SUP><SUB><IT>x</IT></SUB>]</NU><DE>(2&pgr;)<SUP>1/2</SUP>&sfgr;<SUB><IT>x</IT></SUB></DE></FR> · <FR><NU>exp [−(<IT>z</IT> − <IT>x</IT>)<SUP>2</SUP>/2&sfgr;<SUP>2</SUP><SUB><IT>y</IT></SUB>]</NU><DE>(2&pgr;)<SUP>1/2</SUP>&sfgr;<SUB><IT>y</IT></SUB></DE></FR> d<IT>x</IT>](/content/vol85/issue3/fulltext/1123/img019.gif)
|
(A14)
|
which,
after some calculations, is reduced to
![<IT>f</IT><SUB><B>z</B></SUB>(<IT>z</IT>) = [2&pgr; · (&sfgr;<SUP>2</SUP><SUB><IT>x</IT></SUB> + &sfgr;<SUP>2</SUP><SUB><IT>y</IT></SUB>)]<SUP>−1/2</SUP> · exp [−<IT>z</IT><SUP>2</SUP>/2(&sfgr;<SUP>2</SUP><SUB><IT>x</IT></SUB> + &sfgr;<SUP>2</SUP><SUB><IT>y</IT></SUB>)]](/content/vol85/issue3/fulltext/1123/img020.gif)
|
(A15)
|
i.e.,
a new Gaussian density function with zero mean and variance
|
(A16)
|
For
equal variances of the contributing signals the resulting variance is
|
(A17)
|
Repeated
use of the procedure shows that for N
contributions the variance of the resulting Gaussian density function
is
|
(A18)
|
Myoelectric processes and measures.
At low levels of muscle activity, i.e., low motor unit potential
repetition rate and low degree of motor unit recruitment, the
probability of having overlapping action potentials is low. A slight
increase in repetition rate or number of recruited motor units,
therefore, will add potentials that most likely are nonoverlapping. If
we make the observation interval equal to the mean repetition rate of
the motor units, the intensity (I) according to Eq. A1 will be the relevant quantity to describe the muscle
activity at low contraction levels.
Thus, using Eq. A13, with the
assumption that the individual action potential duration
TD is negligible
in comparison with the length of the observation interval, now
considered equal to the repetition interval, and further observing that
zero level peak (z) makes no
contribution to the signal strength measure, we find the following
expressions for the RMS and FRA measures at low rates of
repetition and recruitment
|
(A19)
|
|
(A20)
|
or,
more condensed
|
(A21)
|
At
high levels of muscle activity, i.e., high motor unit repetition rates
and a large number of recruited motor units, the probability of
overlapping potentials is very high: the total signal has the character
of random (Gaussian) noise. Any of the two processes of recruitment or
repetition rate increase will add potentials randomly overlapping other
potentials. Thus also in this case the product of the mean repetition
rate and number of recruited units, i.e., the intensity, is the
relevant quantity to describe the muscle's activity. In the case of a
Gaussian density function, we find, from the equations above, the
following expressions for the RMS and FRA values
|
(A22)
|
|
(A23)
|
which immediately show
that
|
(A24)
|
It
should be observed that in the above derivation of the expressions for
RMS and FRA a dependence between recruitment order and motor unit size
has not been taken into account.
A simulation of potential summation is shown in Fig.
7, where the RMS and FRA values are plotted
as functions of the intensity (I). We observe a consistent difference
(in the log scale) between the RMS and FRA values for high intensities
in accordance with the theory which states that the quotient between
the RMS and FRA (the form factor) values should attain a value of
(
/2)1/2 
1.2 for Gaussian
noise signals. Deviations from this value indicate the presence of
non-Gaussian signals.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 7.
Signal strength measurements as a function of intensity of muscle
activity. Plot of RMS and full rectified average (FRA,
y-axis) as a function of intensity
(x-axis) in log scale.
|
|
| |
APPENDIX B |
Theoretical Explanation for the Relationship Between
Pdi/Pdimax @L and Global Diaphragm
Activation
When all factors affecting the contractile properties of muscle are
constant, any change in neural drive to the muscle (i.e., activation)
will provide a change in muscle force. During nonfatiguing, isometric
contractions, one can assume that
|
(B1)
|
where
kL is a constant
related to muscle length (L) and
activation refers to the combined processes of diaphragm motor unit recruitment and motor unit firing rate. For the diaphragm, therefore, to generate pressure (force) at a given muscle length and radius of
curvature [inferred from chest wall configuration (CWC)]
|
(B2)
|
and
|
(B3)
|
where
maximum refers to a maximum voluntary (100% of max) Pdi or activation.
In other words
|
(B4)
|
If
we want
Pdi/Pdi(100% of max)
|
(B5)
|
Therefore
|
(B6)
|
and
is a measure of relative global diaphragm activation.
| |
ACKNOWLEDGEMENTS |
J. Beck was supported by the Fonds pour la formation de Chercheurs
et l'Aide à la Recherche (FCAR Quebec) and the McGill University
Faculty of Medicine Internal Studentship. C. Sinderby is a recipient of
a fellowship from The Parker B. Francis Families Foundation. We also
acknowledge the support of Inspiraplex of The Respiratory Health
Network of Centres of Excellence, The Medical Research Council of
Canada, The Swedish Association for Traffic and Polio Disabled, and The
Swedish Association for the Neurologically Disabled.
| |
FOOTNOTES |
Address for reprint requests: J. Beck, Pulmonary Function Lab, Pavillon
Notre-Dame, CHUM, 1560 Sherbrooke St. East, I-2158, Montreal, Quebec,
Canada H2L 4M1.
Received 26 September 1997; accepted in final form 27 April 1998.
| |
REFERENCES |
1.
Baydur, A.,
P. K. Behrakis,
W. A. Zin,
M. Jaeger,
and
J. Milic-Emili.
A simple method for assessing the validity of the esophageal balloon technique.
Am. Rev. Respir. Dis.
126:
788-791,
1982[Medline].
2.
Beck, J.,
C. Sinderby,
L. Lindström,
and
A. Grassino.
Diaphragm interference pattern EMG and compound muscle action potentials: effects of chest wall configuration.
J. Appl. Physiol.
82:
520-530,
1997[Abstract/Free Full Text].
3.
Beck, J.,
C. Sinderby,
L. Lindström,
and
A. E. Grassino.
Influence of bipolar esophageal electrode positioning on measurements of human crural diaphragm EMG.
J. Appl. Physiol.
81:
1434-1449,
1996[Abstract/Free Full Text].
4.
Beck, J.,
C. Sinderby,
J. Weinberg,
and
A. E. Grassino.
Effects of muscle-to-electrode distance on the human diaphragm electromyogram.
J. Appl. Physiol.
79:
975-985,
1995[Abstract/Free Full Text].
5.
Bellemare, F.,
and
B. Bigland-Ritchie.
Assessment of human diaphragm strength and activation using phrenic nerve stimulation.
Respir. Physiol.
58:
263-277,
1984[Medline].
6.
Braun, N.,
N. Arora,
and
D. F. Rochester.
Force-length relationship of the normal human diaphragm.
J. Appl. Physiol.
53:
405-412,
1982[Abstract/Free Full Text].
7.
Cooke, I. R.,
M. Soust,
and
P. J. Berger.
Differential recruitment of inspiratory muscles in response to chemical drive.
Respir. Physiol.
92:
167-181,
1993[Medline].
8.
DeTroyer, A.,
B. Leeper,
D. K. McKenzie,
and
S. C. Gandevia.
Neural drive to the diaphragm in patients with severe COPD.
Am. J. Respir. Crit. Care Med.
155:
1335-1340,
1997[Abstract].
9.
Druz, W. S.,
and
J. T. Sharp.
Activity of respiratory muscles in upright and recumbent muscles.
J. Appl. Physiol.
51:
1552-1561,
1981[Abstract/Free Full Text].
10.
Druz, W. S.,
and
J. T. Sharp.
Electrical and mechanical activity of the diaphragm accompanying body position in severe chronic obstructive pulmonary disease.
Am. Rev. Respir. Dis.
125:
275-289,
1982[Medline].
11.
Gandevia, S. C.,
and
D. K. McKenzie.
Human diaphragmatic EMG: changes with lung volume and posture during supramaximal phrenic nerve stimulation.
J. Appl. Physiol.
60:
1420-1428,
1986[Abstract/Free Full Text].
12.
Gauthier, A. P.,
S. Verbanck,
M. Estenne,
C. Segebarth,
P. T. Macklem,
and
M. Paiva.
Three-dimensional reconstruction of the in vivo diaphragm shape at different lung volumes.
J. Appl. Physiol.
76:
495-506,
1994[Abstract/Free Full Text].
13.
Georgopoulos, D.,
I. Mitrouska,
K. Webster,
Z. Bshouty,
and
M. Younes.
Effects of inspiratory muscle unloading on the response of respiratory motor output to CO2.
Am. J. Respir. Crit. Care Med.
155:
2000-2009,
1997[Abstract].
14.
Grassino, A.,
M. D. Goldman,
J. Mead,
and
T. A. Sears.
Mechanics of the human diaphragm during voluntary contractions: statics.
J. Appl. Physiol.
44:
829-839,
1978[Abstract/Free Full Text].
15.
Hershenshorn, M. B.,
Y. Kikuchi,
and
S. H. Loring.
Relative strengths of the chest wall muscles.
J. Appl. Physiol.
65:
852-862,
1988[Abstract/Free Full Text].
16.
Konno, K.,
and
J. Mead.
Measurement of the separate volume changes in rib cage and abdomen during breathing.
J. Appl. Physiol.
22:
407-422,
1967[Free Full Text].
17.
Laghi, F.,
M. J. Harrison,
and
M. J. Tobin.
Comparison of magnetic and electrical phrenic nerve stimulation in assessment of diaphragm contractility.
J. Appl. Physiol.
80:
1731-1742,
1996[Abstract/Free Full Text].
18.
Laporta, D.,
and
A. Grassino.
Assessment of transdiaphragmatic pressure in humans.
J. Appl. Physiol.
58:
1469-1476,
1985[Abstract/Free Full Text].
19.
Lawrence, J. H.,
and
C. J. DeLuca.
Myoelectric signal versus force relationship in different human muscles.
J. Appl. Physiol.
54:
1653-1659,
1983[Abstract/Free Full Text].
20.
Lindström, L.,
and
I. Petersén.
Power spectrum analysis of EMG signals and its applications.
In: Progress in Clinical Neurophysiology. Computer-Aided Electromyography, edited by J. E. Desmedt. Basel: Karger, 1983, vol. 10, p. 1-51.
21.
Lourenco, R. V.,
N. S. Cherniack,
J. R. Malm,
and
A. P. Fishman.
Nervous output from the respiratory centers during obstructed breathing.
J. Appl. Physiol.
21:
527-533,
1966[Free Full Text].
22.
McKenzie, D. K.,
G. M. Allen,
and
S. C. Gandevia.
Reduced voluntary drive to the human diaphragm at low lung volumes.
Respir. Physiol.
105:
69-76,
1996[Medline].
23.
McKenzie, D. K.,
and
S. C. Gandevia.
Phrenic nerve conduction times and twitch pressures of the human diaphragm.
J. Appl. Physiol.
58:
1496-1504,
1985[Abstract/Free Full Text].
24.
Oyer, L.,
S. L. Knuth,
D. K. Ward,
and
D. Bartlett.
Patterns of neural and muscular electrical activity in costal and crural portions of the diaphragm.
J. Appl. Physiol.
66:
2092-2100,
1989[Abstract/Free Full Text].
25.
Papoulis, A.
Functions of two random variables.
In: Probability, Random Variables, and Stochastic Processes. New York: McGraw-Hill, 1965, chapt. 7, p. 187-190.
26.
Person, R. S.,
and
M. S. Libkind.
Simulation of electromyograms showing interference patterns.
Electroencephalogr. Clin. Neurophysiol.
28:
625-636,
1969.
27.
Petit, J. M.,
G. Milic-Emili,
and
L. Delhez.
Role of the diaphragm in breathing in conscious normal man: an electromyographic study.
J. Appl. Physiol.
15:
1101-1106,
1960[Abstract/Free Full Text].
28.
Pollard, M. J.,
D. Megirian,
and
J. H. Sherrey.
Unity of costal and crural diaphragmatic activity in respiration.
Exp. Neurol.
90:
187-193,
1985[Medline].
29.
Sinderby, C.,
J. Beck,
L. Lindström,
and
A. Grassino.
Enhancement of signal quality in esophageal recordings of diaphragm EMG.
J. Appl. Physiol.
82:
1370-1377,
1997[Abstract/Free Full Text].
30.
Sinderby, C.,
L. Lindström,
and
A. Grassino.
Automatic assessment of electromyogram quality.
J. Appl. Physiol.
79:
1803-1815,
1995[Abstract/Free Full Text].
32.
Smith, J.,
and
F. Bellemare.
Effect of lung volume on in vivo contraction characteristics of human diaphragm.
J. Appl. Physiol.
62:
1893-1900,
1987[Abstract/Free Full Text].
33.
Vredenbregt, J.,
and
G. Rau.
Surface electromyography in relation to force, muscle length, and endurance.
In: New Developments in Electromyography and Clinical Neurophysiology, edited by J. E. Desmedt. Basel: Karger, 1973, vol. 1, p. 607-622.
34.
Ward, M.,
D. Eidelmen,
D. G. Stubbing,
F. Bellemare,
and
P. T. Macklem.
Respiratory sensation and pattern of respiratory muscle activation during diaphragm fatigue.
J. Appl. Physiol.
65:
2181-2189,
1988[Abstract/Free Full Text].
35.
Younes, M.,
and
W. Riddle.
A model for the relation between respiratory neural and mechanical outputs.
J. Appl. Physiol.
51:
963-978,
1981[Abstract/Free Full Text].
J APPL PHYSIOL 85(3):1123-1134
8570-7587/98 $5.00
Copyright © 1998 the American Physiological Society
Top
Abstract
Introduction
