Optoelectronic Plethysmography in Intensive
ANDREA ALIVERTI, RAFFAELE DELLACÀ, PAOLO PELOSI, DAVIDE CHIUMELLO, ANTONIO PEDOTTI,
and LUCIANO GATTINONI
Dipartimento di Bioingegneria, Politecnico di Milano, Italy; Centro di Bioingegneria, Fondazione Don Gnocchi IRCCS e Politecnico di Milano, Italy;
and Istituto di Anestesia e Rianimazione, Università di Milano, Ospedale Maggiore IRCCS, Milano, Italy
We used optoelectronic plethysmography to study 11 normal sub- with multiple devices and lie in a supine position, which limits
jects during quiet and deep breathing, six sedated and paralyzed the visibility of markers on the anterior and lateral parts of the
patients with acute lung injury and acute respiratory distress syn- rib cage and abdomen. The applicability of optoelectronic
drome (ALI/ARDS) receiving continuous positive pressure ventila- plethysmography in this environment is therefore question-
tion (CPPV) (positive end-expiratory pressure [PEEP] 10 cm H2O, able. In this work, we set out to test the feasibility of this
tidal volume [VT] 300, 600, 900 ml), and seven ALI/ARDS pa- method in an ICU setting by comparing the volumes mea-
tients receiving pressure support ventilation (PSV) (PEEP 10 cm sured using optoelectronic plethysmography with measure-
H2O, pressure support 5, 10, 15, 25 cm H2O). The volumes mea- ments taken using spirometry and pneumotachography. We
sured using optoelectronic plethysmography were compared with
studied both normal subjects and patients with ALI/ARDS.
measurements taken using spirometry and pneumotachography.
Our aim is to report data regarding the feasibility of using op-
The three methods were highly correlated. The discrepancies found
toelectronic plethysmography on ICU patients and some pre-
were 1.7 5.9%, 1.6 5.4%, and 4.9 6.4% when comparing
optoelectronic plethysmography with spirometry, optoelectronic
liminary optoelectronic plethysmography results concerning
plethysmography with pneumotachography, and spirometry with the distribution of chest wall volume changes found between
pneumotachography, respectively. Accuracy of the compartmental- spontaneously breathing subjects, patients with ALI/ARDS
ization procedure (upper thorax, lower thorax, and abdomen) was receiving pressure support ventilation (PSV), and sedated and
assessed by calculating compartmental volume changes during iso- paralyzed patients with ALI/ARDS receiving continuous posi-
volume maneuvers. The discrepancy from the ideal zero line was tive pressure ventilation (CPPV).
2.1 48.3 ml. Abdominal contribution to inspired volume was
greater for normal subjects than for PSV patients (63 11% METHODS
versus 43 14%, p 0.001). It decreased with VT for normal
subjects (48.5 15%, p 0.05), whereas it increased for CPPV
patients (61 10%, p 0.05). No significant distribution differ- The study population consisted of 24 subjects. Eleven were healthy
ences were found between 5 and 25 cm H2O PSV. We conclude volunteers (5 men and 6 women, age 28.0 4.5 yr, weighing 65.3
that optoelectronic plethysmography is a feasible technique able
11.5 kg and 1.71 0.09 meters tall, body mass index [BMI] 22.18
2.08 kg/m2). Thirteen were patients who, either before or during the
to provide unique data on the distribution of chest wall volume
study, had ALI/ARDS, defined according to the American European
changes in intensive care patients.
Consensus Conference Criteria (7). The patients’ most relevant clini-
cal data are summarized in Table 1.
Chest wall volumes depend on the regional mechanical charac-
teristics of the lungs and the chest wall. Patients with acute lung Experimental Set-up (Figure 1)
injury (ALI) and acute respiratory distress syndrome (ARDS) Optoelectronic plethysmography. Optoelectronic plethysmography
typically have unevenly distributed lesions throughout the lung was carried on the whole study population by analyzing the move-
parenchyma and altered mechanical properties of the chest ments of retro-reflective markers using four television cameras con-
wall, especially after abdominal disease (1). Indeed, it may be nected to an automatic motion analyzer (Elite system; BTS, Milan, It-
interesting to assess the chest wall mechanics and distribution aly). All markers were simultaneously visible to at least two television
of chest wall volume changes of ALI/ARDS patients. cameras so that their three-dimensional positions and displacements
Recently, a plethysmographic method based on optoelec- could be reconstructed using stereo-photogrammetric methods (8).
tronic measurement was described which can be used to assess After a series of preliminary experiments, we found that the best set-
absolute chest wall volumes and their variations in the upper up for the supine position involved using 45 markers. The markers
were placed in order to define three chest wall compartments: upper
and lower rib cage and abdomen (both right and left sides) (2–
thorax, lower thorax, and abdomen, as shown in Figure 2 (see APPEN-
4). The method is based on measuring a finite number of dis- DIX for details).
placements of points on the outer surface of the chest wall. It Spirometry. The normal subjects were connected to a water dis-
is noninvasive and does not involve connections to the patient. placement spirometer fitted with a potentiometer (Model 308; Spec-
So far, it has only been applied to normal subjects in a seated trol Electronics, City of Industry, CA, linearity 0.25%). The spi-
position (5, 6). rometer was filled with room air at ambient temperature, and the
In intensive care units (ICUs), patients are usually covered subject–spirometer circuit was closed without CO2 absorption or oxy-
gen supply. Spirometry was carried out on normal subjects for 30 to 40 s.
(Received in original form March 2, 1999 and in revised form November 11, 1999) For four CPPV and four PSV patients, who were ventilated using a Si-
This work was partly supported by the European Commission – BIOMED II emens Servo 900 C ventilator (Siemens-Elema, Solna, Sweden), the
programme (biomedical technology research project “BREATH” – Biomedical spirometer was connected to the ventilator exhaust port. Expiratory
technology for REspiration Analysis THrough optoelectronics) and by MURST - gases were collected over a 60-s period. The analogue spirometer sig-
Cofin ’98. nal (calibrated beforehand using a 1.5-L calibration syringe) was sent
Correspondence and requests for reprints should be addressed to Andrea Aliverti, to an A/D board (RTI800; Analog Devices, Norwood, MA), synchro-
Ph.D., Centro di Bioingegneria, Politecnico di Milano-Fond.Pro Juventute Don nized with the automatic motion analyzer and digitally recorded at
Gnocchi, Via Gozzadini 7, I-20148 Milano, Italy. E-mail: email@example.com 100 Hz.
Am J Respir Crit Care Med Vol 161. pp 1546–1552, 2000 Pneumotachography. For four CPPV patients and four PSV pa-
Internet address: www.atsjournals.org tients, a pneumotachograph (HR 4700-A; Hans Rudolph Inc., Kansas
Aliverti, Dellacà, Pelosi, et al.: Chest Wall Optoelectronic Plethysmography 1547
ALI/ARDS Study Day
Patients Sex Age Weight BMI PaO2/FIO2 PaO2/FIO2 Day of Study Diagnosis S/D
1 F 70 75 29.3 152 136 1 Pneumonia S
2 F 73 70 27.3 190 230 2 Pneumonia S
3 F 71 70 27.3 141 204 22 Pneumonia D
4 M 81 70 24.2 218 335 1 Postanoxic coma S
5 F 40 48 21.3 292 383 3 Pneumonia S
6 F 64 75 33.3 144 228 20 Pneumonia S
Mean SD 66.5 14.1 68 10.1 27.1 4.1 189.5 58.6 252.7 90.4 8.2 10.0
1 M 60 60 19.6 79 190 15 Sepsis D
2 M 76 48 19.5 168 376 21 Pneumonia S
3 M 55 70 20.5 188 302 76 Sepsis S
4 M 55 80 24.7 107 234 51 Politrauma S
5 M 67 75 24.5 225 245 28 Hemorrhagic shock S
6 F 60 75 29.3 123 225 38 Septic shock S
7 M 65 80 26.1 146 308 7 Cardiac arrest S
Mean SD 62.6 7.5 69.7 11.8 23.4 3.7 148.0 50.0 268.6 63.3 33.7 23.7
Definition of abbreviations: BMI body mass index, kg/m2; CPPV continuous positive pressure ventilation; PSV pressure support ventilation; S survived; D deceased.
City, MO) was inserted between the filter and the connection to the expiratory line (pressing expiratory hold on the Servo 900 C) at the end
ventilator. Its output signal was sent to an A/D board, synchronized of expiration. The patient’s inspiratory effort caused either a contrac-
with the automatic motion analyzer and digitally recorded at 100 Hz. tion of the belly and an expansion of the thoracic cage (“belly in”) or an
Time integration of this signal provided the volume measurement. expansion of the belly and a contraction of the rib cage (“belly out”).
Esophageal pressure. For the seven patients receiving PSV, esoph- Simultaneously, the esophageal pressure swings were recorded.
ageal pressure (Pes) was measured by using an esophageal balloon
(Bicore, Irvine, CA) modified to allow connection to the transducer Data Analysis
(SCX01; Sensym, Milpitas, CA). During measurements, the balloon Volume comparison. A detailed description of the optoelectronic
was inflated with 0.5 to 1 ml of air. The balloon positioning was veri- method used to analyze chest volume in standing and seated positions
fied by chest radiography. has already been published (2–4). The changes made in order to ana-
lyze subjects in a supine position are summarized in the APPENDIX.
Experimental Procedures Three-dimensional reconstruction was based on stereo-photogram-
After an adaptation period, the 11 normal subjects were asked to metry (8) using the data from two or more television cameras to im-
breathe normally and quietly for about 30 s, then to take four deep prove calculation accuracy. On the basis of previously defined geo-
breaths. The six sedated paralyzed patients receiving CPPV were ven- metrical models that describe the whole chest wall and its compartments,
tilated at the fraction of inspired oxygen (FIO2) in use (0.44 0.11), dedicated software was used to calculate the absolute volumes and
(positive end-expiratory pressure [PEEP] 10 cm H2O) and studied at their changes during respiration from changes in the x, y, and z coor-
VT 300, 600, and 900 ml. Their respiratory frequency was kept con- dinates of the markers.
stant, and the data collected in each test were averaged over 60 s. The The expiratory changes in volume ( Volume volume at end of
seven patients receiving PSV were ventilated at the FIO2 in use (0.38 inspiration volume at end of expiration) measured using optoelec-
0.08), (PEEP 10 cm H2O) and studied at 5, 10, 15, and 25 cm H2O PSV. tronic plethysmography, spirometry, and pneumotachography were
To assess the accuracy of optoelectronic plethysmography in mea- averaged for each test (i.e., over approximately 8 to 20 breaths for
suring volume distribution between the three chest wall compartments each test) and the three methods were compared by linear regression
(upper thorax, lower thorax, and abdomen), the seven patients receiv- and Bland-Altman analysis (9). In addition, the mean percentage dis-
ing PSV were made to make isovolume movements by occluding the crepancy between any two methods was calculated as:
Figure 1. Experimental set-up for optoelectronic plethys-
mography, spirometry, and pneumotachography. Opto-
electronic plethysmography and spirometry were com-
pared both for normal subjects and ALI/ARDS patients,
whereas optoelectronic plethysmography and pneumo-
tachography were compared only for ALI/ARDS patients.
1548 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 161 2000
and pneumotachography. Note the excellent correlation of the
several comparisons, the small intercept of the regression line,
and the tight data around that line. There is an increase in the
difference between volumes measured at the mouth and at the
chest wall. This difference increases absolutely (while it de-
creases fractionally) within increasing VT (upper two panels),
but no such relationship for the comparison of spirometric and
The average discrepancy between pneumotachography and
optoelectronic plethysmography was 1.6 5.4%, which was
similar to the discrepancy between spirometry and optoelec-
tronic plethysmography (1.7 5.9%, p not significant [NS]).
The largest discrepancy found was between spirometry and
pneumotachography (4.9 6.4%, p 0.05). The discrepan-
cies between pneumotachography, optoelectronic plethysmog-
raphy, and spirometry were not correlated with the BMI.
Figure 2. Marker positions and chest wall compartments for optical The patients in PSV reacted to the “occlusion” maneuver ei-
analysis in supine position. Transversal lines: clavicular line; manubrio-
sternal joint (angle of Louis); xiphoid process; lower costal margin; up-
ther with a “belly-in” (four patients) or a “belly-out” (three
per abdomen (3 markers); umbilical level; anterior superior iliac crest. patients) movement. The results are shown in Figure 5. As can
Axial lines: two midaxillary lines; two midclavicular lines; two paraster- be seen, the discrepancy from the ideal zero line was on aver-
nal lines (extended to the abdominal region); midsternal line (ex- age 8.3 52.5 ml (i.e., 12.7% of the thoracic volume change)
tended to the abdomen). Chest wall compartments: upper thorax (UT) for patients who performed belly-in movements. The same
or lung-apposed pulmonary rib cage; lower thorax (LT) or diaphragm- discrepancy was on average 10.6 56.7 ml (i.e., 8.9% of the
apposed abdominal rib cage; abdomen (AB).
thoracic volume change) for patients who performed belly-out
movements (not significant).
discrepancy% = [ ( ∆V X – ∆V Y ) ⁄ ∆V X ] × 100 We found a significative difference (p 0.05) in averaged
esophageal pressure swings during occlusion between belly in and
where VX is the volume change measured using spirometry or pneu- belly out patients ( 7.8 5.0 cm H2O and 13.8 7.4 cm H2O,
motachography and VY is the volume change measured using opto- respectively), suggesting that belly out is associated with a greater
electronic plethysmography. When comparing pneumotachography
inspiratory effort. Moreover, we found a correlation between the
with spirometry, VY is the pneumotachographic volume change and
VX, the spirometric volume change. amplitude of the esophageal pressure swings and the deviation
Accuracy. To test the accuracy of the compartmentalization proce- from the ideal zero line, i.e., more negative was the esophageal
dure, we compared the volume changes in each chest wall compart- pressure swings more positive was the difference between the ob-
ment (upper, lower thorax and abdomen) measured during occlusion served changes in chest wall volume and the ideal zero line (slope
tests on patients receiving PSV. The compartmental volume changes 3.5 ml/cm H2O, intercept 40.8 ml, r 0.63, p 0.01).
measured in each test were averaged. Ideally, the sum of the volume
changes in the three compartments (upper thorax, lower thorax, and Compartmental Volume Changes
abdomen) should equal zero.
We observed significant differences in distribution of chest
Distribution of Chest Wall Volume Changes wall volume changes between normal subjects and mechani-
cally ventilated patients and, in the latter case, between pa-
Optoelectronic plethysmography was used on the whole study popu-
lation, both normal subjects and CPPV and PSV patients, to study tients receiving CPPV and patients receiving PSV. Moreover,
volume changes in the upper thorax, lower thorax, and abdomen. Dis- distribution of chest wall volume changes appeared to change
tribution of chest wall volume changes calculated for all subjects in all according to VT. The relevant data are summarized in Figure 6.
ventilatory conditions was expressed as percentage of VT. As can be seen, abdominal contribution to total chest wall
volume changes was 63.1 11.4%, 56.4 9.5%, and 43.3
Statistical Analysis 14% for normal subjects during quiet breathing (VT 631.3
All data are expressed as mean SD unless otherwise indicated. The 148.4 ml), CPPV patients at the lowest VT used (VT 300 ml),
three methods (optoelectronic plethysmography, spirometry, and and PSV patients at the lowest pressure support used (5 cm
pneumotachography) were compared by linear regression and Bland- H2O, VT 341.8 81.1), respectively (normal subjects versus
Altman analysis (9). Spirometry, pneumotachography, and optoelec- CPPV, p NS; normal subjects versus PSV, p 0.001; and
tronic plethysmography accuracy data were compared using Student’s
CPPV versus PSV, p 0.05). Increasing VT for normal sub-
t test, as were the distribution of chest wall volume changes of normal
and mechanically ventilated patients. Probability values of less than
jects (by asking them to breathe deeply, VT 2,695.1 655.0
0.05 were taken as significant. ml), CPPV patients (VT 900 ml), and PSV patients (at 25 cm
H2O, VT 770.2 287.2) led to different abdominal contribu-
RESULTS tions: 48.5 15%, 61.1 10%, and 47.9 10%, respectively
(normal subjects versus CPPV p 0.05; normal subjects ver-
Figure 3 shows typical experimental tracings for normal sub- sus PSV, p NS; and CPPV versus PSV, p 0.05). Indeed,
jects during spontaneous breathing (left panel) and patients when VT was increased, abdominal contribution to total chest
receiving PSV (right panel). wall volume changes decreased significantly for normal sub-
jects (p 0.05), but increased significantly for CPPV patients
Comparisons between Spirometry, Pneumotachography, (p 0.05). No significant differences in abdominal contribu-
and Optoelectronic Plethysmography tion to total chest wall volume changes were found between 5
Figure 4 shows the relationships among chest wall volume and 25 cm H2O PSV. It is interesting to note, however, that
changes as measured by optical plethysmography, spirometry, there was greater upper thorax (and lower abdomen) contri-
Aliverti, Dellacà, Pelosi, et al.: Chest Wall Optoelectronic Plethysmography 1549
Figure 3. (Left panel) a typical experimental tracing obtained for
normal subjects. Vut upper thoracic volume, Vlt lower tho-
racic volume; Vab abdomen volume; Vcw volume of total
chest wall (Vcw Vut Vlt Vab ); Vsp spirometer volume.
(Right panel) a typical experimental tracing obtained for patients
with ARDS receiving PSV. Pao pressure at the airway opening.
The arrow indicates the beginning of isovolume movements ob-
tained by occlusion at the end of expiration. As can be seen, flow
0, volume of the spirometer constant, and Vcw constant.
This particular patient reacted to the occlusion with inspiratory efforts (see Pao deflection) because of the expansion of the abdomen and retraction
of the upper thorax. For simplicity, no experimental tracing is given for ALI/ARDS patients receiving CPPV.
bution to total chest wall volume changes for patients receiv- volume changes and chest wall volume changes. Interestingly,
ing PSV than there was for patients receiving CPPV at all VT gas volume changes were found to be systematically signifi-
levels. cantly higher than chest wall volume changes when VT was in-
creased (Figure 4, upper and middle panels). This may suggest
a progressive increase in blood shift from the thoracic/abdom-
inal complex to the periphery when intrathoracic pressure is
The primary aim of this study was to test the feasibility of us- increased.
ing optoelectronic plethysmography on ICU patients, who are Another potential source of discrepancy stems from the
usually covered with multiple devices which could potentially differences between actual lung gas volume changes and the
interfere with measurements. We found that optoelectronic ple- volume changes measured using spirometry which are caused
thysmography is as feasible for these patients as it is for nor- by temperature, humidity, and pressure differences between
mal subjects. the lungs and the spirometer. In our experimental conditions,
spirometer gas humidity and pressure corrections, in order to
Comparison between Spirometry, Pneumotachography, bring them in line with the standard body conditions of nor-
and Optical Plethysmography mal subjects (i.e., 37 C and 100% humidity), imply a 6 to
To assess the effectiveness of optoelectronic plethysmography 7% gas expansion, according to standard formulas (11). On
in measuring volumes, we used spirometry as a reference the other hand, during mechanical ventilation, the positive
method throughout the whole study population, and in the pressure applied to the lungs induces a gas compression of ap-
case of patients with ALI/ARDS we used both spirometry and proximately 1 to 2%, depending on the pressure used. Fur-
pneumotachography. It is important to point out, however, thermore, spirometry was carried out through closed-circuit
that these comparisons are not straightforward. Optoelec- rebreathing (without CO2 absorption) on normal subjects, and
tronic plethysmography, in fact, measures changes in the chest through expired gas collection on mechanically ventilated pa-
wall, whereas spirometry and pneumotachography measure tients. During rebreathing, oxygen consumption exceeds CO2
changes in lung gas volume. These changes are not necessarily output, which results in an overall decrease in lung–spirome-
the same because there may be blood shifts from the thorax/ ter system gas content (12). This does not occur for mechani-
abdomen to the periphery or vice versa. cally ventilated patients, because the ratio between oxygen
In fact, as has been described for positive pressure ventila- consumption and CO2 output remains unchanged during the
tion (10), blood shifts to the periphery may lead to greater gas test. Faced with all these confounding factors, we chose not to
lung volume changes than chest wall volume changes, and vice introduce any corrections, because most of them are mutually
versa during negative pressure ventilation. Indeed, there is a compensating (i.e., temperature and humidity increase gas vol-
potential physiological source of discrepancy between lung gas ume, whereas pressure and O2 consumption above CO2 output
1550 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 161 2000
Figure 4. Triangles refer to normal subjects during spon-
taneous rebreathing; closed circles refer to patients re-
ceiving CPPV; open circles refer to patients receiving
PSV. (Upper panels) Left: regression line between
volume changes in the spirometer ( Vsp) and volume
changes measured using optoelectronic plethysmog-
raphy ( Vcw) (intercept 28.1 ml, slope 0.93, r
0.99, p 0.0001). Right: Bland-Altman analysis on the
same patient data (mean 13.3 ml). Note the signifi-
cant regression between increasing VT (x-axis) and
the difference between the two methods (y-axis) (inter-
cept 22.5 ml, slope 0.06, r 0.37, p 0.05).
(Middle panels) Left side: regression line between vol-
ume changes measured by the pneumotachograph
( Vpn) and volume changes measured using opto-
electronic plethysmography ( Vcw). This comparison
only includes patients receiving CPPV or PSV, because
pneumotachography was not used on normal subjects
(intercept 46.5 ml, slope 0.92, r 0.99, p
0.0001). Right: Bland-Altman analysis on the same pa-
tient data (mean 1.9 ml). Note the significant re-
gression between increasing VT (x-axis) and the differ-
ence between the two methods (y-axis) (intercept
42.6 ml, slope 0.07, r 0.52, p 0.01). (Lower
panels) Left: regression line between volume changes
in the spirometer ( Vsp) and volume changes measured
by the pneumotachograph ( Vpn) (intercept
27.2 ml, slope 1.00, r 0.99, p 0.0001). Right:
Bland-Altman analysis on the same patient data (mean
26.4 ml). Note the lack of significant regression be-
tween increasing VT (x-axis) and the difference be-
tween Vsp and Vpn (y-axis), which both measure lung
gas volume changes.
decrease gas volume). In fact, we have previously shown that crepancy from the ideal zero line was 2.1 48.3 ml. This, on
these corrections are not necessary in a similar situation where the average, is a satisfactory result; however, we have to explain
volume–pressure curve is performed with a super-syringe (13). the rather large standard deviation we observed. This could be
Indeed, even without any correction, the changes in lung the result of a technical problem in the method or of physio-
gas volume and the changes in chest volume were highly cor- logical phenomena. Interestingly, we found that the amplitude
related with a discrepancy of approximately 2%. Interestingly, of the negative esophageal pressure swings was associated
the highest and most significant discrepancy was found when with a chest wall volume change more positive than expected.
comparing two accepted methods: spirometry and pneumot- This may suggest a blood shift from periphery into the chest
achography (4.9 6.4%). and/or greater lung gas expansion.
With all these limitations in mind, among which possible
blood shift is the most important physiologically, our data sug-
gest that optical plethysmography is an excellent estimate of Distribution of Chest Wall Volume Changes
lung gas volume changes even in the case of ICU patients. We observed significant differences in distribution of chest
wall volume changes between normal subjects and mechani-
Compartmentalization Procedure Accuracy cally ventilated patients and, in the latter case, between pa-
To assess the accuracy of the compartmentalization procedure tients receiving CPPV and patients receiving PSV. For sponta-
in patients receiving PSV, we used the occlusion test which re- neously breathing subjects, distribution of chest wall volume
sulted in belly-in or belly-out movement. Ideally (absence of changes is dictated by the mechanical characteristics of the
blood shifts and negligible lung gas compression/decompres- respiratory system and the relative activity of the diaphragm
sion) changes in thoracic volume should be opposite and equal and inspiratory rib cage muscles, whereas for paralyzed CPPV
to changes in abdominal volume, and their sum should be patients, only the mechanical characteristics of the system are
zero. Considering the overall data, the magnitude of the dis- involved. The case of PSV is more complicated, because spon-
Aliverti, Dellacà, Pelosi, et al.: Chest Wall Optoelectronic Plethysmography 1551
Figure 5. Volume changes recorded for the seven PSV patients during
occlusion maneuvers. Each series of bars refers to one patient. Note
that Patients 1, 5, 6, and 7 reacted to occlusion by expanding the tho-
rax (belly-in) whereas Patients 2, 3, and 4 expanded the abdomen
(belly-out). Top: volume changes measured in the thorax ( Vth which
includes upper thorax and lower thorax). Middle: volume changes
measured in the abdomen ( Vab). Bottom: volume changes of the
chest wall ( Vcw Vth Vab). Ideally Vcw should be 0.
taneous and mechanical ventilation are combined to various Figure 6. Distribution of chest wall volume changes in upper thorax
and abdomen (percent of VT) for normal subjectS (triangles), CPPV pa-
degrees depending on the amount of respiratory support. tients (closed circles), and PSV patients (open circles) as a function of VT.
Chest wall volume changes for normal subjects during quiet For clarity, the lower thorax is not shown but it may easily be calcu-
breathing (631.3 148.4 ml) were preferentially distributed to lated as 100 (Vut% Vab%) and does not change significantly in
the abdominal compartment. This suggests greater compli- any of the groups considered.
ance of this compartment and/or a preferential activation of
the diaphragm. Taking a normal functional residual capacity
(FRC) of 2,200 ml (14) for our normal subjects, “quiet breath- CPPV and spontaneously breathing normal subjects. It may
ing” implies lung expansion of approximately 25 to 30%. be supposed that this preferential contribution of the upper
Shifting to deep breaths (VT 2,695.1 655.0 ml), i.e., ex- thorax was mainly due to the activity of inspiratory rib cage
panding the lung by more than its FRC, led to redistribution muscles. Increasing the support pressure to 10, 15, and 25 cm
of chest wall displacement from the abdomen to the thorax. H2O (and increasing the VT) led to an increase in abdominal
Assuming that abdomen/thorax compliances remain the same, and a decrease in upper thorax contribution to total chest wall
this suggests greater activation of rib cage inspiratory muscles. volume changes. This may be the result of lower inspiratory
The final result is a more uniform distribution of inflation be- rib cage muscle activity and/or an increase in abdominal com-
tween the thorax and the abdomen at high lung volumes, as pliance through mechanisms similar to those reported for
has been previously described for normal subjects (15). CPPV patients. However, even at 25 cm H2O PSV (VT
We measured the FRC of four of the seven patients receiv- 770.2 287.2 ml), a level at which ventilation should mainly
ing CPPV using helium dilution (16): it was found on average be passive, the distribution was significantly different from
to be 1,364 448 ml. In fact, CPPV patients ventilated with that found for CPPV patients. This finding deserves further in-
VT 300 ml expanded their respiratory systems by approxi- vestigation.
mately 25%, the same order of magnitude as normal subjects In conclusion, we believe that optoelectronic plethysmog-
during quiet breathing; their abdominal contribution to total raphy is an excellent tool for measuring chest wall volume
chest wall volume changes was also found to be similar. How- changes in ICUs and that it may be used to assess the physio-
ever, unlike in the case of normal subjects, expanding their logical importance of blood shifts and to study distribution of
respiratory systems to nearly 100% of their FRC (VT 900 ml) chest wall volume changes in different ventilatory conditions.
led to increased contribution of the abdominal compartment.
These data suggest that increasing the VT and pressures used References
to ventilate patients with ALI/ARDS at 10 cm H2O PEEP 1. Gattinoni, L., P. Pelosi, P. M. Suter, A. Pedoto, P. Vercesi, and A. Lis-
may cause increased compliance of the abdominal compart- soni. 1998. Acute respiratory distress syndrome caused by pulmonary
ment, possibly due to intratidal recruitment (17). and extrapulmonary disease: different syndromes? Am. J. Respir. Crit.
Care Med. 158:3–11.
The most interesting results, however, were seen in the case
2. Cala, S. J., C. M. Kenyon, G. Ferrigno, P. Carnevali, A. Aliverti, A. Pe-
of PSV patients. At 5 cm H2O PSV (VT 341.8 81.1 ml), dotti, P. T. Macklem, and D. F. Rochester. 1996. Chest wall and lung
chest wall volume changes were more highly distributed to the volume estimation by optical reflectance motion analysis. J. Appl.
upper thorax than for both ALI/ARDS patients receiving Physiol. 81:2680–2689.
1552 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 161 2000
3. Ferrigno, G., P. Carnevali, A. Aliverti, F. Molteni, G. Beulke, and A. Pe- Geometrical Modeling of the Chest Wall
dotti. 1994. Three-dimensional optical analysis of chest wall motion. J.
Appl. Physiol. 77:1224–1231.
The connections between the points define the geometrical mod-
4. Carnevali, P., G. Ferrigno, A. Aliverti, and A. Pedotti. 1996. A new els that describe the whole surface or a part thereof. Closed
method for 3D optical analysis of chest wall motion. Technol. Health surfaces are usually described by also considering some “vir-
Care 4:43–65. tual” markers whose coordinates are defined from the coordi-
5. Aliverti, A., S. J. Cala, R. Duranti, G. Ferrigno, C. M. Kenyon, A. Pe- nates of the real markers through mathematical relationships
dotti, G. Scano, P. Sliwinski, P. T. Macklem, and S. Yan. 1997. Human (mean value, translation, or a combination of both).
respiratory muscle actions and control during exercise. J. Appl. Phys-
In the case of standing and seated positions, i.e., when the
6. Kenyon, C. M., S. J. Cala, S. Yan, A. Aliverti, G. Scano, R. Duranti, A. whole trunk is visible “all round,” we have found (2) that the
Pedotti, and P. T. Macklem. 1997. Rib cage mechanics during quiet best marker arrangement consists of 86 markers on the chest
breathing and exercise in humans. J. Appl. Physiol. 83:1242–1255. wall (42 anterior, 34 posterior, and 10 lateral) and virtual mark-
7. Bernard, G. R., A. Artigas, R. L. Brigham, J. Carlet, K. Falke, L. Hud- ers are only used to close the surface at the top and bottom. In
son, M. Lamy, J. R. Le Gall, A. Morris, R. Spragg, and the Consensus the supine position, part of the chest wall surface is hidden by a
Committee. 1994. The American-European Consensus Conference on
ARDS: definitions, mechanisms, relevant outcomes and clinical trial
support, such as a bed. Therefore in the case of this constrained
coordination. Am. J. Respir. Crit. Care Med. 149:818–824. posture, only the visible markers can be considered in the chest
8. Ferrigno, G., N. A. Borghese, and A. Pedotti. 1990. Pattern recognition wall geometrical model and the small part of the body which
in 3-D motion analysis. ISPRS J. Photogram. Remote Sensing 45:227–246. cannot be seen by the cameras is assumed to be fixed.
9. Bland, J. M., and D. G. Altman. 1986. Statistical methods for assessing To obtain the maximal possible level of accuracy in volume
agreement between two methods of clinical measurements. Lancet 1: estimation, we use the same marker arrangement on the ante-
10. Hedenstierna, G., A. Strandberg, B. Brismar, H. Lundquist, L. Svensson,
rior chest as already defined for the upright positions. The
and L. Tockics. 1985. Functional residual capacity, thoraco-abdominal geometrical chest wall model is then built by considering a
dimension and central blood volume, during general anesthesia with number of virtual points belonging to a reference plane that
muscle paralysis and mechanical ventilation. Anesthesiology 62:247–254. corresponds approximately to the horizontal surface of the
11. Cramer, D., A. Peacock, and D. Denison. 1984. Temperature corrections bed. The height of this reference plane was estimated to be the
in routine spirometry. Thorax 39:771–774. mean height of the lateral markers minus 5 cm. The positions
12. Sackner, M. A. 1987. Measurement of cardiac output by alveolar gas ex-
change. In A. P. Fishman, L. E. Farhi, S. M. Tenney, and S. R. Geiger,
of these virtual points were determined by projecting the fron-
editors. Handbook of Physiology: The Respiratory System, Vol. IV, tal coordinates of the lateral markers which occurred in the
Section 3. American Physiological Society, Bethesda, MD. 233–255. first acquired image frame onto the reference plane.
13. Gattinoni, L., D. Mascheroni, E. Basilico, G. Foti, A. Pesenti, and L. Avalli.
1987. Volume/pressure curve of the total respiratory system in paralysed
patients: artefacts and correction factors. Intensive Care Med. 13:19–25. Volume Calculation
14. Ibanez, J., and J. M. Raurich. 1982. Normal values of functional residual
capacity in the seated and supine positions. Intensive Care Med. 8:
The geometrical definition of the model allows the enclosed
173–177. volume and its changes due to movement to be calculated by
15. Milic-Emili, J., A. M. Henderson, M. B. Dolovich, D. Trop, and K. surface integration using Gauss’s theorem to obtain a volume
Kaneko. 1966. Regional distribution of inspired gas in the lung. J. integral. In detail, the analytical expression of the theorem is:
Appl. Physiol. 21:749–759.
∫ F ⋅ ndS = ∫ ∇FdV
16. Pelosi, P., M. Cereda, G. Foti, M. Giacomini, and A. Pesenti. 1995. Al-
terations in lung and chest wall mechanics in acute lung injury. Am. J.
Respir. Crit. Care Med. 152:531–537. S V
17. Gattinoni, L., P. Pelosi, S. Crotti, and F. Valenza. 1995. Effects of posi-
where S is the surface considered; V is the volume enclosed by S;
tive end-expiratory pressure on regional distribution of tidal volume →
and recruitment in adult respiratory distress syndrome. Am. J. Respir. F is an arbitrary vector; n is the normal unitary vector at the
Crit. Care Med. 151:1807–1814. various points on S; and ∇ is the divergence operator.
If we choose an arbitrary vector with unitary divergence,
APPENDIX Equation A.1 becomes:
Optoelectronic Plethysmography: Measurement Principles
Optoelectronic plethysmography is based on automatic mo- ∫ F ⋅ ndS = ∫ dV = V (A.2)
tion analysis using passive 6-mm-diameter markers. Each S V
marker consists of a thin film of retro-reflective paper on a and the volume integral is calculated by means of a simpler
plastic hemisphere. The markers are fixed to the skin by surface integral. Passing from a continuous to discrete form,
means of biadhesive hypoallergenic tape. Specially designed Equation A.2 becomes:
TV cameras (solid-state charge-coupled devices) and infrared
scene lighting provide the best possible contrast between the K
marker images and the surrounding environment.
A dedicated processor recognizes and calculates the posi-
∑F ⋅ n A
i i = V (A.3)
tions of the markers recorded by the different cameras in real
time. System calibration, consisting of camera calibration and where K is the total number of triangles; Ai is the area of the i-th
spatial camera location, allows three-dimensional (3D) recon- triangle; and ni is the normal unitary vector of the i-th triangle.
struction of the different markers. Special algorithms used for By considering the geometric model of the whole trunk
marker detection provide the very high accuracy required for surface, it is therefore possible to calculate total chest wall vol-
measuring the micromovements involved in respiration (8). ume changes, whereas by considering only part thereof, the
Once the 3D coordinates of the points belonging to the contributions of the different compartments to total volume
chest wall surface have been acquired, the next step is to cal- changes can also be calculated.
culate the volume of the closed surface obtained by connect-
ing the points to form triangles.