- Email: [email protected]
Computerized hot-wire anemometry — Principles of calculation
Computer Programsin Biomedicine 11 (1980) 113-118 © Elsevier/North-Holland Biomedical Press
COMPUTERIZED HOT-WIRE ANEMOMETRY - PRINCIPLES OF CALCULATION Anders HALD and Bjarne STIGSBY Department of Anaesthesia and Department for Data Processingin Medicine, Herlev Hospital, University of Copenhagen, DK-2730 Herlev, Denmark Principles of calculation of respiratory parameters based on a hot-wire anemometer with special reference to computer monitoring were evaluated. Flow-rate, gas-pressure, and flow-direction signals were recorded simultaneously on magnetic tape. Subsequent quantitative analyses were performed on a general purpose digital minicomputer. An analysis epoch of 256 s was selected from the 3 channels. After identification of one cycle baseline values of flow-rate and pressure were determined.
Different time-lagsin one respiratory cycle (inspiratory time, pause time and expiratory time) could be determined. Inspiratory and expiratory volumes were obtained by integration. Peak pressure and plateau pressure were calculated, Airway resistance and total complianceof the thoracic cage and the lungs were calculated using the above mentioned parameters. Finally, the respiratory frequency was calculated. Anemometry, data processing Resistance, airway
Anemometry, hot-wire Respiratory monitoring
Compliance, of thoracic cage and lmigs
anemometer is obtained from an electrically heated wire. A heat transfer from the wire to the passing gas will take place. In order to keep the wire at a constant temperature adjustment of the current in the wire is necessary. These adjustments are the flow-rate signal. Figure 2, upper cur~,e, shows variations in flow-rate. Gas pressure is obtained from the lumen of the anemometer. The pressure is measured with an Acker pressure transducer type 840 based on the straingauge principle [12]. Figure 2, middle curve, shows variations in gas pressure. Directional indication, 2 V during expiration and 0 V during inspiration, is made by two heated sensors placed in parallel on the inside of the tube, perpendicular to the flow.direction. Figure 2, lower curve, shows directional indication. The hot-wire anemometer and a test procedure for its general useare detailed in [10]. The programs perform automatic digitizing and storage of 256 s signals from the above-mentioned 3 channels.
1. Introduction Studies on computerized respiratory monitoring are few [1-9]. The source of signals in this study is a new respiratory hot-wire constant temperature anemometer which together with simultaneous recording of airway pressure and directional indication of the passing gas can be used in connection with anaesthesia/intensive care monitoring and pulmonary fun:tion tests (fig. 1). Flow-rate in a hot-wire ccnstant temperature LUERLOCKFORPRESSURE FLOWSENSOR
/
TRANSDUCER
E
ISO-- CONNECTIONS TEMP.COMPENSATOR
Flow-rate analysis
2. t3reliminaw processing
N DIRECTIONAL SENSOR
Two respiratory cycles are necessary as calibration signals. Using a spirometer and a water manometer as
Fig. 1. The hot-wire constant temperature anemometer. 113
A. Hald, B. Stigsby, Computerizedhot.wire anemometry
114
d
"'
I FLOW-RATE (l/see)
3. Computational methods 3.1. The dam acquisitDn program
EXPIRATION
INSPIRATION
The computer digitizes the 3 analog channels from the magnetic tape with a sample frequency of 50 Hz. Three 3-sectioned ring buffers with a total capacity of 6.1 s are used. When one buffer section has been filled up it is stored in a magnetic disc file. The analog input routine continues by f'dling sample points in the next buffer section. The final result is 3 parallel disc files with sample points from each channel stored in a separate ['de.
/ PRESSURE (ram Hg)
3. 2. Cycle recognition program This program reads continuously sample points from file 3 (direction of flow). Every time one respiratory cycle has been read the start and end addresses are transferred t'o a cycle analysing program.
DIRECTION (volts) EXPIRATION ! INSPIRATION • PAUSF
/
3.3. Cycle analysing program
' L ,t:_
One cycle
This program performs the proper analysis of 1 cycle.
=~ TIME(see.)
Fig. 2. Showsthe 3 time series(flow-rate,pressure and flowdirection versustime) obtained during one respiratory cycle from a normal intubated and curarisedperson ventilated with a Siemens-ElemaServo 900 B ventilator.
3.3. L Baseline values A baseline for the flow-rate signals (//'flow) is calculated by ranking in increasing order all flow-rate samples within the cycle considered. The baseline value is the value of point no. 10, which is the median of the lowest 19 points. The same procedure is used to determine the baseline of the pressure signal
(Sp,o.). exact references the respiratory volumes (ml) and plateau-pressure (ram Hg) are measured. The plateau. pressure is the pressure between inspiration and expiration. These values are keyed in by the operator as Vkey and Pkey, respectively. The equivalents to Vkey are the areas below the flow-rate curves during either inspiration or expiration, in machine units A [¢al and A Ecah respectively. The equivalent to Pkey is the height of the plateau-pressure in machine units,/)ca] " Aicah AEcal, and Pea are computed as the arithmetic average of the two calibration cycles. The calibration signals as well as the following monitoring signals are recorded on magnetic tape.
3. 3. 2. Threshold In order to determine the start and end of inspiration and beginning of expiration a threshold is necessary. This is obtained by calculating the mean value (M) of the 19 highest points, still using the ranked flowrate sample points. Then the threshold (Tff) is given by: TH = Bflow + 1/3 (M - Bflow )
3.3.3. Duration o f cycle, inspiration, pause and expiration (see fig. 3) The start and end of 1 cycle ( T r T s ) is defined by
A. Hald, B. Stigsby, Computerized hot-wire anemometry
115
3.3.4. Respiratory volumes
I
The inspiratory and expiratory volume-,are calculated as the areas between the flow-rate signal and the baseline:
FLOW-RATE (llsecl t I~-THRESHOLO
FILE 1 ~
Area of inspiration T3
I
T2 F T2SS~RE T3~ ~T~
(flow.rate - Bflow ) dt
AI = f
r2 Area of expiration
FILE 2 I
Ts
~ AE-'- f
(flow-rate -Bflow ) dt
T4 The calculated values of the inspiratory and expiratory areas are converted to milliliters by comparing them with the calculated mean areas of the 2 calibration signals of known volume:
DIRECTION '~vo|ts)
lnspiratory volume
I
FILE 3 1 T !
Ts,l_
One cycle TtME(sec) Fig. 3. Is similar to fig. 1 but does also define the start and end of 1 respiratory cycle. 7"2 = start of inspiration, T 3 = end of inspiration and T4 = start of expiration, are defined as the moments when the flow-rate signal crosses the threshold.
the shift from high to low in fde 3 obtaining: Duration of I cycle
Expiratory volume
VI - Vk~y.A[ A lcal Vkey . AE
Finally, the difference in milliliters between the. 2 volumes and the difference as the percentage of inspiratory volume are calculated. These express the accuracy and tightness of the system. 3.3.5. Plateau and peak pressures
The plateau pressure is the average pressure during the pause time: 1 Tf3 T4 P~iateau = Dp
(pressure-- Bpress) ,it
Dc = Ts - TI
The duration of inspiration, pause and expiration are calculated from the cross-moments of the flowrate signal and the threshold obtaining:
The peak pressure is higher than the plateau pressure due to the dynamic flow resistance during inspiration (see fig. 4):
Duration of inspiration DI = T3 - T2
T3 Large areaAL = J (pressure - Bpress) dt T2
Duration of pause
Dp= T4 - T3
Duration of expiration
D E = O c - (7"4 - 7 2 )
The respiratory frequency defined as cycles/min is given by: F = 60" D~l
Small area As - ½ "Ppl~teau "DI Then the peak.pressure ~ peak is calculated as: P;eak - (AL - As)/DI + P;lateau
116
A. /laid, B. Stigsby, Computerized hot-wire anemomerry 3.3.6. Airway resistance and total compliance The airway resistance due to the inspiratory gasflow is calculated as the difference between peak and plateau pressures divided by the mean inspiratory flow-rate (square flow): h
t2
D! Airway resistance/Caw ='VI" (/)peak -/)plateau)
large area = T2, X,Y, 1"3 small area = T2,Z,T 3
Total compliance - Clara i (which is the compliance of the thoracic cage and the lungs) expresses the elasticity of the system, and is calculated as the volume change per unit pressure change:
Fig. 4. Shows the pressure curve during one respiratory cycle. Peak pressure = Ppeak, plateau pressure = Pplaleau and baseline = Bpress are indicated. The calculation of Ppeak is based on all sample points during inspiration in order to avoid artel'acls.
gl Clara I - _ _ Pplateau
The calculaled values of the plateau and peak pres-
sures are converted to millimeters of mercury by comparing them with the calculated mean plateau pressure of the 2 calibration signals of known pressure: Pplateau = Pkey. epla teau
4. Flowchart Figure 5 shows the s~stem flowchart of the data acquisition and the signal analysing principle. 5. Sample run
Pkey. Ppeak Ppeak = ',PeaI
Figure 6 shows the printer output fro'n the analysis of I cycle. The program analyses the cycles con-
DATA ACQUISITION
SIGNAL ANALYSIS CYCLE NO. X
FLOW
PRESSURE DIRECTION
CYCLE ANALYSING PROGRAM
DISC FILES
FILE 1 FLOW
SPOOLING PROGRAM
Fig. 5. System lion,chart.
FILE 2
PRESSURE
FILE 3
DIRECTION
START & END OF CYCLE
CYCLE RECOGNITll PROGRAM
A. Hald, B. Stigsby, Computerized hot.wire anemometry
117
equipped with DEC VT-52 terminals and a Centronics 301 printer.
C Y C L E NO. 6
Inspiration Duration Volume Mean flow rate
: 1.5 sec. : 744 ml = 29370 ml/min.
25%
: 3.9 sec. = 686 ml : 10670 mllmin.
66%
Ezpiration Duration Volume Mean flow rate
Volume difference = Pause time
=
58
ml
0,5 sec.
8% 9%
Pl'oaauro Area Peak pressure Plateau pressure i
Cycle t i m Frequency Compliance Resistance
= =
=
97 59 36
mm HO x eec. mm HO mm Hg
: = : =
5.9 10 21 47
sec. cycles/rain. mi/mm Hg mm Hg x sec./i
7. Mode o f availability
The source listings of the programs are available from the Department for Data Processing in Medicine, Herlev Hospital, DK-2730 Herlev, Denmark, on request.
Acknowledgements The authors wish to thank Dr. Mogens J~rgensen, Head of the Department for Data Processing in Medicine, Herlev Hospital, for his valuable cooperation, and DISA Elektronik, Skovlunde for the loan of the anemometer equipment.
•
100%
Fig. 6. Printer output after analysis of I cycle.
secutively. It terminates when 256 s o f signals have been analysed, or it may be stopped earlier b y the operator. The average time used for data acquisition and analysis is 3 times real time on the available equipment. This length o f time is dependent on the type o f printer used. An on.line version o f the program will be able to analyse the signals in real time and can be accomplished with a faster computer.
6. Hardwareand software specifications The programs were coded in FORTRAN IV and executed on a HP 21 MX° The acquisition and control system is the RTE version 2 at the Department for Data Processing in Medicine, Herlev Hospital. The computer is a 32 k, 16 bits/word computer. The disc storage capacity is 5 megabytes/disc (4 discs). It is
Referene~es [ 1] G. Fletcher and J.W. Bellville, Online computation of pulmonary compliance and work of breathing, L Appl. Physiol. 21 (1966) 1321-1327. [2] J.H. Miller and D.H. Simmons, Rapid determination of dynamic compliance and resistance, I. Appl. Physiol. 15 (1960) 967-974. [3] J.J. Osborn, LO. Beaumont, J.C.A. Raison, J. Russel and F. Gerbode, Measurement and monitoring of acutely ill patients by digital computer, Surgery 64 (1968) 1057-1070. [4] F. Gerbode, Computerized monitoring of seriously ill patients, J. Thorac. Cardiovasc. Surg. 66 (1973) 167174. [5] H.R. Warner, R.M. Gardner and A.F. Toronto, Computer-based monitoring of cardiovascular functions in post-operative patients, Circ. suppl. 2 0968) 37-38, 68-74. [6] M. Hilberman and R.M. Peters, On.line digital analysis of respiratory mechanics and the automation of respiratozy control, J. Thorac. Cardiovasc. Surg. 58 (1969) 821-828. [7] M. Hilberman, 6. Kamm, M. Tarter and J.J. Osborn, An evaluation of computer-based patient monitoring at Pacific Medical Center, Comic,at. Biomed. Res. 8 (1975) 447-460. [81 S.Z. Turney, W~Blumenfeld, S. Wolf, and R. Denman, Respiratory monitoring: recent developments in automatic monitoring of gas concentrations, flow, p~'essure and temperature. Ann. Thorac. Surg. 16 (1973) 184192.
118
A. Hald, B. Stigsby, Computerized hot-wire anemom¢:.~.
[9] S.Z. Turney and W. Blumenfeld, On-line respiratory wave form analysis using a digital desk calculator, Med. Biol. Eng. 2 (1973) 275-283. [ 10] T. Kann, A. Haid and F.E. J~rgensen, A new transducer
for respiratory monitoring, Acta Anaesth. Scand. 23 (1979) 394-358. [ I 1] D.W. Hill, Electronic techniques in anaesthesia and surgery (Butterworth, London, 1973).
COMPUTERIZED HOT-WIRE ANEMOMETRY - PRINCIPLES OF CALCULATION Anders HALD and Bjarne STIGSBY Department of Anaesthesia and Department for Data Processingin Medicine, Herlev Hospital, University of Copenhagen, DK-2730 Herlev, Denmark Principles of calculation of respiratory parameters based on a hot-wire anemometer with special reference to computer monitoring were evaluated. Flow-rate, gas-pressure, and flow-direction signals were recorded simultaneously on magnetic tape. Subsequent quantitative analyses were performed on a general purpose digital minicomputer. An analysis epoch of 256 s was selected from the 3 channels. After identification of one cycle baseline values of flow-rate and pressure were determined.
Different time-lagsin one respiratory cycle (inspiratory time, pause time and expiratory time) could be determined. Inspiratory and expiratory volumes were obtained by integration. Peak pressure and plateau pressure were calculated, Airway resistance and total complianceof the thoracic cage and the lungs were calculated using the above mentioned parameters. Finally, the respiratory frequency was calculated. Anemometry, data processing Resistance, airway
Anemometry, hot-wire Respiratory monitoring
Compliance, of thoracic cage and lmigs
anemometer is obtained from an electrically heated wire. A heat transfer from the wire to the passing gas will take place. In order to keep the wire at a constant temperature adjustment of the current in the wire is necessary. These adjustments are the flow-rate signal. Figure 2, upper cur~,e, shows variations in flow-rate. Gas pressure is obtained from the lumen of the anemometer. The pressure is measured with an Acker pressure transducer type 840 based on the straingauge principle [12]. Figure 2, middle curve, shows variations in gas pressure. Directional indication, 2 V during expiration and 0 V during inspiration, is made by two heated sensors placed in parallel on the inside of the tube, perpendicular to the flow.direction. Figure 2, lower curve, shows directional indication. The hot-wire anemometer and a test procedure for its general useare detailed in [10]. The programs perform automatic digitizing and storage of 256 s signals from the above-mentioned 3 channels.
1. Introduction Studies on computerized respiratory monitoring are few [1-9]. The source of signals in this study is a new respiratory hot-wire constant temperature anemometer which together with simultaneous recording of airway pressure and directional indication of the passing gas can be used in connection with anaesthesia/intensive care monitoring and pulmonary fun:tion tests (fig. 1). Flow-rate in a hot-wire ccnstant temperature LUERLOCKFORPRESSURE FLOWSENSOR
/
TRANSDUCER
E
ISO-- CONNECTIONS TEMP.COMPENSATOR
Flow-rate analysis
2. t3reliminaw processing
N DIRECTIONAL SENSOR
Two respiratory cycles are necessary as calibration signals. Using a spirometer and a water manometer as
Fig. 1. The hot-wire constant temperature anemometer. 113
A. Hald, B. Stigsby, Computerizedhot.wire anemometry
114
d
"'
I FLOW-RATE (l/see)
3. Computational methods 3.1. The dam acquisitDn program
EXPIRATION
INSPIRATION
The computer digitizes the 3 analog channels from the magnetic tape with a sample frequency of 50 Hz. Three 3-sectioned ring buffers with a total capacity of 6.1 s are used. When one buffer section has been filled up it is stored in a magnetic disc file. The analog input routine continues by f'dling sample points in the next buffer section. The final result is 3 parallel disc files with sample points from each channel stored in a separate ['de.
/ PRESSURE (ram Hg)
3. 2. Cycle recognition program This program reads continuously sample points from file 3 (direction of flow). Every time one respiratory cycle has been read the start and end addresses are transferred t'o a cycle analysing program.
DIRECTION (volts) EXPIRATION ! INSPIRATION • PAUSF
/
3.3. Cycle analysing program
' L ,t:_
One cycle
This program performs the proper analysis of 1 cycle.
=~ TIME(see.)
Fig. 2. Showsthe 3 time series(flow-rate,pressure and flowdirection versustime) obtained during one respiratory cycle from a normal intubated and curarisedperson ventilated with a Siemens-ElemaServo 900 B ventilator.
3.3. L Baseline values A baseline for the flow-rate signals (//'flow) is calculated by ranking in increasing order all flow-rate samples within the cycle considered. The baseline value is the value of point no. 10, which is the median of the lowest 19 points. The same procedure is used to determine the baseline of the pressure signal
(Sp,o.). exact references the respiratory volumes (ml) and plateau-pressure (ram Hg) are measured. The plateau. pressure is the pressure between inspiration and expiration. These values are keyed in by the operator as Vkey and Pkey, respectively. The equivalents to Vkey are the areas below the flow-rate curves during either inspiration or expiration, in machine units A [¢al and A Ecah respectively. The equivalent to Pkey is the height of the plateau-pressure in machine units,/)ca] " Aicah AEcal, and Pea are computed as the arithmetic average of the two calibration cycles. The calibration signals as well as the following monitoring signals are recorded on magnetic tape.
3. 3. 2. Threshold In order to determine the start and end of inspiration and beginning of expiration a threshold is necessary. This is obtained by calculating the mean value (M) of the 19 highest points, still using the ranked flowrate sample points. Then the threshold (Tff) is given by: TH = Bflow + 1/3 (M - Bflow )
3.3.3. Duration o f cycle, inspiration, pause and expiration (see fig. 3) The start and end of 1 cycle ( T r T s ) is defined by
A. Hald, B. Stigsby, Computerized hot-wire anemometry
115
3.3.4. Respiratory volumes
I
The inspiratory and expiratory volume-,are calculated as the areas between the flow-rate signal and the baseline:
FLOW-RATE (llsecl t I~-THRESHOLO
FILE 1 ~
Area of inspiration T3
I
T2 F T2SS~RE T3~ ~T~
(flow.rate - Bflow ) dt
AI = f
r2 Area of expiration
FILE 2 I
Ts
~ AE-'- f
(flow-rate -Bflow ) dt
T4 The calculated values of the inspiratory and expiratory areas are converted to milliliters by comparing them with the calculated mean areas of the 2 calibration signals of known volume:
DIRECTION '~vo|ts)
lnspiratory volume
I
FILE 3 1 T !
Ts,l_
One cycle TtME(sec) Fig. 3. Is similar to fig. 1 but does also define the start and end of 1 respiratory cycle. 7"2 = start of inspiration, T 3 = end of inspiration and T4 = start of expiration, are defined as the moments when the flow-rate signal crosses the threshold.
the shift from high to low in fde 3 obtaining: Duration of I cycle
Expiratory volume
VI - Vk~y.A[ A lcal Vkey . AE
Finally, the difference in milliliters between the. 2 volumes and the difference as the percentage of inspiratory volume are calculated. These express the accuracy and tightness of the system. 3.3.5. Plateau and peak pressures
The plateau pressure is the average pressure during the pause time: 1 Tf3 T4 P~iateau = Dp
(pressure-- Bpress) ,it
Dc = Ts - TI
The duration of inspiration, pause and expiration are calculated from the cross-moments of the flowrate signal and the threshold obtaining:
The peak pressure is higher than the plateau pressure due to the dynamic flow resistance during inspiration (see fig. 4):
Duration of inspiration DI = T3 - T2
T3 Large areaAL = J (pressure - Bpress) dt T2
Duration of pause
Dp= T4 - T3
Duration of expiration
D E = O c - (7"4 - 7 2 )
The respiratory frequency defined as cycles/min is given by: F = 60" D~l
Small area As - ½ "Ppl~teau "DI Then the peak.pressure ~ peak is calculated as: P;eak - (AL - As)/DI + P;lateau
116
A. /laid, B. Stigsby, Computerized hot-wire anemomerry 3.3.6. Airway resistance and total compliance The airway resistance due to the inspiratory gasflow is calculated as the difference between peak and plateau pressures divided by the mean inspiratory flow-rate (square flow): h
t2
D! Airway resistance/Caw ='VI" (/)peak -/)plateau)
large area = T2, X,Y, 1"3 small area = T2,Z,T 3
Total compliance - Clara i (which is the compliance of the thoracic cage and the lungs) expresses the elasticity of the system, and is calculated as the volume change per unit pressure change:
Fig. 4. Shows the pressure curve during one respiratory cycle. Peak pressure = Ppeak, plateau pressure = Pplaleau and baseline = Bpress are indicated. The calculation of Ppeak is based on all sample points during inspiration in order to avoid artel'acls.
gl Clara I - _ _ Pplateau
The calculaled values of the plateau and peak pres-
sures are converted to millimeters of mercury by comparing them with the calculated mean plateau pressure of the 2 calibration signals of known pressure: Pplateau = Pkey. epla teau
4. Flowchart Figure 5 shows the s~stem flowchart of the data acquisition and the signal analysing principle. 5. Sample run
Pkey. Ppeak Ppeak = ',PeaI
Figure 6 shows the printer output fro'n the analysis of I cycle. The program analyses the cycles con-
DATA ACQUISITION
SIGNAL ANALYSIS CYCLE NO. X
FLOW
PRESSURE DIRECTION
CYCLE ANALYSING PROGRAM
DISC FILES
FILE 1 FLOW
SPOOLING PROGRAM
Fig. 5. System lion,chart.
FILE 2
PRESSURE
FILE 3
DIRECTION
START & END OF CYCLE
CYCLE RECOGNITll PROGRAM
A. Hald, B. Stigsby, Computerized hot.wire anemometry
117
equipped with DEC VT-52 terminals and a Centronics 301 printer.
C Y C L E NO. 6
Inspiration Duration Volume Mean flow rate
: 1.5 sec. : 744 ml = 29370 ml/min.
25%
: 3.9 sec. = 686 ml : 10670 mllmin.
66%
Ezpiration Duration Volume Mean flow rate
Volume difference = Pause time
=
58
ml
0,5 sec.
8% 9%
Pl'oaauro Area Peak pressure Plateau pressure i
Cycle t i m Frequency Compliance Resistance
= =
=
97 59 36
mm HO x eec. mm HO mm Hg
: = : =
5.9 10 21 47
sec. cycles/rain. mi/mm Hg mm Hg x sec./i
7. Mode o f availability
The source listings of the programs are available from the Department for Data Processing in Medicine, Herlev Hospital, DK-2730 Herlev, Denmark, on request.
Acknowledgements The authors wish to thank Dr. Mogens J~rgensen, Head of the Department for Data Processing in Medicine, Herlev Hospital, for his valuable cooperation, and DISA Elektronik, Skovlunde for the loan of the anemometer equipment.
•
100%
Fig. 6. Printer output after analysis of I cycle.
secutively. It terminates when 256 s o f signals have been analysed, or it may be stopped earlier b y the operator. The average time used for data acquisition and analysis is 3 times real time on the available equipment. This length o f time is dependent on the type o f printer used. An on.line version o f the program will be able to analyse the signals in real time and can be accomplished with a faster computer.
6. Hardwareand software specifications The programs were coded in FORTRAN IV and executed on a HP 21 MX° The acquisition and control system is the RTE version 2 at the Department for Data Processing in Medicine, Herlev Hospital. The computer is a 32 k, 16 bits/word computer. The disc storage capacity is 5 megabytes/disc (4 discs). It is
Referene~es [ 1] G. Fletcher and J.W. Bellville, Online computation of pulmonary compliance and work of breathing, L Appl. Physiol. 21 (1966) 1321-1327. [2] J.H. Miller and D.H. Simmons, Rapid determination of dynamic compliance and resistance, I. Appl. Physiol. 15 (1960) 967-974. [3] J.J. Osborn, LO. Beaumont, J.C.A. Raison, J. Russel and F. Gerbode, Measurement and monitoring of acutely ill patients by digital computer, Surgery 64 (1968) 1057-1070. [4] F. Gerbode, Computerized monitoring of seriously ill patients, J. Thorac. Cardiovasc. Surg. 66 (1973) 167174. [5] H.R. Warner, R.M. Gardner and A.F. Toronto, Computer-based monitoring of cardiovascular functions in post-operative patients, Circ. suppl. 2 0968) 37-38, 68-74. [6] M. Hilberman and R.M. Peters, On.line digital analysis of respiratory mechanics and the automation of respiratozy control, J. Thorac. Cardiovasc. Surg. 58 (1969) 821-828. [7] M. Hilberman, 6. Kamm, M. Tarter and J.J. Osborn, An evaluation of computer-based patient monitoring at Pacific Medical Center, Comic,at. Biomed. Res. 8 (1975) 447-460. [81 S.Z. Turney, W~Blumenfeld, S. Wolf, and R. Denman, Respiratory monitoring: recent developments in automatic monitoring of gas concentrations, flow, p~'essure and temperature. Ann. Thorac. Surg. 16 (1973) 184192.
118
A. Hald, B. Stigsby, Computerized hot-wire anemom¢:.~.
[9] S.Z. Turney and W. Blumenfeld, On-line respiratory wave form analysis using a digital desk calculator, Med. Biol. Eng. 2 (1973) 275-283. [ 10] T. Kann, A. Haid and F.E. J~rgensen, A new transducer
for respiratory monitoring, Acta Anaesth. Scand. 23 (1979) 394-358. [ I 1] D.W. Hill, Electronic techniques in anaesthesia and surgery (Butterworth, London, 1973).