- Email: [email protected]
An online technique for estimating cerebral carbon dioxide reactivity
An online technique for estimating carbon dioxide reactivity M.C. Patel, M.G. Taylor, S. Kontis, T.S. Padayachee Ultrasonic Angiology Research Group, London Bridge, London SE1 9RT, UK
Division
cerebral
and R.G. Gosling
of Radiological
Sciences,
Guys Hospital,
Received November 1988, accepted February 1990 ABSTRACT
A technique for measuring cerebral reactivity using transcranial pulsed Doppler ultrasound is described; the system includes a spectrum analyser and capnometer. Data acquisition and manipulation is under software control. Main stem middle cerebral artery blood velocity is monitored continuously using the transcranial Doppler technique, whilst the operator initiates data collection and controls the inspired gas composition. The calculation of cerebral CO2 reactivity is based upon linear regression analysis of normal&d, time-averaged middle cerebral velocity on end-tidal p CO2 and is displayed graphically. Measurement of middle cerebral CO, reactivities can be completed within 15min. Resultsporn two subjects, a healthy volunteer and a patient with occlusive disease, are shown to illustrate the technique. Keywords:
CO, reactivity, blood velocity, Doppler ultrasound, cerebral circulation
INTRODUCTION
SYSTEM DESCRIPTION
Cerebral CO2 reactivity is defined as the change in cerebral blood flow (CBF) per unit change in arterial COn tension &CO,). Increasing the level of arterial PCO~ causes a reduction in cerebral resistance and a consequent increase in cerebral blood flow. In normal sub’ects, the relationship between CBF and arterial pC d 2 is sigmoid and is linear in the range 3065 mmHg pC02. The maximum b which flow can increase is equivalent to the cerebra lyreserve capacity. Because the increase in CO2 will vary with baseline CBF, investigators have presented their data as percentage CO2 reactivity. This involves normalization of the CBF data to the CBF obtained at a pCOz of 40 mmHg’. Measurements of cerebral reactivity and cerebral reserve have proved useful in the management of patients with cerebral vasospasm and cerebrovascular disease*. However, such measurements have been limited as most CBF techniques are either invasive or require the use of radioisotopes. Transcranial pulsed Doppler ultrasound enables middle cerebral artery blood velocity to be measured non-invasive1 3. It has subsequent1 been shown that time-average dyvelocity in the midd re cerebral arte can be used as an index of CBF and can provide brly ood flow information for the measurement of cerebral reactivity415. As in subjects with no respiratory problems the endexpiratory pCO2 shows a linear relationship with arterial PCOs, these data may be combined with blood velocity information to measure cerebral reactivity non-invasively. In this paper a microcomputer-based system is described for measuring the cerebral COP reactivity non-invasively.
The system consists of a transcranial
ultrasound unit (TC2-64B, EME, Uerberlingen, FRG); which comprises a 2MHz pulsed Doppler velocimeter and a real-time spectral analyser, a capnometer (Gould Medical, Bilthoven, The Netherlands), an ECG trigger for timing purposes and a microcomputer (BBC model B, Acorn) with dual disc drive facility. The spectrum analyser of the transcranial Doppler s stem outputs maximum Doppler-shift frequency J ata in the form of a 6-bit word every 12ms. This digital output is converted into an analogue signal and then passed to the ‘analogue-in’ port of the microcomputer. The D/A conversion effectively smoothes the data and avoids any ‘handshake’ problems due to timing differences between the two systems. An electronically isolated ECG trigger provides a 25ms pulse to the microcomputer on detecting the RS complex in order to ensure sampling of the b B ood velocity data over complete cardiac cycles. The capnograph measures the instantaneous percentage of CO2 over the range O-10%. The value is displayed digitally and is also output in anal0 e form v ing from O-l V. This analogue sign aY is passed Y irectly to the ‘analogue-in’ port of the microcomputer. Each of the signals from the capnometer and spectrum analyser is alternately digitized under software control to la-bit accuracy. Each conversion takes 10 ms giving a minimum time resolution of 20ms. Software programs that run the system are stored on one floppy disc whilst the second disc is reserved for data storage.
System calibration Correspondence
and reprint requests to: Dr T.S. Padayachee
0 1990 Butterworth-Heinemann for BES 0141~.5425/90/004316-03 316
J. Biomed. Eng. 1990, Vol. 12, July
Software
was written to check
the accuracy
of the
An online technique for estimating cerebral carbon dioxide reactivity: MC. Pate1
analogue outputs of the capnograph and spectral analyser and the A/D converter in the microcomputer. This involved setting the capnograph to test mode, which produced voltage outputs proportional to 0% and 6% COz. These values were compared with the values displayed on the microcomputer screen and any deviation corrected by adjustment of the relevant A/D input attenuator housed in the connector to the computer analogue input port. The spectral analyser output was checked by selecting freeze mode, which enabled a constant and known value of velocity to be output, and this was then displayed on the computer screen and corrected in a similar fashion.
et al.
t Sample
keyboard
Yes Loop
counter
=0 I
i Sample A/D update max CO2
METHODS
I Plot
The subject lay supine whilst ECG
electrodes were applied and the ultrasound transducer was ositioned over the temporal bone to obtain Dopp Per si als from the middle cerebral artery. The tram Jnucer position was maintained constant throughout data acquisition b a monitoring head-band (EME, Uerberlin en, FR E ). A nose-cli was then positioned and the su%ject asked to breath t! rough a mouth piece which was connected to two one-way valves which ensured that inspired and expired gases did not mix. A 2 litre reservoir bag formed part of the circuit and accomodated ra id changes in respiratory rate. The sensing tube of tKe capnometer was connected to the expiratory outlet of this circuit. Initially the subject breathed air and once data collection was initiated a mixture of 10% CO2 in 02 was added to the inspired gas in a step-wise fashion. After each increment, there was an equilibration period during which an y chan%es.in CO2 level and middle cerebral artery b ood ve ocrty were allowed to stabilize before data acquisition was initiated. Gas flow rates were adjusted according to individual subject demand. The addition of the CO:! mixture caused an increase in the end-expiratory pCOz from its baseline value up to 70mmH . A decrease in endexpiratory pCOz was effected % y voluntary hyperventilation.
max
CO2 on
screen
t Sample A/D conversion from spectrum analyser
I Plot
max
F on
screen
Loop counter =loop counter + 1
No
No
Yes
SOFIWARE
Data acquisition program Figure 1 illustrates the principal components of this program. In acquisition mode the computer waits for an ECG trigger signal, once this is received both the velocity and pC!Oz data are digitized and displayed graphically every 25ms. The digitization period continues for at least 5 s and then terminates on the following ECG trigger. Thus during the acquisition period a complete number of cardiac cycles are digitized and an average of all the instantaneous maximum velocity recordings in this time is taken as representative of the mean middle cerebral artery blood velocity. The 5 s period was found to be sufficient to include an expiration cycle in all cases. The maximum value of pCOz recorded in the measurement period was taken to be the end-expiratory PC02 value. This data
Figure 1 reactivity
Flowchart
of software for online
measurement
of CO2
point is then plotted on the velocity/PC02 graph and the program recycles until instructed otherwise by a keystroke. In hold mode the data are processed as above except that data points appear but are not plotted on the screen. This enables the operator to ensure that signals are being collected and recorded faithfully before they are plotted on the screen. The acquisition mode can then be continued without any loss of acquired data.
J. Biomed. Eng. 1990, Vol. 12, July
317
An online technique fir estimating cerebral carbon dioxidereactivity: M.C. Patchet al.
where VOis the observed velocity, V&Jthe velocity at a pCOz of 40 mmHg and V, the normalized velocity expressed as a ercentage. Percentage reactivity is then calculated Kom the plot of normalized velocity against end-tidal pCOs. Linear regression analysis is performed and the correlation coefficient is calculated to provide an indication of the linearity of the relationship. The slope of this line is equal to the cerebral reactivi in units of percentage change in velocity per mm3; g of CO2 and is displayed on the computer screen.
100 90
80 70 60 50 40
G al m
-40 -50
/
20
I
I
I
I
I
30
40
50
60
70
a
CO2
I 80
I
1
90
100
The online nature of this system enables instant review of results. The following examples illustrate the information derived from such measurements. Figure 2a shows the CO* reactivity lots from a healthy normal volunteer who showe B a reactivity value within the expected normal limits of 2-6%. Figure 2b shows a plot for a atient with severe stenosis of the ipsilateral intern P carotid artery. The lowered reactivity value of 0.33% indicates that the cerebral vascular reserve has been diminished in an attempt to improve cerebral blood flow to the middle cerebral territory.
(mmHg1
100 90
80 70 2 EI s
60 50 40 30 20
L ”
0 0 -10 0 10 L-+----
2
5
APPLICATION
ACKNOWLEDGEMENTS 20
30
40
50
60
70
80
90
100
Figure 2 COP reactivity plots. a, Volunteer with a normal COP reactivityof 3.27% (regressioncoefficient0.99); b, patient with severe internal carotid stenosiswith a reduced value of 0.33% (0.72)
Data processing progmm
All data points obtained for each measurement are stored in RAM and saved on the data floppy disc. At the end of each measurement there is a plot of velocity against pCOz displayed on the screen for all data points. At this time it is possible to exclude data points that fall outside the range of end-expiratory PC02 30-65 mmHg (necessary for reactivity calculations) by using the ‘edit’ facility incorporated into the software. However, all other data points should be included in further analyses to avoid biasing of results. Subse uent processing is erformed on the displaye Zl data points to calcuPate the percentage cerebral reactivity. Initially the velocity data are normalized as shown below: v
=
N
318
WI- bo)x 100 V40
J. Biomed. Eng. 1990, Vol. 12, July
are grateful to the Coronary Artery Disease Research Association (CORDA) for financial support to TSP.
We
REFERENCES 1. Harper AM, Glass HJ. Effects of the alterationin the arterial CO2 tensionon the blood flowthroughthe cerebralcortex at normal and low arterial pressures.J Neurol Neurosutg Psychiatt 1965; 28: 449-52. 2. Padayachee TS, Kirkham FJ, Lewis RR, Gillard J, Hutchinson MCE, Gosling RG. Transcranial measurement of blood velocities in basal cerebral arteries using pulsed Doppler ultrasound: a method of assessing the Circle of Willis. UltrasoundMed Biol 1986; 1: 5-14. 3. Aaslid R, Markwalder TM, Nomes H. Noninvasive transcranial Doppler ultrasound recording of flow velocity in basal cerebral arteries. J Neurosurg 1982; 57: 769-74. 4. Kirkbam I$ Padayachee TS, Parsons S, Seargent LS, House FR, Gosling RG. Transcranial measurement of blood velocities in basal cerebral arteries using pulsed Doppler ultrasound: velocity as an index of flow. ultrasound Med Biol 1986; 12: 15-21. 5. Markwalder TM, Grolimund P, Seiler RW, Roth F, Aaslid R. Dependency of blood flow velocity in the middle cerebral artery on the end tidal CO2 partial pressure. A transcranial Doppler study. j CerebBZuodFlow Metub 1984; 4: 368-72.
Division
cerebral
and R.G. Gosling
of Radiological
Sciences,
Guys Hospital,
Received November 1988, accepted February 1990 ABSTRACT
A technique for measuring cerebral reactivity using transcranial pulsed Doppler ultrasound is described; the system includes a spectrum analyser and capnometer. Data acquisition and manipulation is under software control. Main stem middle cerebral artery blood velocity is monitored continuously using the transcranial Doppler technique, whilst the operator initiates data collection and controls the inspired gas composition. The calculation of cerebral CO2 reactivity is based upon linear regression analysis of normal&d, time-averaged middle cerebral velocity on end-tidal p CO2 and is displayed graphically. Measurement of middle cerebral CO, reactivities can be completed within 15min. Resultsporn two subjects, a healthy volunteer and a patient with occlusive disease, are shown to illustrate the technique. Keywords:
CO, reactivity, blood velocity, Doppler ultrasound, cerebral circulation
INTRODUCTION
SYSTEM DESCRIPTION
Cerebral CO2 reactivity is defined as the change in cerebral blood flow (CBF) per unit change in arterial COn tension &CO,). Increasing the level of arterial PCO~ causes a reduction in cerebral resistance and a consequent increase in cerebral blood flow. In normal sub’ects, the relationship between CBF and arterial pC d 2 is sigmoid and is linear in the range 3065 mmHg pC02. The maximum b which flow can increase is equivalent to the cerebra lyreserve capacity. Because the increase in CO2 will vary with baseline CBF, investigators have presented their data as percentage CO2 reactivity. This involves normalization of the CBF data to the CBF obtained at a pCOz of 40 mmHg’. Measurements of cerebral reactivity and cerebral reserve have proved useful in the management of patients with cerebral vasospasm and cerebrovascular disease*. However, such measurements have been limited as most CBF techniques are either invasive or require the use of radioisotopes. Transcranial pulsed Doppler ultrasound enables middle cerebral artery blood velocity to be measured non-invasive1 3. It has subsequent1 been shown that time-average dyvelocity in the midd re cerebral arte can be used as an index of CBF and can provide brly ood flow information for the measurement of cerebral reactivity415. As in subjects with no respiratory problems the endexpiratory pCO2 shows a linear relationship with arterial PCOs, these data may be combined with blood velocity information to measure cerebral reactivity non-invasively. In this paper a microcomputer-based system is described for measuring the cerebral COP reactivity non-invasively.
The system consists of a transcranial
ultrasound unit (TC2-64B, EME, Uerberlingen, FRG); which comprises a 2MHz pulsed Doppler velocimeter and a real-time spectral analyser, a capnometer (Gould Medical, Bilthoven, The Netherlands), an ECG trigger for timing purposes and a microcomputer (BBC model B, Acorn) with dual disc drive facility. The spectrum analyser of the transcranial Doppler s stem outputs maximum Doppler-shift frequency J ata in the form of a 6-bit word every 12ms. This digital output is converted into an analogue signal and then passed to the ‘analogue-in’ port of the microcomputer. The D/A conversion effectively smoothes the data and avoids any ‘handshake’ problems due to timing differences between the two systems. An electronically isolated ECG trigger provides a 25ms pulse to the microcomputer on detecting the RS complex in order to ensure sampling of the b B ood velocity data over complete cardiac cycles. The capnograph measures the instantaneous percentage of CO2 over the range O-10%. The value is displayed digitally and is also output in anal0 e form v ing from O-l V. This analogue sign aY is passed Y irectly to the ‘analogue-in’ port of the microcomputer. Each of the signals from the capnometer and spectrum analyser is alternately digitized under software control to la-bit accuracy. Each conversion takes 10 ms giving a minimum time resolution of 20ms. Software programs that run the system are stored on one floppy disc whilst the second disc is reserved for data storage.
System calibration Correspondence
and reprint requests to: Dr T.S. Padayachee
0 1990 Butterworth-Heinemann for BES 0141~.5425/90/004316-03 316
J. Biomed. Eng. 1990, Vol. 12, July
Software
was written to check
the accuracy
of the
An online technique for estimating cerebral carbon dioxide reactivity: MC. Pate1
analogue outputs of the capnograph and spectral analyser and the A/D converter in the microcomputer. This involved setting the capnograph to test mode, which produced voltage outputs proportional to 0% and 6% COz. These values were compared with the values displayed on the microcomputer screen and any deviation corrected by adjustment of the relevant A/D input attenuator housed in the connector to the computer analogue input port. The spectral analyser output was checked by selecting freeze mode, which enabled a constant and known value of velocity to be output, and this was then displayed on the computer screen and corrected in a similar fashion.
et al.
t Sample
keyboard
Yes Loop
counter
=0 I
i Sample A/D update max CO2
METHODS
I Plot
The subject lay supine whilst ECG
electrodes were applied and the ultrasound transducer was ositioned over the temporal bone to obtain Dopp Per si als from the middle cerebral artery. The tram Jnucer position was maintained constant throughout data acquisition b a monitoring head-band (EME, Uerberlin en, FR E ). A nose-cli was then positioned and the su%ject asked to breath t! rough a mouth piece which was connected to two one-way valves which ensured that inspired and expired gases did not mix. A 2 litre reservoir bag formed part of the circuit and accomodated ra id changes in respiratory rate. The sensing tube of tKe capnometer was connected to the expiratory outlet of this circuit. Initially the subject breathed air and once data collection was initiated a mixture of 10% CO2 in 02 was added to the inspired gas in a step-wise fashion. After each increment, there was an equilibration period during which an y chan%es.in CO2 level and middle cerebral artery b ood ve ocrty were allowed to stabilize before data acquisition was initiated. Gas flow rates were adjusted according to individual subject demand. The addition of the CO:! mixture caused an increase in the end-expiratory pCOz from its baseline value up to 70mmH . A decrease in endexpiratory pCOz was effected % y voluntary hyperventilation.
max
CO2 on
screen
t Sample A/D conversion from spectrum analyser
I Plot
max
F on
screen
Loop counter =loop counter + 1
No
No
Yes
SOFIWARE
Data acquisition program Figure 1 illustrates the principal components of this program. In acquisition mode the computer waits for an ECG trigger signal, once this is received both the velocity and pC!Oz data are digitized and displayed graphically every 25ms. The digitization period continues for at least 5 s and then terminates on the following ECG trigger. Thus during the acquisition period a complete number of cardiac cycles are digitized and an average of all the instantaneous maximum velocity recordings in this time is taken as representative of the mean middle cerebral artery blood velocity. The 5 s period was found to be sufficient to include an expiration cycle in all cases. The maximum value of pCOz recorded in the measurement period was taken to be the end-expiratory PC02 value. This data
Figure 1 reactivity
Flowchart
of software for online
measurement
of CO2
point is then plotted on the velocity/PC02 graph and the program recycles until instructed otherwise by a keystroke. In hold mode the data are processed as above except that data points appear but are not plotted on the screen. This enables the operator to ensure that signals are being collected and recorded faithfully before they are plotted on the screen. The acquisition mode can then be continued without any loss of acquired data.
J. Biomed. Eng. 1990, Vol. 12, July
317
An online technique fir estimating cerebral carbon dioxidereactivity: M.C. Patchet al.
where VOis the observed velocity, V&Jthe velocity at a pCOz of 40 mmHg and V, the normalized velocity expressed as a ercentage. Percentage reactivity is then calculated Kom the plot of normalized velocity against end-tidal pCOs. Linear regression analysis is performed and the correlation coefficient is calculated to provide an indication of the linearity of the relationship. The slope of this line is equal to the cerebral reactivi in units of percentage change in velocity per mm3; g of CO2 and is displayed on the computer screen.
100 90
80 70 60 50 40
G al m
-40 -50
/
20
I
I
I
I
I
30
40
50
60
70
a
CO2
I 80
I
1
90
100
The online nature of this system enables instant review of results. The following examples illustrate the information derived from such measurements. Figure 2a shows the CO* reactivity lots from a healthy normal volunteer who showe B a reactivity value within the expected normal limits of 2-6%. Figure 2b shows a plot for a atient with severe stenosis of the ipsilateral intern P carotid artery. The lowered reactivity value of 0.33% indicates that the cerebral vascular reserve has been diminished in an attempt to improve cerebral blood flow to the middle cerebral territory.
(mmHg1
100 90
80 70 2 EI s
60 50 40 30 20
L ”
0 0 -10 0 10 L-+----
2
5
APPLICATION
ACKNOWLEDGEMENTS 20
30
40
50
60
70
80
90
100
Figure 2 COP reactivity plots. a, Volunteer with a normal COP reactivityof 3.27% (regressioncoefficient0.99); b, patient with severe internal carotid stenosiswith a reduced value of 0.33% (0.72)
Data processing progmm
All data points obtained for each measurement are stored in RAM and saved on the data floppy disc. At the end of each measurement there is a plot of velocity against pCOz displayed on the screen for all data points. At this time it is possible to exclude data points that fall outside the range of end-expiratory PC02 30-65 mmHg (necessary for reactivity calculations) by using the ‘edit’ facility incorporated into the software. However, all other data points should be included in further analyses to avoid biasing of results. Subse uent processing is erformed on the displaye Zl data points to calcuPate the percentage cerebral reactivity. Initially the velocity data are normalized as shown below: v
=
N
318
WI- bo)x 100 V40
J. Biomed. Eng. 1990, Vol. 12, July
are grateful to the Coronary Artery Disease Research Association (CORDA) for financial support to TSP.
We
REFERENCES 1. Harper AM, Glass HJ. Effects of the alterationin the arterial CO2 tensionon the blood flowthroughthe cerebralcortex at normal and low arterial pressures.J Neurol Neurosutg Psychiatt 1965; 28: 449-52. 2. Padayachee TS, Kirkham FJ, Lewis RR, Gillard J, Hutchinson MCE, Gosling RG. Transcranial measurement of blood velocities in basal cerebral arteries using pulsed Doppler ultrasound: a method of assessing the Circle of Willis. UltrasoundMed Biol 1986; 1: 5-14. 3. Aaslid R, Markwalder TM, Nomes H. Noninvasive transcranial Doppler ultrasound recording of flow velocity in basal cerebral arteries. J Neurosurg 1982; 57: 769-74. 4. Kirkbam I$ Padayachee TS, Parsons S, Seargent LS, House FR, Gosling RG. Transcranial measurement of blood velocities in basal cerebral arteries using pulsed Doppler ultrasound: velocity as an index of flow. ultrasound Med Biol 1986; 12: 15-21. 5. Markwalder TM, Grolimund P, Seiler RW, Roth F, Aaslid R. Dependency of blood flow velocity in the middle cerebral artery on the end tidal CO2 partial pressure. A transcranial Doppler study. j CerebBZuodFlow Metub 1984; 4: 368-72.