Ultrasound in Med. & BioL. Vol. 2. pp. 31-36. Pergamon Press, 1975. Printed in Great Britain.
ERRORS A N D ARTIFACTS E N C O U N T E R E D IN THE M O N I T O R I N G OF FETAL RESPIRATORY MOVEMENTS USING ULTRASOUND D. J. FARMAN, G. THOMASand R. J. BLACKWELL Departments of Obstetrics and Gynaecologyand MedicalPhysics,UniversityCollegeHospital. London W.C.1. (First received 21 October 1974; and in final form 5 February 1975)
A ~ t r a c t - - A study is described of the possible sources of error and artifact arising in the monitoring of fetal respiratory movements using ultrasound, and their effect upon the clinical usefulness of the technique. Major artifacts arose from attempts to use the instrument controls to compensate for a poor choice of insonation direction, and from the use of a belt-mounted transducer. Critical requirements of the necessary instrumentation are discussed, together with the limitations of commercially-available equipment. Key words: Fetal monitoring, Respiratory movements, Ultrasound, Artifact.
MHz unfocussed field, was mounted on an abdominal belt incorporating a universal joint so that the direction of insonation could be adjusted without repositioning the belt on the maternal abdomen. As an alternative, a rigid transducer mount was available, coupling between the transducer and the abdomen being implemented by means of a water-bath. The two ultrasonic instruments used were essentially similar in respect of the transmitter, receiver RF amplifier, demodulator and video amplifier, but the custom-built instrument provided more comprehensive and more precise swept-gain facilities than did the commercial instrument. The instruments differed in the method by which target displacement was converted to an analogue voltage output. The circuitry of both instruments enabled a time gate to be displayed as a baseline step superimposed upon an A-mode display. In the commercial instrument the pulse repetition frequency was 1000 Hz, which was considered at this Department to be rather high for prolonged examination periods. The echo movement was derived in relation to the end of the time gate period and visualised by initiating a linear voltage ramp with the first significant echo received during the gate period. The voltage level attained by the ramp at the end of the gate period was stored and made available as the write-out voltage to be recorded on a paper chart. The control adjusting the length of the gate period also adjusted the slope of the
INTRODUCTION
Intrauterine fetal respiratory movements have been postulated and reported several times during the course of the last ninety years (Dawes et al., 1972) and attempts have recently been made to investigate this phenomenon in the human fetus usingultrasonic A-mode techniques (Boddy and Robinson, 1971). It has been suggested that the amount of episodic respiratory movement increases with gestation, and that the absence of such movements or the occurrence of isolated gasping movements may indicate fetal distress (Boddy and Dawes, 1975). To determine the clinical usefulness of monitoring fetal respiratory movements using the ultrasonic technique an independent study of 150 patients was undertaken, using both the Smith Kline Ekoline 20 (the commercial instrument used by other workers) and an instrument designed specifically for the study. Considerable disparity was noted between the recordings obtained using the two instruments: the emphasis-of the study was therefore realigned toward the investigation of possible sources of artifact. INSTRUMENTATION
The basis of the technique was to monitor the time interval between the generation of an ultrasonic pulse and the reception of the echo due to the fetal chest wall, variations in this interval representing displacements of the chest wall. The ultrasonic transducer, producing a 2.5 31
32
D.J. FARMAN.G. THOMASand R. J. BLACKWELL
time-to-voltage conversion ramp, so that the amplitude of the recorded output depended upon the gate period. These design features had two significant consequences. Firstly, if the amplitude of the echoes received within the gate period fell momentarily below the threshold level, the output voltage, which was maintained by the store, did not fall to zero before the next measurement cycle, but decayed slowly, so that an artifactual movement was recorded. Secondly; small amplitude signals misinterpreted as extraneous noise could almost be eliminated from the trace simply by lengthening the gate period. The altered output sensitivity results in a clean trace. The custom-built instrument had a pulse repetition frequency of 50 Hz in order to minimise the ultrasonic dose to the fetus. The time-to-voltage convertor was again based on a linear ramp, but this ramp was initiated by the start of the time gate and terminated by the first echo to exceed a variable threshold level. The resulting voltage was stored for one measurement cycle only, the output being reset to zero if all echoes received during the gate period were of subthreshold amplitude. The recorded amplitude of echo movement was calibrated and independent of the gate duration. Additional features included the provision of a Bessel-characteristic low-pass filter with cut-off frequency 1.67 Hz incorporated into the output circuitry to reject movement signals originating from the fetal heart, and a subtraction circuit allowing the maternal respiratory component to be largely removed from the record. The maternal respiration was monitored using a strain-gauge strapped around the upper abdomen. METHOD
The fetal chest wall was located by one of two alternative methods. The first was that used by other workers. The abdomen was palpated and the approximate orientation of the fetus established. The transducer was then placed on the anterior wall of the maternal abdomen and a search made for echoes characteristic of fetal heart. The transducer was secured in this position and the gate period adjusted to encompass the echo complex immediately proximal or distal to the fetal heart echoes, the complex being selected by its relative clarity. The second method was to use ultrasonic section-scanning techniques to determine the orientation of the fetus, the gate delay required to obtain a chest wall echo which was free from interference by limbs or placenta, and the
necessary orientation of the transducer for the chest wall to be normal to the ultrasonic beam. The transducer was then immediately secured in this position and the time gate set to the echo group indicated. DISCUSSION
Artifactual traces arise largely from endeavouring to use the instrument controls to compensate for a poor choice of transducer orientation. Consider Fig. 1(a). This is a section scan of the fetal thorax showing the direction of insonation chosen using the A-mode display alone, and also the optimum orientation selected from the section scan. Figs. 1(b and c) show samples of recordings of chest wall movements obtained using these two insonation directions. Shown in (c) is low-amplitude chest-wall movement with signal loss caused by movement of the fetal limb shown in the section scan, whilst in (b) there are well-defined, rhythmic chest wall movements. In many examinations the transducer location was obtained first by the A-mode technique and subsequently by section scanning. It was found that using the first technique the target monitored was not the chest wallamniotic fluid interface, but rather chest wallplacenta, chest wall-limb or occasionally a totally incorrect interface such as fetal limbamniotic fluid, the limb-chest wall interface having been missed in the general echo complex where the limb was opposed to the thorax. Even when a clear thorax-amniotic fluid boundary had been located, the direction of insonation was often found to be at other than normal incidence. Under these conditions the component of specular reflection was small, and the RF gain needed to be increased. This amplified to a significant level the small echoes arising throughout the thickness of the chest wall, and resulted in a broad echo complex being displayed. The effect was more noticeable with the commercial instrument, since a degree of video integration was incorporated into the custom-built instrument. Apparent correction could be achieved using the swept-gain controls, and this was found to be a major source of artifact, particularly with the limited swept-gain facilities available on the commercial instrument. This situation is illustrated in Fig. 2. With no swept-gain applied the echo complex is very broad, Fig. 2(a). With a steep swept-gain profile the echo complex has been considerably reduced in size, Fig. 2(b), because the proximal part of the complex has been heavily suppressed. Instead of using swept-gain in the conventional
Errors and artifacts encountered in the monitoring of fetal respiratory movements using ultrasound X
Y
(b)
T 5 m m/
[
i
.i 10
sec
(c)
Fig. 1. Importance of direction of insonation (a) Transverse section scan through fetal thorax, showing optimum (YI and incorrect (X) directions of insonation: (b) Recording of proximal chest wall movement in direction Y: (c) Recording of distal chest wall movement in direction X.
Fig. 2. Incorrect use of swept-gain to "clean up" chest wall echo complex. (al Minimal swept-gain ; (b) Steep swept-gain profile applied to chest wall echo complex.
33
D. J. FARMAN. G. THOMASand R. J. BLACKWELL
34
i
o
10 $ e c
(a)
I
S m m
10 $ e c
Cb)
Fig. 3. Artifactual recordings due to incorrect use of swept-gain. Upper traces obtained with incorrect swept-gain settings, lower traces with correct settings (a) Fetal chest wall movements; (b) Artifacts due to maternal respiratory movements.
manner to correct for attenuation over the area of interest, it has been used to obtain a highly uneven range of video amplitudes. The result is that as individual echoes in the complex move towards the front of the gate they are suppressed below the trigger threshold of the time-to-voltage analogue circuitry and triggering then results from the next significant echo within the gate, until this also is suppressed to a subthreshold level. The output will therefore be artifactual and appear as a small ripple situated at the position of the suppression edge of the sweptgain profile. A large amplitude artifactual output occurs with the commercial instrument when there is no significant succeeding echo, relating to the fetal chest wall, within the gate. In this case the time-to-yoltage convertor may be triggered by an echo from another structure, or may not be triggered at all. The output has the appearance of an almost rectangular wave whose amplitude bears no relation to the amplitude of the true chest wall movement, although the two frequencies may be identical. This effect is enhanced by the provision of storage circuits which, in the event of this momentary loss of the target echo. gives rise to an output which may be mistakenly interpreted as zero target move-
ment. Similarly, periods of signal loss due to slight fetal movement may be incorrectly interpreted as periods of zero respiratory movement. This situation does not arise with the custombuilt instrument since storage circuits are not incorporated. The swept-gain profile may also cause distortion in the amplitude of the time-to-voltage output. The amplitude of the video output varies with the position of the echo relative to the profile. The resulting change in the amplitude of the echo causes the trigger point on the echo to be dependent on this position. The amplitude of chest wall movement due to respiratory efforts is, by analogy with the adult, of the order of 4 r a m and a trigger point error of l mm (which is quite likely) is very significant. It is clear that swept-gain should never be used to "clean up" an echo complex. The effects described above were demonstrated by pulsing the transducer with the transmitter of the commercial instrument and using simultaneously the receiver and processing circuits of both instruments, the custom-built instrument being used correctly and the commercial instrument used by an operator lacking an appreciation of the possible artifacts but otherwise experienced.
Errors and artifacts encountered in the monitoring of fetal respiratory movements using ultrasound
5mml
,
,
35
(a)
10 sec
(b)
Fig. 4. Artifactual recordings due to abdominal wall movements. (a) Fetal chest wall movements at 38 weeks gestation; (b) "'Movements" of uterine wall, 3 days postpartum.
Sample traces are shown in Fig. 3. Figure 3(a) shows genuine fetal chest wall movements recorded by the two instruments. The output of the commercial instrument is seen to consist of a straight baseline with distinct, often rectangular deflections representing chest wall movements. This does not reflect the true physiological situation, since a straight baseline represents zero movement of the fetal target with respect to the ultrasonic transducer, whereas the output of the custom-built instrument shows some degree of movement to occur continuously. Figure 3 (b) shows artifacts recorded by the commercial instrument due to maternal respiration. This is confirmed by the simultaneous recording obtained with the correctly adjusted custom-built instrument. It was often found impossible to place the transducer in an ideal position since, with the occurrence of either maternal or fetal respiratory movement, there may be a slight change in angle of insonation of the fetal chest wall. These changes are accompanied by a reduction in echo amplitude, which can make monitoring difficult. This effect is diminished by using a wide beamwidth and a longer integration time
constant in the video processor. A large effective beamwidth is achieved with a large aperture transducer used with minimum video suppression and accurate swept-gain compensation. These measures, however, introduce further problems since a wide echo complex may be generated by echoes received from points on the target spaced at different distances from the transducer. The varying echo reflection times involved contribute to a broad echo group. The leading echo in the group may be monitored satisfactorily, but the point on the target giving rise to the leading echo may change as the complex moves. The use of the belt-mounted transducer was also found to lead to the production of recording artifacts. It was observed that the recordings did not show well-defined periods of respiratory movement as reported by other workers, but rather almost continuous movements of varying amplitude. In order to investigate the origin of the smaller-amplitude movements, two patients were examined post-natally using echoes from the anterior uterine wall. Figure 4 shows recordings obtained from one of these patients during the 38th week of pregnancy and again,
(a)
''mI '
lO sec
(b)
Fig. 5. Artifactual recordings due to transducer movement. (a) "'Movements'" of wall of non-pregnant, nulliparous uterus: (b) "'Movements" of hepatic structures.
36
D. J. FARMAN. G. THOMASand R. J. BLACKWELL
three days after delivery. It can be seen that small-amplitude movements, similar in rate to fetal respiratory movements, were recorded post-natally. Some of the uterine wall "movements" are synchronous with the maternal pulse but others resemble respiratory movements. Figure 5 shows recordings from the anterior wall of a non-pregnant, nulliparous uterus, and from echo-producing structures within a human liver, both recordings showing small, rhythmic movements occurring at a rate similar to fetal respiratory movements. When these examinations were repeated using the rigidly mounted transducer and water-bath these movements were no longer recorded, suggesting that they may be attributed to transducer displacement secondary to movement of the abdominal wall. The use of the water-bath was found inconvenient in the clinical situation. For recordings intended for detailed analysis its use would be considered mandatory. The small-amplitude movement artifacts described above could be masked by reducing the output sensitivity of the commercial instrument by increasing the gate period. This procedure, however, also masked genuine, low-amplitude fetal respiratory movements so that only largeamplitude movements were recorded. This also constitutes a ~ource of recording error. CONCLUSIONS
Fetal respiratory movements can be monitored by ultrasonic A-mode techniques, but a number of difficulties are encountered. For accurate results the equipment should be specifically adapted and must be used with great care. The output circuitry must be triggered solely and consistently by echoes from the fetal chest wall, which should be insonated at normal incidence.
Although satisfactory recordings may be obtained by an arbitrary direction of insonation, the use of an ultrasonic section scanner to locate the fetal chest wall greatly increases the likelihood of obtaining artifact-free recordings. The incorrect use of swept-gain facilities may destroy the fidelity of the recordings obtained, and steps must be taken to avoid artifacts due to maternal respiration, maternal and fetal pulse, and maternal abdominal movements. Even when these requirements have been met subsequent fetal movement may make it necessary to reposition the transducer several times during the course of a single examination, thus preventing continuous monitoring over long periods of time, especially during labour, when the uterine contractions contribute to the difficulties. It is concluded that A-mode techniques for recording fetal chest wall movements require further sophistication before they can be said to provide reliable information from which the clinical usefulness of the technique may be assessed. Acknowledgements--The authors gratefully acknowledge
the assistance of Professor D. V. 1. Fairweather and the Consultant Staff of the Obstetric Unit, University College Hospital, in permitting their patients to partake in this study, and of Mr. J. S. Clifton. Director. Medical Physics Department, University College Hospital and Medical School, for making available the facilities of his Department. The authors would also like to acknowledge the generous support of the National Fund for Research into Crippling Diseases in providing finance for the equipment used in this study.
REFERENCES Boddy. K. and Dawes, G. S. (1975) Fetal breathing. Br. Med. Bull. 31, 3. Boddy. K. and Robinson, J. S. (1971) External method for detection of fetal breathing in utero. Lancet 2, 1231. Dawes, G. S.. Fox. H. E.. Leduc, B. M., Liggins, G. C. and Richards, R. T. (1972) Respiratory movements and rapid eye movement sleep in the fetal lamb. J. Physiol. 220, 119.
ERRORS A N D ARTIFACTS E N C O U N T E R E D IN THE M O N I T O R I N G OF FETAL RESPIRATORY MOVEMENTS USING ULTRASOUND D. J. FARMAN, G. THOMASand R. J. BLACKWELL Departments of Obstetrics and Gynaecologyand MedicalPhysics,UniversityCollegeHospital. London W.C.1. (First received 21 October 1974; and in final form 5 February 1975)
A ~ t r a c t - - A study is described of the possible sources of error and artifact arising in the monitoring of fetal respiratory movements using ultrasound, and their effect upon the clinical usefulness of the technique. Major artifacts arose from attempts to use the instrument controls to compensate for a poor choice of insonation direction, and from the use of a belt-mounted transducer. Critical requirements of the necessary instrumentation are discussed, together with the limitations of commercially-available equipment. Key words: Fetal monitoring, Respiratory movements, Ultrasound, Artifact.
MHz unfocussed field, was mounted on an abdominal belt incorporating a universal joint so that the direction of insonation could be adjusted without repositioning the belt on the maternal abdomen. As an alternative, a rigid transducer mount was available, coupling between the transducer and the abdomen being implemented by means of a water-bath. The two ultrasonic instruments used were essentially similar in respect of the transmitter, receiver RF amplifier, demodulator and video amplifier, but the custom-built instrument provided more comprehensive and more precise swept-gain facilities than did the commercial instrument. The instruments differed in the method by which target displacement was converted to an analogue voltage output. The circuitry of both instruments enabled a time gate to be displayed as a baseline step superimposed upon an A-mode display. In the commercial instrument the pulse repetition frequency was 1000 Hz, which was considered at this Department to be rather high for prolonged examination periods. The echo movement was derived in relation to the end of the time gate period and visualised by initiating a linear voltage ramp with the first significant echo received during the gate period. The voltage level attained by the ramp at the end of the gate period was stored and made available as the write-out voltage to be recorded on a paper chart. The control adjusting the length of the gate period also adjusted the slope of the
INTRODUCTION
Intrauterine fetal respiratory movements have been postulated and reported several times during the course of the last ninety years (Dawes et al., 1972) and attempts have recently been made to investigate this phenomenon in the human fetus usingultrasonic A-mode techniques (Boddy and Robinson, 1971). It has been suggested that the amount of episodic respiratory movement increases with gestation, and that the absence of such movements or the occurrence of isolated gasping movements may indicate fetal distress (Boddy and Dawes, 1975). To determine the clinical usefulness of monitoring fetal respiratory movements using the ultrasonic technique an independent study of 150 patients was undertaken, using both the Smith Kline Ekoline 20 (the commercial instrument used by other workers) and an instrument designed specifically for the study. Considerable disparity was noted between the recordings obtained using the two instruments: the emphasis-of the study was therefore realigned toward the investigation of possible sources of artifact. INSTRUMENTATION
The basis of the technique was to monitor the time interval between the generation of an ultrasonic pulse and the reception of the echo due to the fetal chest wall, variations in this interval representing displacements of the chest wall. The ultrasonic transducer, producing a 2.5 31
32
D.J. FARMAN.G. THOMASand R. J. BLACKWELL
time-to-voltage conversion ramp, so that the amplitude of the recorded output depended upon the gate period. These design features had two significant consequences. Firstly, if the amplitude of the echoes received within the gate period fell momentarily below the threshold level, the output voltage, which was maintained by the store, did not fall to zero before the next measurement cycle, but decayed slowly, so that an artifactual movement was recorded. Secondly; small amplitude signals misinterpreted as extraneous noise could almost be eliminated from the trace simply by lengthening the gate period. The altered output sensitivity results in a clean trace. The custom-built instrument had a pulse repetition frequency of 50 Hz in order to minimise the ultrasonic dose to the fetus. The time-to-voltage convertor was again based on a linear ramp, but this ramp was initiated by the start of the time gate and terminated by the first echo to exceed a variable threshold level. The resulting voltage was stored for one measurement cycle only, the output being reset to zero if all echoes received during the gate period were of subthreshold amplitude. The recorded amplitude of echo movement was calibrated and independent of the gate duration. Additional features included the provision of a Bessel-characteristic low-pass filter with cut-off frequency 1.67 Hz incorporated into the output circuitry to reject movement signals originating from the fetal heart, and a subtraction circuit allowing the maternal respiratory component to be largely removed from the record. The maternal respiration was monitored using a strain-gauge strapped around the upper abdomen. METHOD
The fetal chest wall was located by one of two alternative methods. The first was that used by other workers. The abdomen was palpated and the approximate orientation of the fetus established. The transducer was then placed on the anterior wall of the maternal abdomen and a search made for echoes characteristic of fetal heart. The transducer was secured in this position and the gate period adjusted to encompass the echo complex immediately proximal or distal to the fetal heart echoes, the complex being selected by its relative clarity. The second method was to use ultrasonic section-scanning techniques to determine the orientation of the fetus, the gate delay required to obtain a chest wall echo which was free from interference by limbs or placenta, and the
necessary orientation of the transducer for the chest wall to be normal to the ultrasonic beam. The transducer was then immediately secured in this position and the time gate set to the echo group indicated. DISCUSSION
Artifactual traces arise largely from endeavouring to use the instrument controls to compensate for a poor choice of transducer orientation. Consider Fig. 1(a). This is a section scan of the fetal thorax showing the direction of insonation chosen using the A-mode display alone, and also the optimum orientation selected from the section scan. Figs. 1(b and c) show samples of recordings of chest wall movements obtained using these two insonation directions. Shown in (c) is low-amplitude chest-wall movement with signal loss caused by movement of the fetal limb shown in the section scan, whilst in (b) there are well-defined, rhythmic chest wall movements. In many examinations the transducer location was obtained first by the A-mode technique and subsequently by section scanning. It was found that using the first technique the target monitored was not the chest wallamniotic fluid interface, but rather chest wallplacenta, chest wall-limb or occasionally a totally incorrect interface such as fetal limbamniotic fluid, the limb-chest wall interface having been missed in the general echo complex where the limb was opposed to the thorax. Even when a clear thorax-amniotic fluid boundary had been located, the direction of insonation was often found to be at other than normal incidence. Under these conditions the component of specular reflection was small, and the RF gain needed to be increased. This amplified to a significant level the small echoes arising throughout the thickness of the chest wall, and resulted in a broad echo complex being displayed. The effect was more noticeable with the commercial instrument, since a degree of video integration was incorporated into the custom-built instrument. Apparent correction could be achieved using the swept-gain controls, and this was found to be a major source of artifact, particularly with the limited swept-gain facilities available on the commercial instrument. This situation is illustrated in Fig. 2. With no swept-gain applied the echo complex is very broad, Fig. 2(a). With a steep swept-gain profile the echo complex has been considerably reduced in size, Fig. 2(b), because the proximal part of the complex has been heavily suppressed. Instead of using swept-gain in the conventional
Errors and artifacts encountered in the monitoring of fetal respiratory movements using ultrasound X
Y
(b)
T 5 m m/
[
i
.i 10
sec
(c)
Fig. 1. Importance of direction of insonation (a) Transverse section scan through fetal thorax, showing optimum (YI and incorrect (X) directions of insonation: (b) Recording of proximal chest wall movement in direction Y: (c) Recording of distal chest wall movement in direction X.
Fig. 2. Incorrect use of swept-gain to "clean up" chest wall echo complex. (al Minimal swept-gain ; (b) Steep swept-gain profile applied to chest wall echo complex.
33
D. J. FARMAN. G. THOMASand R. J. BLACKWELL
34
i
o
10 $ e c
(a)
I
S m m
10 $ e c
Cb)
Fig. 3. Artifactual recordings due to incorrect use of swept-gain. Upper traces obtained with incorrect swept-gain settings, lower traces with correct settings (a) Fetal chest wall movements; (b) Artifacts due to maternal respiratory movements.
manner to correct for attenuation over the area of interest, it has been used to obtain a highly uneven range of video amplitudes. The result is that as individual echoes in the complex move towards the front of the gate they are suppressed below the trigger threshold of the time-to-voltage analogue circuitry and triggering then results from the next significant echo within the gate, until this also is suppressed to a subthreshold level. The output will therefore be artifactual and appear as a small ripple situated at the position of the suppression edge of the sweptgain profile. A large amplitude artifactual output occurs with the commercial instrument when there is no significant succeeding echo, relating to the fetal chest wall, within the gate. In this case the time-to-yoltage convertor may be triggered by an echo from another structure, or may not be triggered at all. The output has the appearance of an almost rectangular wave whose amplitude bears no relation to the amplitude of the true chest wall movement, although the two frequencies may be identical. This effect is enhanced by the provision of storage circuits which, in the event of this momentary loss of the target echo. gives rise to an output which may be mistakenly interpreted as zero target move-
ment. Similarly, periods of signal loss due to slight fetal movement may be incorrectly interpreted as periods of zero respiratory movement. This situation does not arise with the custombuilt instrument since storage circuits are not incorporated. The swept-gain profile may also cause distortion in the amplitude of the time-to-voltage output. The amplitude of the video output varies with the position of the echo relative to the profile. The resulting change in the amplitude of the echo causes the trigger point on the echo to be dependent on this position. The amplitude of chest wall movement due to respiratory efforts is, by analogy with the adult, of the order of 4 r a m and a trigger point error of l mm (which is quite likely) is very significant. It is clear that swept-gain should never be used to "clean up" an echo complex. The effects described above were demonstrated by pulsing the transducer with the transmitter of the commercial instrument and using simultaneously the receiver and processing circuits of both instruments, the custom-built instrument being used correctly and the commercial instrument used by an operator lacking an appreciation of the possible artifacts but otherwise experienced.
Errors and artifacts encountered in the monitoring of fetal respiratory movements using ultrasound
5mml
,
,
35
(a)
10 sec
(b)
Fig. 4. Artifactual recordings due to abdominal wall movements. (a) Fetal chest wall movements at 38 weeks gestation; (b) "'Movements" of uterine wall, 3 days postpartum.
Sample traces are shown in Fig. 3. Figure 3(a) shows genuine fetal chest wall movements recorded by the two instruments. The output of the commercial instrument is seen to consist of a straight baseline with distinct, often rectangular deflections representing chest wall movements. This does not reflect the true physiological situation, since a straight baseline represents zero movement of the fetal target with respect to the ultrasonic transducer, whereas the output of the custom-built instrument shows some degree of movement to occur continuously. Figure 3 (b) shows artifacts recorded by the commercial instrument due to maternal respiration. This is confirmed by the simultaneous recording obtained with the correctly adjusted custom-built instrument. It was often found impossible to place the transducer in an ideal position since, with the occurrence of either maternal or fetal respiratory movement, there may be a slight change in angle of insonation of the fetal chest wall. These changes are accompanied by a reduction in echo amplitude, which can make monitoring difficult. This effect is diminished by using a wide beamwidth and a longer integration time
constant in the video processor. A large effective beamwidth is achieved with a large aperture transducer used with minimum video suppression and accurate swept-gain compensation. These measures, however, introduce further problems since a wide echo complex may be generated by echoes received from points on the target spaced at different distances from the transducer. The varying echo reflection times involved contribute to a broad echo group. The leading echo in the group may be monitored satisfactorily, but the point on the target giving rise to the leading echo may change as the complex moves. The use of the belt-mounted transducer was also found to lead to the production of recording artifacts. It was observed that the recordings did not show well-defined periods of respiratory movement as reported by other workers, but rather almost continuous movements of varying amplitude. In order to investigate the origin of the smaller-amplitude movements, two patients were examined post-natally using echoes from the anterior uterine wall. Figure 4 shows recordings obtained from one of these patients during the 38th week of pregnancy and again,
(a)
''mI '
lO sec
(b)
Fig. 5. Artifactual recordings due to transducer movement. (a) "'Movements'" of wall of non-pregnant, nulliparous uterus: (b) "'Movements" of hepatic structures.
36
D. J. FARMAN. G. THOMASand R. J. BLACKWELL
three days after delivery. It can be seen that small-amplitude movements, similar in rate to fetal respiratory movements, were recorded post-natally. Some of the uterine wall "movements" are synchronous with the maternal pulse but others resemble respiratory movements. Figure 5 shows recordings from the anterior wall of a non-pregnant, nulliparous uterus, and from echo-producing structures within a human liver, both recordings showing small, rhythmic movements occurring at a rate similar to fetal respiratory movements. When these examinations were repeated using the rigidly mounted transducer and water-bath these movements were no longer recorded, suggesting that they may be attributed to transducer displacement secondary to movement of the abdominal wall. The use of the water-bath was found inconvenient in the clinical situation. For recordings intended for detailed analysis its use would be considered mandatory. The small-amplitude movement artifacts described above could be masked by reducing the output sensitivity of the commercial instrument by increasing the gate period. This procedure, however, also masked genuine, low-amplitude fetal respiratory movements so that only largeamplitude movements were recorded. This also constitutes a ~ource of recording error. CONCLUSIONS
Fetal respiratory movements can be monitored by ultrasonic A-mode techniques, but a number of difficulties are encountered. For accurate results the equipment should be specifically adapted and must be used with great care. The output circuitry must be triggered solely and consistently by echoes from the fetal chest wall, which should be insonated at normal incidence.
Although satisfactory recordings may be obtained by an arbitrary direction of insonation, the use of an ultrasonic section scanner to locate the fetal chest wall greatly increases the likelihood of obtaining artifact-free recordings. The incorrect use of swept-gain facilities may destroy the fidelity of the recordings obtained, and steps must be taken to avoid artifacts due to maternal respiration, maternal and fetal pulse, and maternal abdominal movements. Even when these requirements have been met subsequent fetal movement may make it necessary to reposition the transducer several times during the course of a single examination, thus preventing continuous monitoring over long periods of time, especially during labour, when the uterine contractions contribute to the difficulties. It is concluded that A-mode techniques for recording fetal chest wall movements require further sophistication before they can be said to provide reliable information from which the clinical usefulness of the technique may be assessed. Acknowledgements--The authors gratefully acknowledge
the assistance of Professor D. V. 1. Fairweather and the Consultant Staff of the Obstetric Unit, University College Hospital, in permitting their patients to partake in this study, and of Mr. J. S. Clifton. Director. Medical Physics Department, University College Hospital and Medical School, for making available the facilities of his Department. The authors would also like to acknowledge the generous support of the National Fund for Research into Crippling Diseases in providing finance for the equipment used in this study.
REFERENCES Boddy. K. and Dawes, G. S. (1975) Fetal breathing. Br. Med. Bull. 31, 3. Boddy. K. and Robinson, J. S. (1971) External method for detection of fetal breathing in utero. Lancet 2, 1231. Dawes, G. S.. Fox. H. E.. Leduc, B. M., Liggins, G. C. and Richards, R. T. (1972) Respiratory movements and rapid eye movement sleep in the fetal lamb. J. Physiol. 220, 119.
