British Journal of Obstetrics and Gynaecology
II ~~
NEW SERIES
NOVEMBER 1976
I’
I ‘I
THE THEORY, FEASlBILlTY AND ACCURACY OF AN ULTRASONIC METHOD OF ESTIMATlNG FETAL WEIGHT* BY
J. MORRISON, Senior Lecturer AND
M. J. MCLENNAN, Medical Student Department of Obstetrics and Gynaecology, University of Queensland Mater Misericordiae Mothers’ Hospital, South Brisbane Summary
We present the theory, method, feasibility and accuracy of estimating fetal weight by measuring the fetal volume using compound ultrasonic scanning. The two parameters had a very high coefficient of correlation (0.9794), and the standard error in one series of 20 patients was only &lo6 g. The correlation coefficient achieved by one of us who was new to ultrasound techniques was 0.82, which compared favourably with correlations between fetal weight and bjparietal diameter measurements (r = 0.26). A discussion as to the correction factors required to allow for the unknown value of the velocity of ultrasound in fetal tissues, the clinical use of this method, and the possible means by which the accuracy of estimating fetal weight may be further improved is included. Morrison, 1972; Morrison el al, 1972) it was believed that a method using this technique could be developed which would allow estimates of fetal weight which were sufficiently accurate to be of practicaI use. We now discuss the theory underlying this method, the method itself, its feasibility, and its accuracy.
AN accurate estimate of fetal weight is one of the prime necessities of modern obstetrics. To date, the best methods have been based on ultrasonic scanning techniques, using either the biparietal diameter as reported by Willocks et a1 (1964) and Taylor et a1 (1967), or a combination of skull and chest circumferences, diameters, or cross sectional areas, as reported by Taylor et a1 (1967) and Garrett and Robinson (1971). However, the correlation between these measurements and fetal weight is not sufficiently reliable to be of clinical value. From previous experience with compound ultrasonic scanning (Morrison, I969 ; Morrison and Blackwell, 1969; Morrison et a,, 1969a, 6 , c and d ; Morrison et al, 1970a and 6 ;
THEORY The parameter which is most likely to correlate well with fetal weight is fetal volume (mass = volume x density or M = V x D) provided that fetal density remains fairly constant. Table 1 shows the density of the individual tissue components of the fetus; it is apparent that they tend towards unity. Possibly the greatest change in overall fetal density will occur during the last few weeks of gestation with the accumulation of body fat. However, it is probable that the total density will only shift
* The
material in this article formed the basis of a William Blair-Bell Memorial Lecture by J. Morrison on 31st October, 1974. 833
834
MORRISON AND MCLENNAN
TABLEI
Medium
Density (103kg’m3)
Water Blood Brain Fat Liver Muscle Bone
1.00
1.06 1.03 0.97 1.06 1.03 1.80
volume in cc x fetal density = (fetal cross-sectional areas on echogram) xrxtxuxd where r = correction factor for electronic scale reduction (constant) t = echogram intervals (can be made constant and of such dimensions that the crosssectional area is the mean area for the interval) u = correction factor for the differencz between assumed and true velocities of ultrasound in fetal tissues d = fetal density The formula contains two variable correction factors, namely a compromise velocity of ultrasound in fetal tissues and fetal density. The compromise velocity of ultrasound will obviously lie between 1500 to 1600 m/second (Table I), the precise value depending on the relative proportions of tissue present in each fetus. The most likely variable is fat, and any significant increase in fat content will reduce the compromise velocity of ultrasound in the fetus, giving an erroneously large ultrasonic display. Under these circumstances, the density of the fetus will fall, and thus these two variables will tend to cancel each other out. Fetal weight in g
Acoustic properties of tissues
Velocity of ultrasound (m/second) 1525 1560 1520 1440 1590 1590 3360
marginally and be less than f0.01. A variation of this magnitude (0.01) would in fact only give an error of f10 g, 1 2 0 g, f30 g, f 4 0 g in fetuses respectively weighing 1000 g, 2000 g, 3000 g, and 4000 g in a calculation based on the formula M L- V x D. Fetal volume measurement presents a major difficulty. Compound ultrasonic scanner display methods produce only a two-dimensional cross sectional representation of the underlying tissues. If an irregular object is divided into a number of ‘slices’, then the total volume of the object is the sum of the volumes of each individual slice, and the volume of each slice is the mean cross-sectional area of that slice multiplied by its mean ‘thickness’. Thus, serial echograms taken in parallel planes along the length of the fetus at set intervals will provide an assessment of its volume. But the accuracy of the cross-sectional area displayed on the echogram is influenced by several factors. Firstly, in order to demonstrate complete areas in one plane on the small oscilloscope screen, the area is reduced electronically by a standard value, which in the case of our apparatus is 1 : 9. Secondly, the beam width of the transducer is of finite size and the echogram is a self-averaging display over this width, which is approximately 2 cm with our apparatus. Finally, the size of the ultrasonic display is based on a presumed value for the velocity of ultrasound in the tissues scanned. At present there is no known composite value for the fetus and, for any empirical value used, the display size will differ quantitatively from life size by a factor related to the difference between these two velocities. As Mass = VolumexDensity, we may use the following formula to calculate fetal weight:
= fetal
METHODS AND PATIENTS Serial transverse parallel echograms at 2 cm intervals covering the entire length of the fetus were recorded, using a modified (Morrison, 1974) Kretztechnik Combison Unit (compound ultrasonic scanner), with a 1 mH transducer and an empirical velocity of ultrasound of 1500 m/second. Examinations were conducted on patients with a full bladder (Donald, 1968) and the biparietal diameters were measured using the method described by Campbell (1968). The cross-sectional fetal areas on the echograms were calculated with an Aristo planimeter and the fetal volume was calculated by the formulae given previously. No correction was made for the true compromise velocity of ultrasound in fetal tissues. The feasibility study was carried out entirely
835
ULTRASONIC ESTIMATION OF FETAL WEIGHT
by M. J. McLennan on 25 patients attending the Mater Misericordiae Mothers’ Hospital antenatal clinic who were selected by the obstetric Registrars for probable diversity of fetal weights, and the likelihood that delivery would occur within two days of the ultrasound examination. No communication as to the clinical evaluation of fetal size was permitted between M. J. McLennan and the Registrars. All ultrasonic calculations were done by M . J. McLennan and the birth weight of the infant was checked and recorded by the nursing staff. To test the accuracy of the method, 25 other patients were similarly selected, but both authors were involved in making ultrasound measurements and calculations.
RESULTS Feasibility study The first three patients in the study were used to familiarize M. J. McLennan with the ultrasonic apparatus, the methods of measuring fetal volume and biparietal diameter, and the identification of fetal tissues. These three patients and one more (No. 21, who was delivered more than 48 hours after ultrasound examination) were excluded from the final analysis. The correlation between birth weight of the infant and the ultrasonic measurement of fetal volume was highly significant (p< The coefficient of correlation was 0.8178 and the regression line between these two factors is expressed by the equation: y = 234.4+0.409~ where y is the ultrasound fetal volume (in cc) and x is the birth weight in grams. The relationship between fetal volume and birth weight is shown in Figure I , and the regression coefficient (0.409) and its standard error (fO.063) demonstrates that this line is a good fit for the data. The standard error of the regression line is +230.86 g. The correlation between birth weight of’ the injant and the ultrasonic measurement of the fetal biparietal diameter was much less significant (p = 0.66) than the correlation between fetal volume and birth weight. The coefficient of correlation was 0.2622. The regression line (Figure 2) in this series is expressed by the equation : y = 456*7+281.3~
1-6
I
I
2.0
24
I
I
I
1
I
2.0
3.2
3.6
4-0
4.4
Birth weight (kg)
FIG.1 Correlation between fetal volume (measured by ultrasound) and birth weight.
where y is the birth weight in grams and x is the biparietal diameter in cm. The standard error ( f O ~063) of the regression coefficient (0 -2622) confirmed that the biparietal diameter had a predictive value in estimating fetal weight, but the standard error of the regression line was f494 g.
Accuracy of method Of the 25 patients, 20 were retained for analysis. The remaining 5 were excluded because
-10-0-
E
w
L
w
E
.-
0
1
c
9.0-
0 I ._
m
. .
836
MORRISON AND MCLENNAN
they were delivered more than 48 hours after the ultrasound examination. The comparison of ultrasonic fetal volume and birth weight in the 20 patients is shown in Table I1 and
-
TABLEI1 Fetal volume and birth weight
Patient
Ultrasonic fetal volume (cc)
Birth weight (gm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
I270 1714 1456 1463 1837 1175 1585 1533 1476 1611 1312 1523 1674 1762 1778 I545 863 1550 2098 1495
2409 3658 2770 3060 3689 2352 3009 2975 2900 3220 2825 2975 3231 3577 3435 3090 1701 3029 4253 3033
16
20
24
28 32 36 Birth welght ( k g )
Figure 3. The correlation coefficient between these two parameters is 0-9794 which is very highly significant (p< I O-6). The regression line in the series is expressed by the following equation: birth weight = (-65.84)+2.03 (ultrasonic volume) and the standard error of this line is 106 84 g.
40
44
FIG.3 Correlation between fetal volume (measured by ultrasound) and birth weight.
DISCUSSION The accuracy of estimating fetal weight by ultrasonic measurement of fetal volume (even by someone relatively inexperienced in ultrasound) is clearly demonstrated. The best standard error achieved was such that in a normal term infant it would be 5 3 per cent of the birth weight. This compares very favourably with estimation of birth weight from biparietal diameter; our standard error for that regression line was h494 g, a value similar to that found by Willocks et aZ(1964), Thomson et aZ(1965), and also those authors who additionally used measurements of the cross-sectional area of the fetal trunk in their calculations (Garrett and Robinson, 1970). The weekly weight gain by the fetus over the last four weeks of gestation is of the order of 250 g and so the volumetric method of estimating fetal weight can be of clinical value in assessing growth rates. If the information gained is used in conjunction with biparietal diameter measurements, which more clearly reflect gestational maturity, we have available a method for the antenatal detection of dysmaturity. The estimation of fetal cross-sectional areas is, in our opinion, much more liable to error from operator inexperience than is the measurement of the biparietal diameter. As any error made in the former measurements is magnified some 36 times in calculating the volume of the fetus, it is probable that the most accurate correlations between fetal volume and weight would be achieved by those experienced in ultrasound. The apparently large difference between the ultrasonic fetal volume and the actual weight of the fetus is probably due to a lack of knowledge of the compromise velocity of ultrasound in fetal tissues. For example, if we suppose that the density of fetal tissues is 1 .OO and that our empirical value of ultrasound velocity is exactly 1500 m/second, then we have an apparent loss 5
-
ULTRASONIC ESTIMATION OF FETAL WEIGHT
of cross-sectional area of 100per cent (correlation coefficient 2 -03). However, due to electronic reduction of image size (I : 9) and 2 cm interval echograms, any error is magnified 18 times. The true error in cross-sectional area is therefore 100/18 per cent = 5.55 per cent, and the error in ultrasound velocity is therefore 5.55 per cent = 2 - 4 per cent. The true compromise velocity of ultrasound in fetal tissues under these circumstances is therefore 1500+2 a 4 per cent m/second, i.e. 1536 m/second. Since such a minor error can make a major difference in the final result, it is apparent that the accuracy of this technique should be further improved. The interval between echograms could be 1 or 0.5 cm, which would reduce the correction factor and give a more representative series of fetal cross-sectional areas. Tncreased clarity of echograms, and the elimination of reduction scale factor would also help. The latter could be achieved by direct imaging on large screen oscilloscopes (at present this would mean some loss of clarity), by optical enlargement of the small screen image prior to measuring, or by direct analysis of the ultrasonic image by computer. ACKNOWLEDGEMENTS We record our gratitude to the National Health and Medical Research Council who supplied both the apparatus and funds for modifying the apparatus, and to the obstetrical Registrars, Dr L. Yared and T. Di Francesco, for their helpful co-operation in this project. We also thank the Honorary Obstetricians of the Mater Misericordiae Mothers' Hospital for their kindness in allowing access to their patients.
837
REFERENCES Campbell, S. (1968):Journalof Obstetricsand Gynaecology of the British Commonwealth, 75, 568. Donald, I. ( 1 968) : British Medical Bulletin, 24,7 1. Garrett, W. J., and Robinson, D. E. (1970): Ultrasoundin Clinical Obstetrics. Charles C. Thomas, Springfield, Illinois, p 47. Garrett, W. J., and Robinson, D. E. (1971): Obstetrics and Gynecology, 38, 527. Morrison, J. (1969): Proceedings of the 8th International Conference of Medical and Biological Engineering, Chicago Session, 32, 4. Morrison, J., and Blackwell, R. J . (1969): Australian and New Zealand Journal of Obstetrics and Gynaecology, 9, 201. Morrison, J., Kohorn, E. I., Secker-Walker, R. H., and : Journal of Obstetrics Campbell, S. ( 1 9 6 9 ~ )American and Gynecology, 103, 868. Morrison, J., Kohorn, E. I., Ashford, D., and Blackwell, R. J. (19696): Obstetrics and Gynecology, 34, 515. Morrison, J., Kohorn, E. I., and Secker-Walker, R. H . ( 1 9 6 9 ~ )Proceedings : of the Royal Society of Medicine, 62, 446. Morrison, J., Kohorn, E. I., Ashford, C., Tredgold, C., Secker-Walker, R. H., and Blackwell, R. J. (19694'): Australian and New Zealand Journal of Obstetrics and Gynaecology, 9, 206. Morrison, J., Kohorn, E. I., and Blackwell, R. J . ( 1 9 7 0 ~ ) : Australian and New Zealand Journal of Obstetrics and Gynaecology, 10, 4. Morrison, J., Kohorn, E. I., and Blackwell, R. J . (1970b): Australian and New Zealand Journal of Obstetrics and Gynaecology, 10,4. Morrison, J . (1972): Australian and New Zealand Journul of Obstetrics and Gynaecology, 12, 182. Morrison, J., Lachelin, G . C., and Blackwell, R. J . (1972): Australian and New Zealand Journal of Obstetrics and Gynaecology, 12, 220. Morrison, J. (1974): Ulfrasonics,1, 132. Taylor, E. S., Thompson, H. E., Gottesfeld, K. R., and Holmes, J. H. ( I 967): American Journal of Obstetrics and Gynecology, 99, 671. Thompson, H., Holmes, J. H., Gottesfeld, K. R., and Taylor, E. S. (1965): American Journal of Obstetrics and Gynecology, 92, 44. Willocks, J., Donald, I., Duggan, T., and Day, N. (1964): Journal of Obstetrics and Gynaecology of the British Commonwealth, 71, I I .
II ~~
NEW SERIES
NOVEMBER 1976
I’
I ‘I
THE THEORY, FEASlBILlTY AND ACCURACY OF AN ULTRASONIC METHOD OF ESTIMATlNG FETAL WEIGHT* BY
J. MORRISON, Senior Lecturer AND
M. J. MCLENNAN, Medical Student Department of Obstetrics and Gynaecology, University of Queensland Mater Misericordiae Mothers’ Hospital, South Brisbane Summary
We present the theory, method, feasibility and accuracy of estimating fetal weight by measuring the fetal volume using compound ultrasonic scanning. The two parameters had a very high coefficient of correlation (0.9794), and the standard error in one series of 20 patients was only &lo6 g. The correlation coefficient achieved by one of us who was new to ultrasound techniques was 0.82, which compared favourably with correlations between fetal weight and bjparietal diameter measurements (r = 0.26). A discussion as to the correction factors required to allow for the unknown value of the velocity of ultrasound in fetal tissues, the clinical use of this method, and the possible means by which the accuracy of estimating fetal weight may be further improved is included. Morrison, 1972; Morrison el al, 1972) it was believed that a method using this technique could be developed which would allow estimates of fetal weight which were sufficiently accurate to be of practicaI use. We now discuss the theory underlying this method, the method itself, its feasibility, and its accuracy.
AN accurate estimate of fetal weight is one of the prime necessities of modern obstetrics. To date, the best methods have been based on ultrasonic scanning techniques, using either the biparietal diameter as reported by Willocks et a1 (1964) and Taylor et a1 (1967), or a combination of skull and chest circumferences, diameters, or cross sectional areas, as reported by Taylor et a1 (1967) and Garrett and Robinson (1971). However, the correlation between these measurements and fetal weight is not sufficiently reliable to be of clinical value. From previous experience with compound ultrasonic scanning (Morrison, I969 ; Morrison and Blackwell, 1969; Morrison et a,, 1969a, 6 , c and d ; Morrison et al, 1970a and 6 ;
THEORY The parameter which is most likely to correlate well with fetal weight is fetal volume (mass = volume x density or M = V x D) provided that fetal density remains fairly constant. Table 1 shows the density of the individual tissue components of the fetus; it is apparent that they tend towards unity. Possibly the greatest change in overall fetal density will occur during the last few weeks of gestation with the accumulation of body fat. However, it is probable that the total density will only shift
* The
material in this article formed the basis of a William Blair-Bell Memorial Lecture by J. Morrison on 31st October, 1974. 833
834
MORRISON AND MCLENNAN
TABLEI
Medium
Density (103kg’m3)
Water Blood Brain Fat Liver Muscle Bone
1.00
1.06 1.03 0.97 1.06 1.03 1.80
volume in cc x fetal density = (fetal cross-sectional areas on echogram) xrxtxuxd where r = correction factor for electronic scale reduction (constant) t = echogram intervals (can be made constant and of such dimensions that the crosssectional area is the mean area for the interval) u = correction factor for the differencz between assumed and true velocities of ultrasound in fetal tissues d = fetal density The formula contains two variable correction factors, namely a compromise velocity of ultrasound in fetal tissues and fetal density. The compromise velocity of ultrasound will obviously lie between 1500 to 1600 m/second (Table I), the precise value depending on the relative proportions of tissue present in each fetus. The most likely variable is fat, and any significant increase in fat content will reduce the compromise velocity of ultrasound in the fetus, giving an erroneously large ultrasonic display. Under these circumstances, the density of the fetus will fall, and thus these two variables will tend to cancel each other out. Fetal weight in g
Acoustic properties of tissues
Velocity of ultrasound (m/second) 1525 1560 1520 1440 1590 1590 3360
marginally and be less than f0.01. A variation of this magnitude (0.01) would in fact only give an error of f10 g, 1 2 0 g, f30 g, f 4 0 g in fetuses respectively weighing 1000 g, 2000 g, 3000 g, and 4000 g in a calculation based on the formula M L- V x D. Fetal volume measurement presents a major difficulty. Compound ultrasonic scanner display methods produce only a two-dimensional cross sectional representation of the underlying tissues. If an irregular object is divided into a number of ‘slices’, then the total volume of the object is the sum of the volumes of each individual slice, and the volume of each slice is the mean cross-sectional area of that slice multiplied by its mean ‘thickness’. Thus, serial echograms taken in parallel planes along the length of the fetus at set intervals will provide an assessment of its volume. But the accuracy of the cross-sectional area displayed on the echogram is influenced by several factors. Firstly, in order to demonstrate complete areas in one plane on the small oscilloscope screen, the area is reduced electronically by a standard value, which in the case of our apparatus is 1 : 9. Secondly, the beam width of the transducer is of finite size and the echogram is a self-averaging display over this width, which is approximately 2 cm with our apparatus. Finally, the size of the ultrasonic display is based on a presumed value for the velocity of ultrasound in the tissues scanned. At present there is no known composite value for the fetus and, for any empirical value used, the display size will differ quantitatively from life size by a factor related to the difference between these two velocities. As Mass = VolumexDensity, we may use the following formula to calculate fetal weight:
= fetal
METHODS AND PATIENTS Serial transverse parallel echograms at 2 cm intervals covering the entire length of the fetus were recorded, using a modified (Morrison, 1974) Kretztechnik Combison Unit (compound ultrasonic scanner), with a 1 mH transducer and an empirical velocity of ultrasound of 1500 m/second. Examinations were conducted on patients with a full bladder (Donald, 1968) and the biparietal diameters were measured using the method described by Campbell (1968). The cross-sectional fetal areas on the echograms were calculated with an Aristo planimeter and the fetal volume was calculated by the formulae given previously. No correction was made for the true compromise velocity of ultrasound in fetal tissues. The feasibility study was carried out entirely
835
ULTRASONIC ESTIMATION OF FETAL WEIGHT
by M. J. McLennan on 25 patients attending the Mater Misericordiae Mothers’ Hospital antenatal clinic who were selected by the obstetric Registrars for probable diversity of fetal weights, and the likelihood that delivery would occur within two days of the ultrasound examination. No communication as to the clinical evaluation of fetal size was permitted between M. J. McLennan and the Registrars. All ultrasonic calculations were done by M . J. McLennan and the birth weight of the infant was checked and recorded by the nursing staff. To test the accuracy of the method, 25 other patients were similarly selected, but both authors were involved in making ultrasound measurements and calculations.
RESULTS Feasibility study The first three patients in the study were used to familiarize M. J. McLennan with the ultrasonic apparatus, the methods of measuring fetal volume and biparietal diameter, and the identification of fetal tissues. These three patients and one more (No. 21, who was delivered more than 48 hours after ultrasound examination) were excluded from the final analysis. The correlation between birth weight of the infant and the ultrasonic measurement of fetal volume was highly significant (p< The coefficient of correlation was 0.8178 and the regression line between these two factors is expressed by the equation: y = 234.4+0.409~ where y is the ultrasound fetal volume (in cc) and x is the birth weight in grams. The relationship between fetal volume and birth weight is shown in Figure I , and the regression coefficient (0.409) and its standard error (fO.063) demonstrates that this line is a good fit for the data. The standard error of the regression line is +230.86 g. The correlation between birth weight of’ the injant and the ultrasonic measurement of the fetal biparietal diameter was much less significant (p = 0.66) than the correlation between fetal volume and birth weight. The coefficient of correlation was 0.2622. The regression line (Figure 2) in this series is expressed by the equation : y = 456*7+281.3~
1-6
I
I
2.0
24
I
I
I
1
I
2.0
3.2
3.6
4-0
4.4
Birth weight (kg)
FIG.1 Correlation between fetal volume (measured by ultrasound) and birth weight.
where y is the birth weight in grams and x is the biparietal diameter in cm. The standard error ( f O ~063) of the regression coefficient (0 -2622) confirmed that the biparietal diameter had a predictive value in estimating fetal weight, but the standard error of the regression line was f494 g.
Accuracy of method Of the 25 patients, 20 were retained for analysis. The remaining 5 were excluded because
-10-0-
E
w
L
w
E
.-
0
1
c
9.0-
0 I ._
m
. .
836
MORRISON AND MCLENNAN
they were delivered more than 48 hours after the ultrasound examination. The comparison of ultrasonic fetal volume and birth weight in the 20 patients is shown in Table I1 and
-
TABLEI1 Fetal volume and birth weight
Patient
Ultrasonic fetal volume (cc)
Birth weight (gm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
I270 1714 1456 1463 1837 1175 1585 1533 1476 1611 1312 1523 1674 1762 1778 I545 863 1550 2098 1495
2409 3658 2770 3060 3689 2352 3009 2975 2900 3220 2825 2975 3231 3577 3435 3090 1701 3029 4253 3033
16
20
24
28 32 36 Birth welght ( k g )
Figure 3. The correlation coefficient between these two parameters is 0-9794 which is very highly significant (p< I O-6). The regression line in the series is expressed by the following equation: birth weight = (-65.84)+2.03 (ultrasonic volume) and the standard error of this line is 106 84 g.
40
44
FIG.3 Correlation between fetal volume (measured by ultrasound) and birth weight.
DISCUSSION The accuracy of estimating fetal weight by ultrasonic measurement of fetal volume (even by someone relatively inexperienced in ultrasound) is clearly demonstrated. The best standard error achieved was such that in a normal term infant it would be 5 3 per cent of the birth weight. This compares very favourably with estimation of birth weight from biparietal diameter; our standard error for that regression line was h494 g, a value similar to that found by Willocks et aZ(1964), Thomson et aZ(1965), and also those authors who additionally used measurements of the cross-sectional area of the fetal trunk in their calculations (Garrett and Robinson, 1970). The weekly weight gain by the fetus over the last four weeks of gestation is of the order of 250 g and so the volumetric method of estimating fetal weight can be of clinical value in assessing growth rates. If the information gained is used in conjunction with biparietal diameter measurements, which more clearly reflect gestational maturity, we have available a method for the antenatal detection of dysmaturity. The estimation of fetal cross-sectional areas is, in our opinion, much more liable to error from operator inexperience than is the measurement of the biparietal diameter. As any error made in the former measurements is magnified some 36 times in calculating the volume of the fetus, it is probable that the most accurate correlations between fetal volume and weight would be achieved by those experienced in ultrasound. The apparently large difference between the ultrasonic fetal volume and the actual weight of the fetus is probably due to a lack of knowledge of the compromise velocity of ultrasound in fetal tissues. For example, if we suppose that the density of fetal tissues is 1 .OO and that our empirical value of ultrasound velocity is exactly 1500 m/second, then we have an apparent loss 5
-
ULTRASONIC ESTIMATION OF FETAL WEIGHT
of cross-sectional area of 100per cent (correlation coefficient 2 -03). However, due to electronic reduction of image size (I : 9) and 2 cm interval echograms, any error is magnified 18 times. The true error in cross-sectional area is therefore 100/18 per cent = 5.55 per cent, and the error in ultrasound velocity is therefore 5.55 per cent = 2 - 4 per cent. The true compromise velocity of ultrasound in fetal tissues under these circumstances is therefore 1500+2 a 4 per cent m/second, i.e. 1536 m/second. Since such a minor error can make a major difference in the final result, it is apparent that the accuracy of this technique should be further improved. The interval between echograms could be 1 or 0.5 cm, which would reduce the correction factor and give a more representative series of fetal cross-sectional areas. Tncreased clarity of echograms, and the elimination of reduction scale factor would also help. The latter could be achieved by direct imaging on large screen oscilloscopes (at present this would mean some loss of clarity), by optical enlargement of the small screen image prior to measuring, or by direct analysis of the ultrasonic image by computer. ACKNOWLEDGEMENTS We record our gratitude to the National Health and Medical Research Council who supplied both the apparatus and funds for modifying the apparatus, and to the obstetrical Registrars, Dr L. Yared and T. Di Francesco, for their helpful co-operation in this project. We also thank the Honorary Obstetricians of the Mater Misericordiae Mothers' Hospital for their kindness in allowing access to their patients.
837
REFERENCES Campbell, S. (1968):Journalof Obstetricsand Gynaecology of the British Commonwealth, 75, 568. Donald, I. ( 1 968) : British Medical Bulletin, 24,7 1. Garrett, W. J., and Robinson, D. E. (1970): Ultrasoundin Clinical Obstetrics. Charles C. Thomas, Springfield, Illinois, p 47. Garrett, W. J., and Robinson, D. E. (1971): Obstetrics and Gynecology, 38, 527. Morrison, J. (1969): Proceedings of the 8th International Conference of Medical and Biological Engineering, Chicago Session, 32, 4. Morrison, J., and Blackwell, R. J . (1969): Australian and New Zealand Journal of Obstetrics and Gynaecology, 9, 201. Morrison, J., Kohorn, E. I., Secker-Walker, R. H., and : Journal of Obstetrics Campbell, S. ( 1 9 6 9 ~ )American and Gynecology, 103, 868. Morrison, J., Kohorn, E. I., Ashford, D., and Blackwell, R. J. (19696): Obstetrics and Gynecology, 34, 515. Morrison, J., Kohorn, E. I., and Secker-Walker, R. H . ( 1 9 6 9 ~ )Proceedings : of the Royal Society of Medicine, 62, 446. Morrison, J., Kohorn, E. I., Ashford, C., Tredgold, C., Secker-Walker, R. H., and Blackwell, R. J. (19694'): Australian and New Zealand Journal of Obstetrics and Gynaecology, 9, 206. Morrison, J., Kohorn, E. I., and Blackwell, R. J . ( 1 9 7 0 ~ ) : Australian and New Zealand Journal of Obstetrics and Gynaecology, 10, 4. Morrison, J., Kohorn, E. I., and Blackwell, R. J . (1970b): Australian and New Zealand Journal of Obstetrics and Gynaecology, 10,4. Morrison, J . (1972): Australian and New Zealand Journul of Obstetrics and Gynaecology, 12, 182. Morrison, J., Lachelin, G . C., and Blackwell, R. J . (1972): Australian and New Zealand Journal of Obstetrics and Gynaecology, 12, 220. Morrison, J. (1974): Ulfrasonics,1, 132. Taylor, E. S., Thompson, H. E., Gottesfeld, K. R., and Holmes, J. H. ( I 967): American Journal of Obstetrics and Gynecology, 99, 671. Thompson, H., Holmes, J. H., Gottesfeld, K. R., and Taylor, E. S. (1965): American Journal of Obstetrics and Gynecology, 92, 44. Willocks, J., Donald, I., Duggan, T., and Day, N. (1964): Journal of Obstetrics and Gynaecology of the British Commonwealth, 71, I I .
