Ultrasound has become the essential tool of modern obstetric practice. With advances in technology and computer processing, what was once a mere curiosity has become crucial for the assessment of the placenta, membranes, fluid, and fetal anatomy, as is covered in the other portions of this text. The assignment of pregnancy age is the first task placed before the care provider, and ultrasound is the key modality used for this purpose. Additionally, ultrasound is far superior to the clinical examination for determining adequacy of interval fetal growth. With the development of Doppler technology, one can now assess fetal status to determine pathology earlier than as evidenced by abnormal fetal cardiotocography (CTG). As a result, Doppler ultrasound has become crucial for making management decisions in some high risk settings. The methods for dating a pregnancy, measuring growth, and assessing fetal status using Doppler technology are covered in this chapter.
Sonography provides an enormous amount of useful information to the practicing obstetrician. Arguably, the single most useful piece of information that obstetric sonography provides is an accurate determination of menstrual age. It is difficult to imagine a clinical problem encountered during a pregnancy in which an accurate menstrual age is not highly desired before proceeding with an appropriate management plan.
It is important to establish from the outset what is meant by the term menstrual age and why it is so important in clinical obstetrics. Fetal age actually begins at conception, and an equivalent term is conceptional age. However, classically and by current convention, obstetricians date pregnancies in menstrual weeks, beginning from the 1st day of the last normal menstrual period. The appropriate term for this method of dating is menstrual age, and this term is used exclusively in this chapter when referring to the duration of a pregnancy.1 Many obstetricians also use the term gestational age (GA). Although GA should be equivalent to conceptional age, in clinical obstetric practice GA is used interchangeably with menstrual age. The term conceptional age should be reserved to describe pregnancies in which the actual date of conception is known; this is uncommon and is usually restricted to patients who have undergone assisted reproduction (e.g., in vitro fertilization, artificial insemination). Even if conceptional age is known, the menstrual age should be calculated from the conceptional age, based on the assumption of midcycle ovulation (menstrual age = conceptional age + 14 days). Once this has been done, the menstrual age is established and should never be changed later in pregnancy. Subsequent fetal measurements then become an index of fetal growth rather than menstrual age.
Knowledge of menstrual age is important to the obstetrician because it affects clinical management in a number of important ways. First, knowledge of menstrual age is used in early pregnancy for the scheduling of invasive procedures such as chorionic villus sampling and genetic amniocentesis, and in the interpretation of biochemical tests such as expanded maternal serum biomarker screening (“quad-screen”) for risk assessment for neural tube defects and chromosomal anomalies (e.g., trisomy 21 and 18) in which the normal range of values changes over time. Second, knowledge of menstrual age allows the obstetrician to anticipate normal spontaneous delivery or to plan elective delivery within the time frame of a term pregnancy (37 to 42 weeks); this also allows the physician to institute measures that will optimize fetal outcome when labor ensues before 37 weeks or fails to ensue after 42 weeks. Third, knowledge of menstrual age is important in evaluating fetal growth because the normal range for the size of any fetal parameter changes with advancing age.1 Thus, a fetal weight of 2000 g would be normal at 33 weeks but would indicate growth restriction at 36 weeks. When an anomaly is discovered sonographically, the mother’s choices are heavily influenced by menstrual age. Virtually all important clinical decisions require knowledge of the menstrual age.
Before the advent of sonography, menstrual age was established by the patient’s menstrual history, corroborated, preferably during early pregnancy, by physical examination of uterine size, and confirmed in the postnatal period by physical examination of the neonate.2,3 All three of these parameters alone or in combination were notoriously inaccurate (Table 7-1), but the menstrual history could be especially misleading for a number of reasons. First, many women may not accurately recall the 1st day of the last normal menstrual period (LMP), particularly if they are not trying to conceive. Also some women commonly misunderstand the question posed and report the last day instead of the 1st day of their last period. Moreover, for those who do remember their LMP, it may be unreliable or misleading because of oligomenorrhea, abnormal bleeding events, use of hormonal contraceptives, becoming pregnant in the first ovulatory cycle after a recent delivery, or ovulating very early (<day 11) or very late (>day 21) in the menstrual cycle. This latter point may be particularly important, because Matsumoto, Nogami, and Ohkuri4 reported that early or late ovulation occurs in approximately 20% of the population (Fig. 7-1). Indeed, many pregnancy dating formulae and calculator wheels are based on a 280-day gestation starting from the time of the LMP, which is, in turn, based upon a regular, 28-day menstrual cycle composed of 14 days in the follicular and luteual phases each. It is not surprising, then, that Campbell et al5 and Waldenström, Axelsson, and Nilsson6 found that, even in patients with optimal menstrual histories, a single second trimester biparietal diameter (BPD) measurement was more predictive of the estimated date of delivery (EDD) than the date of delivery calculated from the 1st day of the last normal menstrual period. One often hears the statements “This patient has good dates” or “That patient has bad dates.” What constitutes “good” as opposed to “bad” dates? Optimal menstrual histories are present when the patient has a certain last normal menstrual period (preferably recorded on a calendar), regular menses, no exposure to hormonal contraceptives, and no unusual bleeding. Thus, it should not be surprising that the most common indications for obstetric sonograms are related to uncertainty regarding the menstrual age.
|Clinical or Sonographic Parameter||Variability Estimate (+2 SD)|
|In vitro fertilization*||± 1 day|
|Ovulation induction*||± 3 days|
|Artificial insemination*||± 3 days|
|Single intercourse record*||± 3 days|
|Basal body temperature record*||± 4 days|
|First trimester physical examination||± 2 weeks|
|Second trimester physical examination||± 4 weeks|
|Third trimester physical examination||± 6 weeks|
|First trimester sonographic examination (crown rump ± 18% of the estimate length)|
|Second trimester sonographic examination (head ± 18% of the estimate circumference, femur length)|
|Third trimester sonographic examination (head ± 8% of the estimate circumference, femur length)|
SD, standard deviation.
Adapted from James D. Bowie, M.D., Duke University Medical Center.
FIGURE 7-1 Length of the follicular phase of the menstrual cycle in a large number of patients studied by Matsumoto, Nogami, and Ohkuri.4 Note that the distribution is skewed to the right, with a higher number of patients ovulating late in the cycle (>day 21) than earlier in the cycle (< day 11). This corresponds to the findings in the study by Waldenstrom, Axelsson, and Nilsson,6 in which an early midtrimester ultrasound study agreed with the menstrual history in only about 80% of the cases. In 3%, the age by ultrasound evaluation was greater than expected from the optimal menstrual history (corresponding to early ovulation), and in about 17%, the age by ultrasound evaluation was less than expected from the optimal menstrual history (indicating late ovulation).
(Adapted from Matsumoto S, Nogami Y, Ohkuri S: Statistical studies on menstruation: A criticism on the definition of normal menstruation. Gumma J Med Sci 11:294, 1962.)
Sonographic studies designed to evaluate the duration of pregnancy are based on measurements of the fetus, using size as an indirect indicator of menstrual age.1,7–32 These studies usually relied on cross-sectional evaluation of large numbers of patients with known dates of the beginning of the last normal menstrual period and no compounding variables in the menstrual history to question its validity. Rossavik and Fishburne33 demonstrated that such populations are equivalent to populations with known conception dates for studies of this type. Most studies also eliminated patients with multiple gestations and those with a history that might adversely affect fetal growth. In a properly designed cross-sectional analysis of any fetal biometric parameter, measurements are made in a large number of fetuses evenly distributed over the entire range of menstrual ages, with each fetus being measured only once in gestation; the latter point is important in avoiding bias in variability estimates. The data are then analyzed using regression analysis, with menstrual age as the dependent variable, and equations are generated that will predict menstrual age for any given measurement or set of measurements. Most of the published tables that provide predictions of menstrual age from sonographic measurements have been generated in this way.1
The value of any given studied biometric parameter (e.g., BPD, head circumference [HC], femur length [FL], ear lobe length, nostril width) is based on ease of obtaining the measurement and the accuracy with which it predicts menstrual age. A measurement that is easily obtained but inaccurate for judging menstrual age is of little value. As well, a measurement that accurately predicts menstrual age but is very difficult to obtain is also usually not valuable. In inexperienced hands, a difficult measurement is often inaccurately obtained, thus nullifying its salutary effect on age prediction. The measurements currently recommended by the American Institute of Ultrasound in Medicine (BPD, HC, abdominal circumference [AC], and FL) are adequate for the purpose of estimating menstrual age; any additional measurement to predict menstrual age must improve current predictors, and thus far none has emerged.
Regardless of the number of sonographic measurements that one uses in predicting menstrual age, it is very important to remember that this is an inference of age from size, and to understand the variability that is associated with any such estimate. The variability, usually the result of measurement error or actual biologic variability in size, is expressed as ±2 standard deviations, which should be applicable to 95% of the fetuses in a normal population. One must always keep in mind, however, that 5% of the time estimates will be outside this range. Inspection of the magnitude of the maximum errors observed in the original regression data will give a general idea about the largest mistake one could make prospectively predicting menstrual age in a clinical case setting.
Reporting a menstrual age estimate to one or two decimal places for a given fetal measurement or set of measurements may result in an unreasonable assumption regarding the degree of accuracy by the patient’s primary clinician. Importantly, when reporting the menstrual age estimate, it is wise to avoid potential misunderstanding. Cardinal numbers (1, 2, 3, and so on) are used for counting. Ordinal numbers are used to express succession (first, second, third, and so on). Although often used interchangeably, they are not. A fetus in the 20th menstrual week of pregnancy is between the ages of 19 weeks 0 days and 19 weeks 6 days, whereas a fetus that is 20 menstrual weeks is 20 weeks 0 days. The cardinal number is more precise. Sonographic estimates of menstrual age should be reported in cardinal numbers.34
As noted earlier, the BPD5,6 was more predictive of the EDD than the EDD calculated from the 1st day of the last normal menstrual period. Thus, the gold standard that was used to test the validity of sonographic fetal biometry was less accurate than the sonographic measurement itself. The result is that the breadth of the variation about the mean for any sonographic measurement is in part due to the inaccuracy imposed by using the LMP to construct the axis of the curve defining the sonographic measurement. Understanding variation in measurements is very important. However, in practice, one should always choose the mean measurement for clinical case management. The principal reason to know the variation of a given measurement prediction is to gauge the potential danger to the patient if the mean estimate is very far from reality. During the remainder of this section, clues for gauging the potential for error will be provided.
The earliest unequivocal sign of pregnancy using sonographic evaluation is the demonstration of the “gestational sac.”35–41 Interestingly, the term gestational sac is largely an invention of the early sonologists, who used the term to describe the ring-like structure they identified within the womb of women with positive pregnancy tests. The more precise term is chorionic sac. It is the developing chorionic villi that generate the bright echogenic ring noted near the endometrial cavity. The fluid contained therein is almost totally chorionic fluid during very early development. With early articulated-arm equipment (static imaging) and transabdominal real-time equipment the gestational sac could not be visualized until approximately 6 menstrual weeks, but with the new high-resolution real-time equipment, particularly those equipped with transvaginal probes, the gestational sac can usually be seen by 5 menstrual weeks. At this early stage in gestation, the average internal diameter of the gestational sac, calculated as the mean of the anteroposterior diameter, the transverse diameter and the longitudinal diameter (the so-called mean sac diameter [MSD]), can provide an estimation of menstrual age in a normally developing pregnancy (Table 7-2). Importantly, the measurement of the MSD is obtained from the interface of the chorionic villi and the chorionic fluid. Therefore, it represents an estimate of the mean internal diameter of the chorionic sac. The wall of the sac is not included.
|Mean Gestational Sac Diameter (mm)||Predicted Age Range (week) = 95% CI*||Predicted hCG (mIU/mL) Range = 95% CI†|
|2||5.0 (4.5–5.5)||1164 (629–2188)|
|3||5.1 (4.6–5.6)||1377 (771–2589)|
|4||5.2 (4.8–5.7)||1629 (863–3036)|
|5||5.4 (4.9–5.8)||1932 (1026–3636)|
|6||5.5 (5.0–6.0)||2155 (1226–4256)|
|7||5.6 (5.1–6.1)||2704 (1465–4990)|
|9||5.9 (5.4–6.3)||3785 (2085–6870)|
|10||6.0 (5.5–0.5)||4470 (2400–0075)|
|11||6.1 (5.6–6.6)||5297 (2952–9058)|
|12||6.2 (5.8–6.7)||6267 (3502–11,218)|
|13||6.4 (5.9–6.8)||7415 (4145–13,267)|
|14||6.5 (6.0–7.0)||8773 (4894–15,726)|
|15||6.6 (6.2–7.1)||10,379 (5767–18,682)|
|16||6.7 (6.3–7.2)||12,270 (6776–22,235)|
|17||6.9 (6.4–7.3)||14,528 (7964–26,501)|
|18||7.0 (6.5–7.5)||17,188 (9343–31,621)|
|19||7.1 (6.6–7.6)||20,337 (10,951–37,761)|
|20||7.3 (6.8–7.7)||24,060 (12,820–45,130)|
|21||7.4 (6.9–7.8)||28,464 (15,020–53,970)|
|22||7.5 (7.0–8.0)||33,675 (17,560–64,570)|
|23||7.6 (7.2–8.1)||39,843 (20,573–77,164)|
|24||7.8 (7.3–8.2)||47,138 (24,067–93,325)|
* Predicted age from mean sac diameter is from Daya S, Wood S, Ward S, et al: Early pregnancy assessment with transvaginal ultrasound scanning. Can Med Assoc J 144:441, 1991. Reprinted with permission from the Canadian Medical Association.
† Predicted hCG from mean sac diameter is from Nyberg DA, Filly RA, Filho DLD, et al: Abnormal pregnancy; Early diagnosis by US and serum chorionic gonadotropin levels. Radiology 158:393, 1986. (The hCG was calibrated against the Second International Standard.) CI, confidence interval; hCG, human chorionic gonadotropin.
There is some controversy in the literature regarding the precise age at which sonography can first detect a chorionic (gestational) sac; estimates range from 3 to 5 weeks.38,42,44 There is less controversy regarding the size of the gestational sac when it is first observed.44 This is now thought to be approximately 2 to 3 mm MSD. Similarly, most observers agree that the MSD increases about 1 mm per day in early gestation.36,42,45,46 Controversy returns when one studies age estimates of MSD by various authors.38,42,43 Embryologic data and recent data gathered by de Crespigny, Cooper, and McKenna44 leave little doubt that the MSD equals 2 to 3 mm at 4 weeks and 3 to 4 days.41 It is reasonably safe to assume that a gestational sac reaches 5 mm at 5 weeks.43 Thus (until an MSD of 25 mm is reached), GA in days can be calculated by adding 30 to the MSD (i.e., MSD at 5 weeks or 35 days = 5 mm).36,46
The early embryo cannot be seen at this time, but there are two features that differentiate the gestational sac from the pseudogestational sac of an ectopic gestation. One is the double sac sign, which is created by visualization of the deep layer of the decidua parietalis and the early villi separated by the less echogenic layer and more superficial layer of decidua parietalis (Fig. 7-2).18,50 Also of value is the intradecidual sign, wherein the developing sac, imbedded in the decidua, deviates the endometrial cavity reflection (Fig. 7-3).49 Visualization of an embryo or embryonic structure is a more reliable sign than either of the features just mentioned. The earliest embryonic structure detectable by sonography is the yolk sac. This can be seen using high-resolution vaginal probes during the 5th menstrual week (MSDs usually range from 6 to 12 mm when a yolk sac is seen in the absence of a concomitant embryo) (Fig. 7-4).
FIGURE 7-2 The double decidual sac sign is a misnomer. This sonographic feature typical of early gestations simply should be called the double sac sign because of the origin of only one of the decidua. The origin of the inner of the double rings (the endometrial cavity [EC]) is less echogenic than the deeper layer (closer to the myometrium). It is the less echogenic superficial layer of the decidua vera (DV) that provides the “separation” between the more echogenic chorionic villi (C) and deeper decidua vera rings.
FIGURE 7-3 Transvaginal sonogram demonstrating the “intradecidual sign.” Note that the chorionic sac (CS) displaces the endometrial cavity (EC), indicating that the chorionic sac actually resides within the decidual vera and not within the endometrial cavity.
FIGURE 7-4 This high-resolution transvaginal real-time image demonstrates a yolk sac before the embryo is visible. Thus, a crown rump length cannot be measured at this stage of embryonic development. A-C. Instead, a mean diameter of the chorionic sac is determined, the “mean sac diameter.” Three measurements-length (A), depth (B), and width (C) are obtained, summed, and then divided by three. The length and depth are measured on the same longitudinally oriented image, whereas the width is measured on a transversely oriented image. Note as well that the measurements are obtained from the interface of the chorionic fluid with the rim of chorionic villi; the “wall” is not included.
The MSD becomes progressively less reliable for predicting menstrual age as the first trimester of pregnancy advances. Once the embryo can be visualized, the measurement of choice for estimation of menstrual age becomes the crown rump length (CRL). The developing embryo can be consistently detected with transvaginal transducers when the CRL reaches 5 mm and can be detected when it is as small as 2 mm. If the embryo can be visualized and measured, then MSD is no longer used to predict age. The embryo achieves a CRL of 5 mm when the MSD equals approximately 14 mm. MSDs of 14 mm or less are very precise for predicting menstrual age in normal pregnancies. The accuracy of MSD measurements only deteriorates after this time. As a general principle, embryonic or fetal measurements are more precise than measurements of the gestational sac. In addition, the earlier the measurement, the more accurate it will be. Therefore, MSD measurements between 2 and 14 mm (i.e., before the embryo can be seen) are highly reliable because they represent the earliest possible sonographic measurement.
By the 6th menstrual week, one can usually identify the early embryo and usually cardiac activity, as well, in normally developing pregnancies (Fig. 7-5A). However, better visualization of the early embryo can be made between 7 and 13 menstrual weeks (see Fig. 7-5B). Warren et al41 demonstrated the developmental landmarks of the embryo during this time frame (Fig. 7-6). Although these anatomic features can provide clues to the age of the fetus, better estimates of menstrual age can be made by measurement of the CRL.23,29,31,39,51–60 Strictly speaking, investigators and clinical sonologists do not truly measure the CRL when determining the length of first trimester embryos and fetuses.61 The early embryo/fetus is curved (see Fig. 7-5B). The conventional measurement that is obtained is the maximal straight-line length of the fetus (see Fig. 7-5B).61 The true CRL is depicted in Figure 7-5C. An additional measurement that can be considered is the maximum axial length of the fetus depicted in Figure 7-5D. However, despite the minor inaccuracy of nomenclature, the maximal straight-line length as measured by virtually all practicing sonologists will be referred to in this chapter as the CRL. When using the CRL to predict menstrual age, one should use the average CRL measurement from three satisfactory images. Modern instruments compute the menstrual age from this measurement. Alternatively, one may refer to published tables for estimation of menstrual age (Table 7-3).
FIGURE 7-5 A. This high resolution transvaginal real-time image demonstrates the early embryo (E), which measures approximately 3 mm in this case, corresponding to a menstrual age of approximately 6 weeks ±8% (3 days). The yolk sac (YS) is identified immediately adjacent to the embryo because there is no yolk stalk at this stage of embryonic development. B. This high-resolution transvaginal real-time image demonstrates the typical measurement of the crown rump length. This measurement is the maximum straightline length and not a true crown rump length. C. Cursor position to measure the true crown rump length. D. Method of measuring the maximal axial length.
FIGURE 7-6 Sequential appearance of morphologic features as the chorionic sac, yolk sac, embryo, and fetus develop in the first trimester of pregnancy. Although used by the authors, “fetal pole” is an inexact term. Embryo and fetus are preferred terms. Note that the chorionic (gestational) sac is identified in all normal pregnancies beginning at 5 menstrual weeks. The yolk sac is seen in all normal pregnancies from 6 menstrual weeks onward, and the embryo is seen in all normal pregnancies from 7 menstrual weeks onward. Although of some interest scientifically, the appearance times of the cerebral “ventricle” (vesicle), falx, and midgut herniation are not of value in precise pregnancy dating.
(From Warren WB, Timor-Tritsch I, Peisner DB, et al: Dating the early pregnancy by sequential appearance of embryonic structures. Am J Obstet Gynecol 161:747, 1989.)
In general, there has been extreme uniformity in the CRL data from various centers dating back to the original studies of Robinson51 and Robinson and Fleming,52 and it has been demonstrated that measurements made with static image equipment, transabdominal real-time equipment, and transvaginal real-time equipment demonstrate no significant differences. Additionally, Silva et al59 evaluated patients with known dates of conception using high-resolution transvaginal probes, and their data correspond closely with those of the early investigators. The only difficulty with the older studies is that they do not provide data before 7 weeks. This posed a problem when estimating the menstrual age of embryos detected earlier than 7 weeks. In a comprehensive study of CRL, Hadlock et al23 developed measurement tables establishing menstrual age for CRL as small as 2 mm and extended the range of CRL, data up to measurements as large as 12 cm (see Table 7-3). The majority of the early studies on CRL demonstrated that the accuracy of the method in predicting menstrual age was 3 to 5 days (± 2 standard deviations [SD]).51–53
In a study by MacGregor et al,56 however, the accuracy of the technique was demonstrated to decrease as pregnancy advanced into the late first trimester. Hadlock et al23 concurred that accuracy decreases as the first trimester of pregnancy nears its end. Because MacGregor’s study population had known dates of conception, this increase in variability with advancing pregnancy was believed to represent early biologic variability in embryonic or fetal size. In an effort to simplify the reporting of variability estimates, Hadlock et al evaluated variability as a percentage of the predicted age and demonstrated that the variability is relatively uniform at 8% for CRL measurements between 2 mm and 12 cm. Thus, for a CRL menstrual age prediction of 8 weeks, the 95% confidence interval is 8 weeks ± 8% = 8 weeks ± 0.64 week. Similarly, for a CRL age estimation of 15 weeks, the variability would be 15 weeks ± 8% = 15 weeks ± 1.2 weeks. The optimal time for prediction of menstrual age from CRL measurements is between 6 and 9 weeks.
Benson and Doubilet8 recommended the following rules of thumb for visual estimates of early first trimester menstrual age. If a chorionic (gestational) sac with no yolk sac or embryo is seen, estimate the age at 5 menstrual weeks. If a chorionic (gestational) sac with a yolk sac but no embryo is seen, estimate the age at 5.5 menstrual weeks. If a chorionic (gestational) sac with a tiny embryo (<5 mm) adjacent to the yolk sac is seen, estimate the age at 6 menstrual weeks. Although these are simple visual estimates, their accuracy is uncanny.
Other measurements of the fetus can also be made in the first trimester of pregnancy. For example, Bovicelli et al54 in 1981 evaluated the fetal BPD in comparison with the first trimester CRL for predicting menstrual age between 7 and 13 weeks, and in 1982 Selbing39 reported a similar study. Both groups demonstrated that the first trimester BPD is an accurate predictor of menstrual age but is not more accurate than the CRL and adds little, if anything, to the age estimate based on the CRL. Reece et al31 demonstrated similar results using early fetal “abdominal” (torso) circumference in the first trimester of pregnancy. Most recently, Sladkevicius et al62 compared 21 different CRL-based dating formulae to three different BPD-based dating formulae obtained at 12 to 14 weeks GA by abdominal ultrasound in 167 singleton pregnancies conceived following in vitro fertilization, and thus had definitively known pregnancy ages. Their findings showed that five of the CRL formulae generated very accurate pregnancy ages, and although the BPD formulae were also quite accurate, they had smaller random measurement errors than the CRL formulae. This led the authors to conclude that the BPD is a superior measurement than the CRL at 12 to 14 weeks for dating purposes. No measurements of the AC, HC, or FL were performed in this study, and thus, comments cannot be made about them in comparison to the CRL. In practice, the use of high-resolution vaginal probes allows very acceptable images of the head, abdomen, and femur for measurements of HC, BPD, AC, and FL from 10 weeks on. However, because these additional measurements are not necessarily more accurate than the CRL length in predicting age from 10 to 13 weeks and their use in conjunction with the CRL does not further improve age estimation, it is difficult to justify their use. Moreover, they are technically more difficult to obtain than the CRL measurement, and at least thus far their routine use is not warranted in the first trimester. The transition between first and second trimesters (13 to 14 weeks) is the appropriate time to make the transition from CRL to BPD, HC, AC, and FL.
In summary, the accuracy of first trimester fetal measurements in predicting menstrual age is well documented; there is very little biologic size variability during this time. This is in contrast to the third trimester of pregnancy, in which individual genetic expressions in fetal size can result in a very heterogeneous population. It is also well established that once menstrual age has been determined or corroborated by very early MSD (2 to 14 mm) or embryonic or fetal CRL in the first trimester of pregnancy, the menstrual age of the pregnancy is established and should never be changed based on biometric measurements made later in pregnancy.
Indeed, the same can be said of sonographic estimation of age at any time. An earlier measurement supersedes a later measurement, and sonographic estimates of age before 20 weeks gestation are highly reliable for pregnancy dating. In nearly all cases, sonographic estimates of age before 20 weeks gestation (except those with assisted conception and those maintaining basal body temperature charts) represent the most accurate scientific information establishing the menstrual age.
An important issue is the timing of the sonogram when the dates are uncertain. It has already been stated that the earlier the measurement is obtained, the more accurate is the estimate because biologic variability increases throughout gestation. Unfortunately, an age estimate at 7 menstrual weeks will be extremely accurate but provides little other useful information about the fetus, placenta, amniotic fluid volume, and cervix. An alternative is to wait until the second trimester is under way before performing the sonogram to establish dates (15 to 18 weeks). Because dates are in question, the uterine size estimate is important in planning the timing of the early second trimester examination. If this second option is elected, dating accuracy will suffer. However, the accuracy at 15 to 18 weeks is still excellent and adequate for all clinical purposes. The value of waiting is that a large amount of additional information about the health of the pregnancy is obtained. Of course, there is also the possibility of obtaining both first trimester and second trimester sonograms, but this is associated with a significant increase in cost. The greatest accuracy per medical dollar spent is achieved in the first trimester. The greatest value per medical dollar spent is achieved in the early second trimester. If the dates of a pregnancy are clinically truly uncertain, the earliest possible ultrasound is advised to ensure the highest accuracy in assigning GA. This information may ultimately prove to essential for the long-term management of the pregnancy, and therefore worth the additional expenditure up front.
In the second trimester of pregnancy, the fetus has grown sufficiently in size so that remarkable anatomic detail can be visualized (see Chapter 9). Many structures can be identified and measured during this time,22,23,33,63,64 but the basic fetal measurements used to estimate age are the BPD, HC, AC, and FL.1 With training and practice, these measurements can be obtained with a high degree of consistency and accuracy. One must always remember, however, that the images and measurements must be made with great care in every case, and one must be certain to duplicate the technique of the investigator whose data one is using. Using measurements from poor images or images that depict fetal anomalies should be avoided.
Modern instrumentation computes the age estimate virtually in real time on the viewing screen because the cursors are fit to the structure being measured. The examiner should avoid the temptation to “massage” the endpoints of the measurement because the age computation on the viewing screen is smaller or larger than their preconceived notion of what the biometry should be. Instead, it is crucial to learn the rules for obtaining each measurement and to adhere to them rigorously; only then can the results be interpreted in a rational way.
Crucial decisions may be made on the basis of fetal biometric measurements. Therefore, each clinical specialist must be well schooled in the proper techniques of obtaining these measurements. It is not necessary to repeat every erroneously taken measurement because correcting the measurement probably will not influence the interpretation of the case if the error is small. However, it is important that the interpreter can recognize the error and has a sense of both the magnitude and the direction of the error. For example, suppose that a BPD was erroneously measured in a fetus of 21 weeks menstrual age, known by early embryonic CRL. The erroneous measurement is 4.7 cm and the predicted age by BPD is, therefore, 20.2 weeks. Analysis of the error leads one to recognize that the BPD was slightly undermeasured. One can recognize that a BPD predicting 20.2 weeks is within the normal range. Correcting the error would push the measurement toward the known menstrual age. Clearly, there is no reason to spend the time correcting the error. Note that this comment is only applicable for the occasional slight error in measurement and is not appropriate to systematic or large-scale errors that should be rectified expeditiously.
Potential errors in biometric measurements are numerous and can be engendered by equipment misregistration, aberrations of transducers, and type of transducer to name a few. It is important to ensure through the equipment manufacturer that the instrument is properly adjusted to make accurate linear measurements in every direction on the viewing screen, that biometric tables are accurately entered in the machine computation package, and finally, that subsequent computations are accurate. Although one does not have direct control over these adjustments, the wise sonologist never assumes that instrumentation is always accurate in this arena. Aside from optimizing the image, one can only control two aspects of biometric measurement. One can choose the plane of section, and one can choose the endpoints of measurement (Figs. 7-7 to 7-12).
FIGURE 7-7 An accurate biparietal diameter can be obtained through any plane of section that intersects the third ventricle (TV) and thalami (T). The margins of the calvaria (C) must be symmetric. The first criterion ensures that the plane of section is taken at the proper craniocaudal plane. The second criterion ensures that the transducer is oriented perpendicular to the central axis of the head. Note the measurement is 74.6 mm.
FIGURE 7-8 Image of same fetus as in Figure 7-7 taken minutes apart. This biparietal diameter measurement meets the criterion of being taken through the third ventricle (TV) and thalami (T) but does not meet the criterion of demonstrating symmetric calvaria (C). The BPD measured from this plane (72.7 mm) is 2 mm smaller than the properly measured BPD of 74.6 mm (see Fig. 7-7). This degree of error is probably not clinically significant.
FIGURE 7-9 Image of same fetus as in Figure 7-7 taken minutes apart. As with Figure 7-7, this plane of section fulfills the two criteria for measuring a biparietal diameter (BPD) (T, thalamic nucleus; TV, third ventricle). A. However, it also fulfills the third criterion necessary to measure a head circumference; the plane is properly oriented to the skull base. Demonstrating the cavum sepli pellucidi (CSP) anteriorly and the tentorial hiatus posteriorly (Tm, tentorium) documents proper orientation of the plane of section to the skull base. Because this plane of section accurately meets all criteria for an HC measurement and because the ellipse is carefully fit to the calvarial margins, one may correctly assume that 263.6 mm is highly accurate. B. Note that the BPD measurement is 74.2 mm, only a 0.4-mm difference from that of Figure 7-7, demonstrating that any plane through the thalami and third ventricle with symmetric calvaria accurately measures the BPD.
FIGURE 7-10 Image of same fetus as in Figure 7-7 taken minutes apart. A. One reason for the development of the long linear transducer was to encompass the entire fetal calvaria in late pregnancy to measure both the biparietal and an occipitofrontal diameters necessary to compute a head circumference. Later, as measurement technology advanced, this requirement was also deemed necessary so that the computer-generated ellipse could be seen to “fit” the calvarial margin throughout. An error, however, was made during this measurement. The sonographer fit the ellipse to the scalp margin rather than the calvarial margin (C). B, With shorter linear transducers, the ellipse approximated a portion of the calvarial margin. This does not create a significant error. There is no need to purchase a long linear transducer solely for the purpose of demonstrating the entire perimeter of a third trimester fetal head.
FIGURE 7-11 This is the same image as Figure 7-7 now used to measure the head circumference (HC). The plane of section meets all necessary criteria for biparietal diameter measurement but is not properly oriented to the skull base for HC measurement. Note that the cerebellum (Ce) rather than the tentorial hiatus is demonstrated posteriorly. The HC measured from this plane of section is 256.9 mm compared with 263.6 mm measured from the proper plane (compare with Fig. 7-9A).
FIGURE 7-12 Image of same fetus as in Figure 7-7 taken minutes apart, A. This biparietal diameter (BPD) measurement fulfills the criterion of symmetric calvaria but was not obtained through the thalami and third ventricles. It was obtained too near the vertex. The three “lines” seen in this image have been commonly displayed in the ultrasound literature. The central line is the interhemispheric fissure (IH). The lateral lines are composite reflections from the deep medullary veins (DMV) draining the periventricular white matter. Therefore, this plane is obtained cephalad to the lateral ventricular bodies, BPD measurement in this plane underestimates age. Compare this measurement of 71.9 mm with that obtained in the correct plane as illustrated in Figure 7-7 (74.6 mm). B. The head circumference (HC) measured in this plane also underestimates age. Compare the measurement of 250.7 mm with the correctly measured HC of 263.6 mm obtained in this fetus (compare with Fig. 7-9A).
The BPD can be appropriately measured through any plane of section that traverses the third ventricle and thalami (see Fig. 7-7).65 Early in the history of BPD measurement, the BPD was defined as the widest distance between the parietal eminences. The “widest” BPD is not always obtained though a plane of section that traverses the third ventricle and thalami. Occasionally, the “widest” BPD is obtained in a more cephalad location.65 However, rather than dutifully searching for the widest diameter, it is now convention to measure all fetuses at the same anatomic plane in all centers across the country. The advantages of every practitioner obtaining the BPD in the same way are obvious. Searching for the widest BPD is likely to engender errors because there will be a tendency to accept an erroneous measurement wherein the error overestimates the measurement.
Any plane of section through a 360-degree arc that passes through the thalami and third ventricle is acceptable for measuring the BPD. This means that an accurate measurement can be obtained through an infinite number of planes (the significance of this point is made clear later). The rules for measuring the BPD are as follows. First, the correct plane of section is through the third ventricle and thalami, as noted. Second, the calvaria are smooth and symmetric bilaterally. Third, the cursors are consistently positioned in one of the three following ways: outer edge of near calvarial wall to inner edge of far calvarial wall, inner edge of near calvarial wall to outer edge of far calvarial wall, or middle of near calvarial wall to middle of far calvarial wall. Measuring from the outer edge of the near calvarial wall to the outer edge of the far calvarial wall is inappropriate. The first two criteria define the precise plane of section. Because there are two criteria that must be met to obtain a proper plane of section, the BPD measurement should be considered as a “two-dimensional” measurement. The third criterion describes the proper endpoints of measurement. Because the calvaria are brightly echogenic and symmetric and the thalami are symmetric about the third ventricle, the BPD is a measurement that can be obtained with great consistency and accuracy. The symmetry makes it easy to see that the correct plane of section has been obtained. Common errors in BPD are illustrated in Figures 7-8 and 7-12A.
Recall that the BPD can be obtained through an infinite number of planes. However, in the strictest sense, the HC is best obtained through a single plane of section. Thus, the HC is a more difficult measurement to obtain consistently correctly. The correct plane of section parallels the base of the skull. Therefore, the plane is more cephalad anteriorly than it is posteriorly. Whereas when obtaining a properly measured BPD the transducer must be correctly oriented in two planes, the HC requires that the transducer be properly oriented in three planes. Because there are three criteria that must be met to obtain a proper plane of section, the HC measurement should be considered as a three-dimensional measurement. To measure the BPD the transducer must be (1) perpendicular to the parietal bones and (2) positioned at the correct cephalocaudal position to intersect the third ventricle and thalami. To measure the HC, the two criteria for the BPD must be fulfilled, and the transducer plane must be properly oriented to the skull base. Because the skull base is very irregularly shaped, it is not used to orient the plane of the HC. Instead brain anatomy landmarks are used.
The rules for measuring the HC are as follows. The correct plane of section is through the third ventricle and thalami in the central portion of the brain (as with the BPD), but the cavum septi pellucidi must be visible in the anterior portion of the brain and the tentorial hiatus must be visible in the posterior portion of the brain. Some refer to this anatomy as an “arrow.” The cavum septi pellucidi and frontal horns are the “feathers.” The third ventricle and sylvian aqueduct are the “shaft.” The ambient and quadrigeminal cisterns and the tentorial hiatus are the “arrowhead.” The calvaria must be smooth and symmetric bilaterally. After the proper plane of section is obtained, the cursors are positioned at the outer edge of the near calvarial wall and the outer edge of the far calvarial wall. Most modern equipment then will allow a computer-generated ellipse to be fit to the calvarial margins (see Fig. 7-9A). The entire perimeter of the calvaria need not be demonstrated to accurately measure the HC.66 The ellipse adequately estimates the head perimeter even when it is not entirely imaged (see Fig. 7-10).
Note several features. First, a properly measured BPD can be obtained on the same image as a properly oriented HC measurement (see Fig. 7-9B). The reverse is not necessarily true (see Figs. 7-11 and 7-12). Second, the proper position of the cursors to measure the HC is inappropriate for BPD measurements (compare Fig. 7-9A and B). Thus, if the BPD is measured first on the image, the cursors must then be moved before the computer-generated ellipse is fit to the calvarial margins for HC determination. Finally, it is important to be certain that the ellipse is fit to the calvaria and not the skin of the scalp (see Fig. 7-10A). Again, an analysis of an observed error should give the interpreter a sense of both the direction and magnitude of the error. If, for example, the sonographer fit the ellipse to the skin of the scalp and not to the calvaria, the error will increase the HC measurement (i.e., the “direction” of the error), and in a late, large fetus the error will be great (i.e., the “magnitude” of the error). In a young fetus the magnitude of the error will be small.
The FL measurement is technically the easiest of the common biometric measurements. This is due to the essentially one-dimensional nature of the measurement. The transducer need only be aligned to the long axis of the bone to obtain a proper plane of section. No other transducer adjustment is necessary. However, that does not mean that errors in measurement or observational errors are rare.10,67,68 Quite the contrary is true.
The first feature to understand about the FL measurement is that one does not actually measure the entire femur. Only the ossified portions of the diaphysis and metaphysis are measured (Fig. 7-13).10 The cartilaginous ends of the femur are excluded. The ossified portion of the femur is more visible sonographically than the nonossified ends. Nonetheless, the cartilaginous ends of the femur are readily demonstrated (Fig. 7-14). Although the cartilaginous ends are excluded, they are the keys to accuracy and consistency in femoral length measurements. To obtain the measurement accurately, the transducer must be aligned to the long axis of the diaphysis. How does one know that the transducer is properly aligned? One method is to accept the longest femoral length as the most accurate measurement. This assumes that the only potential error is to undermeasure the bone because the transducer is improperly aligned to the bone. This is a false assumption.
FIGURE 7-13 Radiograph of the femur from a term neonatal autopsy specimen. Sonographic measurements of the “femur” include only the ossified portion of the diaphysis (D) and metaphysis (M). The cartilaginous femoral head (FH), greater trochanter (GT), and distal epiphysis (DE) are not included in the measurement. When it ossifies, the distal femural epiphyseal (DFE) secondary ossification center is also not included in the measurement.
FIGURE 7-14 A. Sonogram of the proximal fetal femur. Compare with Figure 7-13 The cartilaginous greater trochanter (GT) and femoral head (FH) are not included in the femoral measurement but are clearly seen, B. Sonogram of the distal femur (compare with Fig. 7-13). The cartilaginous distal epiphysis (DE) is also clearly seen but is not included in the femoral measurement. Although not included in the measurements, the cartilaginous ends of the femur are the keys to obtaining highly accurate and reproducible femoral measurements. See text.
Proper alignment of the transducer to the long axis of the femur is ensured by demonstrating that both the femoral head or the greater trochanter and the femoral condyle are simultaneously in the plane of section (Fig. 7-15).10 If these two structures are seen, the plane is unambiguously through the femoral long axis. Thus, once can be confident that the proper plane of section to obtain an accurate measurement has been chosen. The only remaining task is to position the electronic cursors properly at the correct endpoints of measurement. Again, it is the cartilaginous ends that direct precise placement of the measurement cursors. The cursors are positioned at the junction of the bone with the cartilage (see Fig. 7-15). It is important not simply to choose the end of the brightest reflection as the endpoint of measurement because of an entity that has been called the “distal femoral point.” The distal femoral point is not part of the bony metaphysis.10 Including it significantly overmeasures the femur. Therefore, the rules for measurement of the femur are as follows. First, align the transducer to the femur and freeze the plane that shows both the cartilaginous femoral head and distal condyle. Then place the measurement cursors at the junction of the cartilage and bone, being careful to avoid the distal femoral point.
FIGURE 7-15 This sonogram demonstrates the femur of a middle third trimester fetus. This is an ideal image from which to measure the femur. Because the greater trochanter (GT) and lateral condyle (LC) of the distal epiphysis are seen, one may be confident that the proper plane through the long axis of the femoral diaphysis has been obtained. The only remaining task is to position the endpoints of measurement properly (i.e., the electronic cursors). The proximal cursor is easily placed at the junction of bone and cartilage (short arrow). However, a choice must be made distally. Should the “point” (long arrow) along the lateral margin be included? Compare with Figure 7-13. No such ossified “point” exists. This structure, only seen sonographically, should not be included in the measurement and introduces a significant overestimate of femur size if included. The only way to choose properly the correct endpoint for placing the distal measurement cursor is to demonstrate clearly the junction of the distal epiphysis with the metaphysis (short arrow). As well, this image clearly depicts the ossification center (DFE) of the distal epiphysis.
Finally, the AC is the most difficult of the four measurements that are ordinarily obtained. As with the HC, the AC is a three-dimensional measurement. Unfortunately, the abdominal anatomy is not symmetric like the brain anatomy. Further, there is no bright calvarial margin to check the perpendicularity of the planar axis and to provide easily seen endpoints of measurement.
The AC is measured in a location that estimates liver size. The liver is the largest organ in the fetal torso, and its size reflects aberrations of growth, both growth restriction and macrosomia. Therefore, one does not measure the circumference at the fetal “waist” (umbilicus) as one would in an adult. The fetal AC is measured at the position where the transverse-diameter of the liver is the greatest. This can be determined sonographically as the position where the right and left portal veins are continuous with one another (Fig. 7-16).69 Some refer to this anatomic confluence of the intrahepatic portal veins as “the hockey stick.”
FIGURE 7-16 Umbilical venous circulation through the fetal liver. A. Plane of section depicting the umbilical vein (UV) in short axis (correlate with Fig. 7-17A). This plane is too caudal for abdominal circumference measurement. B. Plane of section through the junction of the left (LPV) and right (RPV) portal veins (correlate with Fig. 7-17B). This is the correct level for AC measurement (DV, ductus venosus). C. Plane of section aligned along the course of the LPV (correlate with Fig. 7-17C). Note that this plane is too inclined in a craniocaudal axis.
(Illustration by James A. Cooper, MD, San Diego, CA.)
Therefore, the rules of measurement of the fetal AC are as follows. The correct cephalocaudal plane is the position where the right and left portal veins are continuous with one another. Second, the appearance of the lower ribs is symmetric. Finally, the shortest length of the umbilical segment of the left portal vein is depicted. If a long segment is seen, then the transducer is erroneously angled inferiorly instead of perpendicular to the midline (see Fig. 7-16C).69 After this plane of section is frozen on the screen, the ellipse is fit to the skin edge. Note that this is distinctly different from the HC, where one specifically does not fit the ellipse to the skin edge but rather to the calvarial edge. Importantly, the skin margin may abut other soft tissue structures like the placenta or myometrium and be relatively inconspicuous. Remember that the rib margin is easily seen, and one may mistakenly fit the ellipse margin to the rib instead of the skin. This will significantly undermeasure the AC. Although this error will have relatively little effect on the estimate of menstrual age, it will have a much greater effect on the weight estimate.
There will be many times when the landmarks for fetal AC measurement are less than optimally documented. In this circumstance, the reader should rely on the following rule: “round” covers a multitude of sins. This rule implies that, when the desired anatomic landmarks are difficult to demonstrate, the circumference estimate, wherein the transverse and anteroposterior diameters of the abdomen are equal or nearly so, is likely to be the more accurate compared with estimates wherein the diameters are disparate. In compliance with this concept, excessive pressure with the transducer should be avoided because it distorts the shape of the abdomen.
The BPD has received the greatest amount of attention in the literature as a means of establishing menstrual age.12,22,23,28,33,63,64,70 The measurement technique and pitfalls were described earlier. The menstrual age can be determined by using a standard reference table (Table 7-4). However, modern instruments immediately compute an age as the measurement is being obtained.
All reports on the BPD have demonstrated it to be an accurate predictor of menstrual age before 20 weeks. For example, Hadlock, Harrist, and Martinez-Poyer22 demonstrated the variability to be ±1 week (2 SD) in a population of 1771 patients with optional menstrual histories seen between 14 and 20 weeks (Table 7-5). Persson and Weldner64 and Rossavik and Fishburne33 reported similar results in patients with known dates of conception, as did Crespigny and Speirs70 in a large series of patients whose dates were confirmed by CRL in the first trimester of pregnancy. In addition, Campbell et al5 and Waldenström, Axelsson, and Nilsson,6 in independent studies, demonstrated that a BPD obtained between 14 and 20 menstrual weeks is a better predictor of the estimated date of confinement than an optimal menstrual history. Virtually all studies demonstrated a progressive increase in variability from 20 weeks to term, but the degree to which the variability increases in the late third trimester of pregnancy has been a subject of some disagreement in the literature.12,28,70,71 Most early authors concluded that the variability during this time frame is approximately ± 3½ weeks (2 SD), but Kurtz et al28 reported the variability to be ± 2 weeks during this time. This difference most likely is a statistical phenomenon, because the mathematical evaluation by Kurtz et al was performed on mean values from a number of different centers and does not directly include any of the raw data from the more than 25,000 patients who form the basis of their report. The variability they reported, then, represents the confidence interval of the mean and should not be comparable to the standard deviation reported by others.12 In studies of patients with optimal menstrual histories, the variability of late third trimester BPD age predictions has been consistently demonstrated to be approximately ± 3½ weeks. This has been confirmed in a large series of patients in Australia in whom the menstrual ages were confirmed in early pregnancy by CRL.70 Both of the later studies, however, eliminated patients with head shape abnormalities, multiple gestation, or diseases likely to adversely affect fetal growth. The large variability associated with third trimester use of the BPD has been confirmed in a study by Benson and Doubilet7 of patients whose menstrual histories had been established early in pregnancy by CRL; in this study, however, no attempt was made to eliminate multiple gestations or patients with potential growth disturbances. These authors found that the variability in predicting menstrual age using BPD reached a peak of approximately 4.1 weeks (2 SD) in the late third trimester of pregnancy (Table 7-6).
In certain circumstances (e.g., ruptured membranes, breech presentations, multiple gestations), shape changes in the fetal head may lead to even greater errors than those mentioned here. If one suspects that the shape of the calvarium is other than typical, one should measure the cephalic index of the head to assess head shape (see Fig. 7-5). The cephalic index is calculated from the BPD and the fronto-occipital diameter (FOD) measured from the outer edge of the calvaria to the outer edge of the calvaria:
Technically, manufacturers usually compute a cephalic index from the HC ellipse long and short axes. Because the short axis of the HC ellipse is not truly a BPD, a slight and clinically insignificant error is introduced into cephalic indexes computed in this way. In the first in utero study on this subject, Hadlock et al11 evaluated 316 patients between 14 and 40 menstrual weeks and demonstrated that the cephalic index was essentially age independent over time, with a mean value of 78.3 and an SD of 4.4. They concluded that a cephalic index greater than 1 SD above or below the mean (<74, >83) may be associated with a significant alteration in the BPD measurement expected for a given menstrual age and that the HC can be used effectively as an alternative means of establishing age in such cases. Gray et al72