Chapter 5: FETAL BIOMETRY

There is likely no more important piece of information used to manage pregnancies than correct gestational age dating. Gestational age allows the practitioner to recommend the correct age-appropriate serum screening, diabetic screening, administration of anti-D immunoglobulin, group B streptococcal testing, and timing of delivery. The use of the patient's last menstrual period (LMP) to establish dating is fraught with imprecision, particularly if the woman has had irregular ovulation, such as from the use of hormonal contraception, lactation, or oligoovulation. A woman's memory of her LMP may also be imprecise. Consequently, the term certain LMP is valid if specific criteria are present (Table 5-1). Even with a certain LMP, the gestational age is more accurately established with sonographic parameters in the first or early second trimester. Careful and meticulous tables that correlate biometric measurements with gestational age are useful.

The literature contains many tables and nomograms that describe the normal growth of various fetal parameters, such as the biparietal diameter, abdominal diameters, femur, and orbits. Some of these tables were established with great care and respect as to basic mathematical principles. Others, however, were prepared in a less careful manner. Deciding which table to use and knowing its limitations are important in everyday practice. For example, using a table whose "confidence limits" were graphically drawn instead of mathematically computed (or worse, a table that does not provide confidence limits) may be imprecise. The basic principles involved are well established, and the software required for analysis is currently available for most microcomputers. Knowing the basic concepts described in this chapter enables an understanding of the seemingly complicated and esoteric language used in sonographic literature, and thereby allows a judgment as to whether or not a study has been performed correctly.

Before the development of ultrasound, fetal dimensions were measured using radiological techniques. The development of ultrasound made it possible to measure the bones and soft tissue structures of the fetus faster and more reliably than with x-rays. Fetal growth is so rapid that parameters such as the fetal biparietal diameter (BPD) and femur length (FL) change significantly within 1 to 2 weeks. The use of such measurements answers the following questions:

1. What is the age of the fetus?

2. Is the fetus of appropriate size for its age?

3. Are there any malformations?

The first question is crucial in the modern practice of obstetrics and is one of the most frequent reasons for referral in countries where routine scanning in not the practice.

The evaluation of fetal growth and the detection of intrauterine growth restriction are also major concerns because the growth-restricted fetus is at a higher risk for morbidity and mortality.

With the decrease in the traditional causes of perinatal mortality, the detection of congenital malformation has become more important. The presence of fetal anomalies may affect the fetal measurements; also, in the course of obtaining fetal biometric measures, a fetal organ survey should be undertaken. The American Institute of Ultrasound (AIUM) has published guidelines describing these organs (Table 5-2).

How Are Normal Values Derived?

The normal values (mean, lower, and upper limits of normal) of a given parameter at different gestational ages are defined by measuring that parameter in fetuses of normal patients with well-established gestational age. Data collection for this purpose can be either cross-sectional or longitudinal. The difference between the two is explained below.

Selection of Patients

The patient population with the best-known dates are those who have undergone in vitro fertilization, with a gestational age of ±1 day. Those women who undergo ovulation induction and intrauterine insemination likewise have a small error of gestation age (±3 days). Gestational age is established with certainty if a record of the basal body temperature has been kept for the conceptional cycle or if the date of conception is known. Another acceptable cohort would be the combination of a regular menstrual history, a well-defined LMP, and confirmation with early ultrasound (<14 weeks gestation). Patients with doubtful dates should not be included in a study of fetal biometry. Besides criteria that may impact the accuracy of the gestational age, patients should not have had a history of medical, surgical, or obstetric complications. Optimally, infants should have been delivered between 38 and 42 weeks with no evidence of growth retardation or congenital anomalies.

Types of Studies: Cross-Sectional or Longitudinal?

Fetal growth can be analyzed by 2 different kinds of studies: cross-sectional or longitudinal. In a cross-sectional study, many features are examined only once during gestation. In a longitudinal study, a small number of fetuses are investigated serially.

Cross-sectional studies offer some important advantages: (1) they can be performed in a relatively short period of time, (2) the data are easier to collect because the patients are scanned only once, and (3) statistical analysis is easier. They also have important drawbacks: (1) they fail to characterize individual growth (they describe "population" growth); (2) the stability of the statistical analysis that can be performed on cross-sectional data is not always as good as that from longitudinal studies; (3) they are extremely susceptible to inclusion of fetuses with abnormal growth patterns, poorly established gestation age, or both; and (4) the follow-up of newborns is complicated by the large number of cases, which often represent the extreme values of the curve: the early and late gestation.

A very common error in the literature is failure to respect the principle that each fetus should contribute only 1 datum point to the study. Otherwise, difficult variance assumptions are violated and the accuracy of the confidence limits is questionable. Nonetheless, one often finds statements in articles such as "150 patients were scanned for a total of 215 examinations."

Longitudinal studies have several advantages: (1) the gestational age has to be defined in fewer patients, (2) age is established only in early pregnancy, (3) abnormal growth curves are easily recognized, and (4) the statistical analysis allows stronger curve fitting and computation of velocity curves and so forth. Unfortunately, by their very design, these studies require that the same fetus be scanned throughout gestation. This considerably increases the time needed to collect data and calls for high motivation on the part of the mother. There are many mathematical, biological, and epidemiological arguments to favor results obtained by longitudinal studies.

Prediction from Equations

The purpose of fetal biometry is to predict information concerning a fetus and then verify how closely that fetus conforms to the predictions. The reference parameter represented on the x-axis of graphs, or at the extreme left column of nomograms (eg, the gestational age), is called the observed value. The y-axis value is called the predicted value. A common mistake consists of using the predicted value to obtain the observed value. For example, the BPD is measured and read from right to left to the gestational age column to obtain the age. This "two-directional" reading is mathematically incorrect. Among other reasons, the confidence limits will be expressed in the wrong scale (or be impossible to find). It appears a bit confusing that, when data are collected, the relation between the observed variable and the predicted variable is different from that between the predicted variable and the observed variable! One tends to assume that if a BPD of 45 mm corresponds to a gestational age of 19 weeks, then for 19 weeks the mean BPD would be 45 mm. The prediction is, in fact, 46 mm (Table 5-3) because of the sum-of-squares technique described above.

How to Compute the Confidence Limits

Besides being able to describe the mean growth, one must be able to answer the question, how far from the mean can an individual be and still be considered normal? Or, in statistical terms, what is the dispersion around the mean? The statistical parameter that measures this dispersion is the standard deviation. The smaller the standard deviation, the less the variability of the sample around the mean. The standard deviation is also used to define the statistical limits of normality. These intervals are called confidence limits (Figures 5-1 and 5-2). Traditionally, the confidence limits are set at the 5th and 95th percentiles (±1.66 standard deviations) or at the 1st and 99th percentiles (±2.38 standard deviations). Two standard deviations correspond to 95% of the population, 2.5% being below the normal limits and 2.5% being above. Outside the lowest and highest percentiles, the parameter is considered abnormal. With the commonly used 5th and 95th percentiles, it is important to remember that 10% of patients tested will be outside normal limits. This does not mean, however, that the patient is abnormal. The predictive value of a given equation is greater if the standard deviation does not change significantly with variations of the observed value. The absence of variation of the standard deviation can be investigated by the Bartlett test or Levene test.

Study methodology and statistical analysis may affect the reliability of the biometric information, and the best mathematical model for biometry–gestational age correlation. The decision of which equation to use for one's ultrasound practice is an important one; one should choose an equation derived from a study based on a large sample of data covering a wide range of values for the independent variable. It should adequately represent the extremes of the curve. If there are many equations that fulfill these criteria, select the curve based on longitudinal rather than on cross-sectional studies. Be careful to check that the study spans the range of gestational age being covered. For example, some studies describe the growth of the femur from 12 to 22 weeks; using these studies outside those limits would be inaccurate.

Do not use the same table to determine gestational age from a parameter and to test the normality of this parameter against the gestational age.

The equations that follow are those of Jeanty and associates, except when specified. The rationale for selection of the curve is expressed above. It is beyond the scope of this chapter to discuss all the curves that have been published. The guidelines should help the reader to decide which curves are correct. "Composite" curves, which average excellent curves with less optimal ones, should be avoided.

In all nomograms presented, the observed value is listed in the left column and the predicted value in the right column. A special effort toward standardization has been made. All measurements are expressed in the same units: millimeters, weeks plus days from the first day of the LMP, and grams.

Definition

Different ways of estimating the duration of pregnancy can be used. The conceptual age is calculated from the first day of the LMP. Gestational age, which is calculated from theoretical time of ovulation, plus 2 weeks, is generally the accepted parameter. This has the advantage over menstrual age of eliminating the problems associated with oligomenorrhea and delayed ovulation.

The weeks are always counted as completed weeks, not as current weeks. A patient whose LMP started on January 1 therefore will be in her fourth week on February 1.

Parameters Proposed for Gestational Age Assessment

In all the tables in this section, gestational age is expressed in weeks and days from the (theoretical) onset of the normal menstrual period (conception date minus 15 days). Tenths of weeks are not used to avoid difficulties in conversion to days. All measurements are expressed in millimeters instead of centimeters to be consistent with the International System of Units in which secondary units are related to the primary unit by 10 raised to the power of 3 (thousand, million, billion, thousandth, micro, pico, and so forth) and in which the use of centimeters is not recommended.¹

The parameters proposed to establish gestational age are detailed below.

Gestational Sac

Prior to an embryo or yolk sac being seen, an echolucency surrounded by an echogenic ring can be seen on sonography, typically endovaginal probe. Three measurements in orthogonal planes are utilized and a mean diameter is calculated. This mean sac diameter (MSD) is then compared to gestational age tables. The MSD is fairly precise up to 14 mm, but once the embryo can be measured (5 mm), a crown-rump length (CRL) is more accurate. A gestational sac must be differentiated from fluid within the endometrial cavity, the so-called "pseudogestational sac" that may be present with ectopic pregnancies. Table 5-4 compares the characteristics of a true gestational sac versus the pseudosac.

Crown-Rump Length

The CRL (Table A5-1) is the longest demonstrable length of the embryo or fetus, excluding the limbs and yolk sac.² The reason for the high accuracy of the CRL is the excellent correlation between length and age in early pregnancy when growth is rapid and minimally affected by pathological disorders. Even if differences do occur during that time, they are too small to be detected by ultrasound. Although originally described with static scanners, CRL measurement is much faster and just as reliable when obtained in real time. In a scan that demonstrates a longitudinal section of the fetus, which ideally should be the midline including the fetal spine, the calipers should be placed at the outer edge of the cephalic pole and the outer edge of the fetal rump (Figure 5-3). Because the embryo can display flexion or extension, at least 3 measurements should be obtained and averaged. The limbs and yolk sac should not be included. It is predictive of menstrual age with an error of 3 days (90% confidence limits) from 7 to 10 weeks.

As growth progresses, however, the curvature of the fetus changes and the linear measurements that can be obtained with the calipers are less accurate. The error increases to 5 days between 10 and 14 weeks of gestation. After 14 weeks gestation, multiple biometric parameters should be used.

Biparietal Diameter

Historically, BPD (Table A5-2) was the first parameter used to assess gestational age. Its accuracy is greatest between 12 and 28 weeks. The consensus that has been reached is to measure the BPD at the level of the thalami (Figure 5-4). Measurement rostral to the thalami (below the cerebral peduncles) may result in the underestimation of the BPD and, consequently, of the gestational age. Usually, the BPD is the widest transverse diameter of the fetal skull; however, the transthalamic landmark is more accurate. Biparietal diameter charts are based on measurement from the outer table of the proximal skull to the inner table of the distal skull,^3,4 corresponding to the "leading edge to leading edge." Although the BPD may change with the gestation age (and move slightly rostrally), using the same landmark throughout the pregnancy is preferred to avoid unnecessary confusion. It is ill advised to change the plane in late pregnancy because the reference charts were established with the fixed plane described above.

The fetal head can occasionally be flattened and elongated (dolichocephaly) and the BPD thereby artificially decreased (Figure 5-5). To check for this, the cephalic index (CI) should be obtained. The CI is the ratio of the BPD divided by occipitofrontal diameter (OFD), and its normal range is 0.75 to 0.85. When the CI is close to or beyond either end of the confidence limits, the BPD should not be used to assess gestational age. In general, the head circumference (HC) is more accurate than the BPD when the fetal head shape is abnormal.

Doubilet and Greenes correct the BPD when it is deformed.⁵ Their procedure is different from the CI. What they have attempted to do is to provide the normal measurement of BPD if the head had not been deformed. They use one of two approaches: either calculate the area of the head and derive the BPD from it or use a circumference correcting the BPD. Although they recommended using the area corrected by BPD, we have been uneasy with that approach because the cross-sectional area of a volume that is compressed is not constant. For example, if you compress a coffee cup between your fingers, the distance between your fingers will decrease, and the cross-sectional area of the cup remains the same. The same is true of the fetal head. If you press the head side to side, the BPD will decrease, the OFD will increase, and the head area will decrease, but the head perimeter will not change. For these reasons we do not recommend using the error-corrected BPD. If either of these methods is used, the perimeter-corrected BPD would probably be of more value. However, in practice, that would increase the number of calculations needed and would not offer significant advantage over the head perimeter method described below.

Wolfson et al showed that, in fetuses that have premature rupture of the membranes, the BPD is not very reliable in assessing gestational age.⁶ This is no surprise in view of the discussion on dolichocephaly. In fetuses that present for the first time with premature rupture of the membrane, measurement of the femur and humerus should be obtained. An average of these 2 measurements usually provides the best available assessment.

O'Keeffe et al conducted a study similar to that of Wolfson et al. O'Keeffe et al arrived at the same conclusion concerning the use of the BPD and suggested the use of the head perimeter and femur measurements.⁷ However, O'Keeffe et al also suggested the use of the abdomen perimeter. In view of the fact that premature infants are often delayed in growth, the abdomen perimeter should not be used because the abdomen perimeter could be less than expected, resulting in some underestimation of fetal age.

Because shape, growth disturbances, and individual variation affect the size of the head to an increasing degree after 28 weeks of gestation, the BPD should be used with some caution after this point.⁸

In the early 1980s, a technique called growth-adjusted sonographic age was proposed to correct the gestational age in serially studied fetuses that were suspected of growth disturbances. Although the technique was an excellent idea from a conceptual viewpoint, it was often impractical to use. It has been shown that the technique did not offer any advantage over a single BPD measurement in a population of well-controlled patients.⁹

Head Perimeter or Head Circumference

The HC is optimally imaged through the thalami and cavum septi pellucida, with the falx cerebri perpendicular to the insonation beam (falx horizontal on image). When the falx is parallel to the scanning plane, turning the patient to the side or waiting a short time may be helpful. A transthalamic plane showing intact fetal calvarium in an elliptical shape, absence of fetal cerebellum, or fetal orbits is optimal. The head perimeter is influenced by growth disorders but to a lesser extent than the BPD. Ott compared the mean error of the head perimeter to the mean error of BPD and found it to be significantly smaller.¹⁰ It is not influenced by dolichocephaly or brachycephaly. The head perimeter is measured in the same transthalamic plane as the BPD, and in general should be measured in the same image used for the BPD. One should make certain that the longest (anteroposterior) length is obtained, which implies that the cavum pellucidum or the roof of posterior fossa is included in the scan. Sometimes the indistinct calvarium of the far-field skull may cause an inadvertent overmeasuring; careful attention to the calvarium thickness such as in the near field allows for correct placement of the calipers around the entire skull. In a laboring patient and in some circumstances of fetal head engagement, a BPD or an HC measure is not possible, and a suboptimal measure should not be used. The head perimeter is either measured with electronic calipers (see Figure 5-6) or computed by using the formula:

A nomogram that allows calculation of gestational age from the head perimeter is provided in Table A5-3.

Femur and Humerus Lengths

Femur length (Table A5-4) was originally measured to diagnose limb dwarfism. It was subsequently observed that femur length was an excellent parameter to determine fetal age.^{3,11,12,13, and 14} The femur can be measured from 10 weeks onward. The femur is measured from the origin to the distal end of the shaft (Figure 5-7). The femoral head and distal epiphysis are not included in the measurement. The optimal technique is to measure the femur closest to the near field, with the FL perpendicular to the insonation beam (horizon rather than vertical), and equal echogenicity throughout its length. A common imaging error of the FL is to demonstrate an incomplete bony diaphysis. Noting distinct shadowing on both ends is sometimes helpful in assuring a correct measurement (see Figure 5-7). The femoral head and distal epiphysis are not included in the measurement. The humerus is measured in the same way (Figure 5-8). The humerus (Table A5-4) is also commonly measured.

Other Parameters

Crown-rump length, BPD, and femur and humerus lengths are by no means the only parameters useful in estimating gestational age. Among the others that have been proposed are other fetal long bones,¹⁴ binocular distance,¹⁵ head perimeter,^16,17 the abdominal perimeter,¹⁶ clavicle, and the size and shape of the fetal ears.¹⁸ The abdominal perimeter has rapidly been abandoned because it is too sensitive to variations of fetal growth.

Among the other long bones, measurement of the humerus is generally the easiest to obtain and the most reproducible. The tibia comes next; the radius and ulna should be used only when confusing results are obtained from other methods. In practice, the bone-derived gestational ages are averaged and compared with the BPD-derived gestational age. If the difference is greater than 11 days, using the bone-derived gestational age is preferred. Table A5-5 (humeral and ulnar length and tibia and clavicular length may be used to determine the gestational age from the long bones of the fetus (Table A5-6).

The binocular distance is a parameter that has occasionally proven useful. To obtain the right plane, one should start from the conventional section of the BPD and move the transducer caudally until the orbits are visualized. In the correct plane, both eyes should have the same diameter, and the image should be symmetrical (Figure 5-9). The largest diameter of the eye should be used; the interocular distance should be the smallest. Table A5-7 can be used to derive the gestational age from the binocular distance.

The cerebellar measurement was proposed by Reece et al,¹⁸ who found that the measurement of the cerebellum correlated well with the gestational age and was not much affected by growth retardation (Figure 5-10). Further, if the measurement is expressed in millimeters, the value is very similar to the gestational age. For instance, if a cerebellum is measured to be 22 mm, the fetus is about 22 weeks in gestational age. This similarity in value makes it easy to use the cerebellum as a rapid check of the gestational age.

The clavicle is also proposed as a measurement to arrive at gestational age. The clavicle has an intramembranous ossification and not an endochondral ossification. This differentiates it from other long bones of the body. The clavicles are affected to a very different degree by diseases such as achondroplasia, thanatophoric dysplasia, and so forth. The clavicle is occasionally useful in such circumstances. Another interesting point is that the length of a clavicle expressed in millimeters is very close to the gestational age expressed in weeks.¹⁹

Ossification Center

Many investigators have also attempted to determine the gestational age based on identification and measurement of ossification centers such as the distal femoral, proximal tibial, and proximal humeral. This method has been associated with a few problems. The first is that the observation of an ossification center by itself is not sufficiently precise. When only the presence of the ossification center is assessed, the fetus can be any age after the age at which the ossification center appears, information that is of little use in clinical practice.

The measurement of the ossification center is also a problem. What appears as the ossification center on ultrasound is only the interface on the proximal side of the center of the ossification center and not the whole ossification center. The thickness and length are not visible unless the ossification center lies exactly perpendicular to the ultrasound beam. This position is difficult to obtain in clinical practice. The ossification centers are small, and a small error in measurement on any of them is equivalent to a large absolute variation in the assessment of gestational age. Therefore, the use of the ossification center is of little value in clinical practice; even in late gestation, the measurement of the femur and humerus is preferred.²⁰

Selection of an Appropriate Table

In view of the large number of different parameters, it is important to know which one to measure and when. Different parameters have different reliability and ease of measurement at different gestational ages. In the list that follows, the parameters are given in decreasing order of preference. This order was established from their reliability, confidence limits, and ease of measurement. It should not be regarded as definitive and can be adapted to specific circumstances.

• From 7 to 10 weeks: CRL

• From 10 to 14 weeks: CRL, BPD, FL humerus length

• From 15 to 28 weeks: BPD, FL, humerus length, head perimeter, binocular distance, other long bone lengths

• After 28 weeks (more accurate for dating): femur length, humerus length, binocular distance, BPD (check that BPD is correct by evaluating the CI), other long bone lengths, head perimeter

When Different Parameters Have Discrepancies in Estimates

The general rule is that the earlier the estimation of the age, the more accurate it is. In other words, the first ultrasound typically establishes the dates. The exception is that with an extremely early gestational age, there can be errors of the measuring technique (ie, calipers are bigger than the distance measured). If estimate of the fetal age has been made at 15 weeks and a follow-up scan at 27 weeks yields a different estimate, do not change the original estimate—it is more accurate. Rather, the follow-up sonogram should indicate fetal growth. Ages are more or less equivalent when they are within 11 days of each other (this is an arbitrary limit). Before 20 weeks, parameters should be remeasured when the difference exceeds 1 week. Clinical information should also be used in assigning gestational age such as the certainty of the LMP, maternal or fetal conditions that affect the biometric parameters (such as gestational diabetes increasing the fetal abdominal circumference [AC]), and the gestational age of the baby.

One important time period to consider is the third trimester, where the rule of thumb is to be very cautious in changing the gestational age, because there is greater fetal biological variation. A growth-restricted fetus may be erroneously assigned a lesser gestational age. In general, with a discrepancy in LMP-ultrasound, the dates should not be definitively assigned on the basis of one ultrasound, but rather a follow-up sonogram can be helpful: a follow-up showing normal fetal growth would be consistent with wrong dates, whereas a follow-up showing little growth would be consistent with fetal growth restriction.

Careful methodology with multiple measures should be used. Although it is reassuring when two similar parameters agree (eg, 2 different bones), the sonographer and sonologist should refrain from "forcing" an image to meet a predicted expectation. When parameters do agree, an average gestational age can calculated. When a few parameters provide estimates that are in the same range with only one discordant, reassess the outlier for methodological errors. If it is still abnormal, do not include it in the final estimate, and consider the reason for its disparity.

We are strongly in favor of reporting the lower 5th and the upper 95th confidence limits on the prediction of each measurement.²¹ For example, the gestational age should be reported at 35 weeks (between 33 and 37 weeks) because this signifies that there is a 95% chance that the age is within 33 to 37 weeks and that the most likely age is 35 weeks. This is especially important in late gestation and may have legal implications.

The fetal abdominal perimeter or circumference (AC) should be determined at the skin line, in other words, at the outer perimeter (Figure 5-11). It should be on a true transverse view at the level of the junction of the umbilical vein, portal sinus, and fetal stomach when visible. Fetal kidneys should not be visible. Noting "whole ribs" of the abdomen assists in verifying the true transverse plane rather than an oblique plane. The AC may be used for gestational age, but caution should be exercised due to growth-restricted fetuses (small AC), or large for gestational age (large AC). The ratio between the HC and AC (HC/AC ratio) may indicate disproportionate AC size. The sonologist must be careful not to inappropriately include the small AC of a growth-restricted fetus, which would appear younger than true age and not growth restricted. A nomogram of the abdominal perimeter is provided in Table A5-8.

The following biometric parameters have also been measured, and interested readers are encouraged to obtain the original report if needed. These include the orbit,²² nose,²³ cheek,²⁴ chin length and upper lip width,²⁵ frontal lobe,²⁶ transcerebellar diameter,²⁷ thyroid,²⁸ lungs,²⁹ heart diameter,³⁰ liver,³¹ sacrum,³² iliac bone,³³ iliac angle,^34,35 and thigh volume.^36,37 Many more have been published, and the interested reader should do a Medline search to identify the most recent articles.