Three-dimensional ultrasound in the assessment of fetal cerebellar transverse and antero-posterior diameters

PII S0301-5629(99)00123-4

● Original Contribution

THREE-DIMENSIONAL ULTRASOUND IN THE ASSESSMENT OF FETALCEREBELLAR TRANSVERSE AND ANTERO-POSTERIOR DIAMETERS

CHIUNG-HSIN CHANG,* FONG-MING CHANG,* CHEN-HSIANG YU,* HUEI-CHEN KO,† andHSI-YAO CHEN

‡

*Department of Obstetrics and Gynecology and†Research Institute of Behavior Medicine, National Cheng KungUniversity Medical College, Tainan, Taiwan; and‡Department of Obstetrics and Gynecology, National Taiwan

University Medical College, Taipei, Taiwan

(Received11 May 1999; in final form 9 August1999)

Abstract—Fetal cerebellum scanning by prenatal ultrasound (US) is very important for early detection of fetalcentral-nervous–system anomaly, as well as for the determination of gestational age (GA). Due to the small organsize and the unique shape of the fetal cerebellum (CL), accurate measurement of the dimensions of CL bytwo-dimensional (2-D) US is not easy if the appropriate plane cannot be reached. With the advent of three-dimensional (3-D) US, the disadvantages of 2-D US in assessing the fetal CL dimensions can be avoided. Thepurpose of this study was to assess the fetal cerebellar transverse diameter (CTD) and cerebellar antero-posteriordiameter (CAD) using 3-D US. First, we compared the reproducibility of 2-D and 3-D US on the assessment offetal cerebellar dimensions. Second, we prospectively measured CTD and CAD in 223 healthy fetuses using across-sectional design with an attempt to establish the normal growth charts of fetal CL. Our results showed 3-DUS is superior to 2-D US in the reproducibility test of fetal cerebellar dimensions. In addition, with GA as thedependent variable, polynomial regression analysis showed that the best-fit equations for both CTD vs. GA andCAD vs. GA were the first-order. The best-fit predictive equation of GA by CTD was GA (week)5 9.028110.585333 CTD (mm) (r 5 0.95,n 5 223, SE5 1.82 weeks,p < 0.0001), and the best-fit predictive equation ofGA by CAD was GA (week)5 10.8551 1.16723 CAD (mm) (r 5 0.82,n 5 223, SE5 3.41 weeks,p < 0.0001).Furthermore, all the correlation coefficients of CTD or CAD vs. the common fetal growth indexes were alsohighly significant (all p < 0.0001). In conclusion, our data of fetal CL dimensions assessed by 3-D US may serveas a useful reference in assessing fetal CL growth, dating GA or detecting fetal CL anomalies. © 2000 WorldFederation for Ultrasound in Medicine & Biology.

Key Words:Three-dimensional ultrasound, Fetus, Cerebellar transverse diameter, Cerebellar antero-posteriordiameter.

INTRODUCTION

Accurate assessment of the fetal organ dimensions byprenatal ultrasound (US) is very crucial in the evaluationof fetal growth and maturation (Chitty et al. 1994a,1994b, 1994c). However, using two-dimensional (2-D)US to precisely assess the growth of small organs, suchas cerebellum (CL), is not an easy task. Although thefetal CL is scanned now for screening the central-ner-vous–system (CNS) anomaly (Benacerraf et al. 1989;

Persutte et al. 1997; Robins et al. 1998) and for deter-mining the gestational age (GA) (Goldstein et al. 1987).In fact, to measure the accurate “largest” diameters offetal CL in an attempt to determine the cerebellar trans-verse diameter (CTD) (Goldstein et al. 1987) or cerebel-lar antero-posterior diameter (CAD) (Smith et al. 1986)in utero is relatively difficult by 2-D US. Because thefetal CL is a small organ with a unique shape, to measurethe “largest” diameters of CL by 2-D US is prone tosubstantial errors if the accurate plane for the measure-ment of CTD and CAD is not reached.

With the advent of three-dimensional (3-D) US, itbecomes an easy technique to access any possible viewsand planes whenever the scanned 3-D volume is obtained(Kelly et al. 1992; Kuo et al. 1992; Lee et al. 1994; Merzet al. 1995; Riccabona et al. 1996). Moreover, precise

Address correspondence to: Dr. Fong-Ming Chang, Departmentof Obstetrics and Gynecology, National Cheng Kung University Med-ical College, 138 Victory Road, Tainan 70428 Taiwan. E-mail:[email protected]

This paper has been reported in part as a poster in the 1998Annual Meeting of American Institute of Ultrasound in Medicine(AIUM) in Boston, MA, March 1998.

Ultrasound in Med. & Biol., Vol. 26, No. 2, pp. 175–182, 2000Copyright © 2000 World Federation for Ultrasound in Medicine & Biology

Printed in the USA. All rights reserved0301-5629/00/$–see front matter

175

quantitative measurement of fetal organ dimensions be-comes feasible when the 3-D volume is retrieved (Lee etal. 1994; Merz et al. 1995; Riccabona et al. 1996). Inaddition to our primary application of 3-D US in obstet-rics (Kuo et al. 1992), we have recently reported a seriesof fetal organ assessments using 3-D US; for example,fetal liver, heart, upper arm and thigh, from early secondtrimester to third trimester, and obtained more accurateresults (Chang et al. 1997a, 1997b, 1997c; Liang et al.1997).

Because 3-D US can overcome the obstacle of 2-DUS in reaching the accurate plane for measuring fetal CLdimensions, this study was undertaken to investigate twopurposes. First, we compared the reproducibility of 2-Dand 3-D US in the assessment of fetal cerebellar dimen-sions. Second, we attempted to establish the normalgrowth charts of fetal cerebellar transverse diameter(CTD) and cerebellar antero-posterior diameter (CAD)during gestation, using the new modality of 3-D US.

MATERIALS AND METHODS

SubjectsThe inclusion criteria of subjects in this study were

as follows: 1. patients with defined last menstrual period(LMP) and confirmed by a dating US examination inearly pregnancy, either by crown-rump length (CRL) orbiparietal diameter (BPD); 2. singleton pregnancies withGA ranging from 20 to 40 weeks; 3. healthy pregnancieswithout maternal or fetal complications; and 4. the USexamination of each fetus being selected only once inthis series (a cross-sectional study). The setting was atthe Ultrasound Unit of the Department of Obstetrics andGynecology, National Cheng Kung University Hospital.All the fetuses scanned in this study were followed todelivery to ensure that they were born healthy.

3-D ultrasonographyWe used the 3.5- or 5-MHz transabdominal Volu-

son transducer of the 3-D US equipment (Combison530D, Kretz-Technik, Zipf, Austria) for fetal cerebellarscanning. The details of the 3-D US scanning were aspreviously described (Chang et al. 1997a, 1997b, 1997c;Liang et al. 1997; Merz et al. 1995). Initially, the high-resolution, real-time 2-D US scanner was applied forscanning of the typical plane for fetal CL described bySmith et al. (1986). Then, we turned on the 3-D trans-abdominal Voluson sector transducer to scan the cere-bellar hemispheres and vermis with the normal velocitymode (which swept 40° automatically within 4 s) whenthe fetus was at rest. The 3-D volume can be evaluatedslice by slice in any arbitrary plane. The built-in 3-Dview software permitted the simultaneous display of the3-D volume in three perpendicular planes on the monitor.

As shown in Fig. 1, when the fetal cerebellum wasrotated to a horizontal position, the cerebellar transversediameter (CTD) was measured at the largest diameter bytracing axially in the cross-sectional planes of 3-D US. Ina similar manner, the cerebellar antero-posterior diame-ter (CAD) was measured at the largest diameter bytracing axially in the cross-sectional planes of 3-D US.The data set was further saved into the built-in computeror in the laser disks for further retrieval and processing,such as volume determination or 3-D image reconstruc-tion.

In addition, we measured the fetal biparietal diam-eter (BPD), occipito-frontal diameter (OFD), head cir-cumference (HC), abdominal circumference (AC) andfemur length (FL) according to the methods described inprevious reports (Chitty et al. 1994a, 1994b, 1994c;Chang et al. 1996, 1998). Estimated fetal weight (EFW)was calculated from the weight-predicting equation com-posed of BPD, AC and FL, reported by Hsieh andcolleagues, derived from a Taiwanese population (Hsiehet al. 1987).

StatisticsAll the data of cerebellar measurements were put

into an IBM-compatible personal computer (PC) for finalanalysis. We used the SPSS-PC statistical package (Chi-cago, IL) to perform statistical calculation. Pairedt-tests(with mean, standard deviation, SD, and standard error,SE, of the differences) and correlation coefficients wereused to compare the superiority of reproducibility (Daw-

Fig. 1. 3-D US in the measurement of fetal cerebellar transversediameter (CTD, between the black and white arrow head in theupper-left panel) and cerebellar antero-posterior diameter(CAD, between the white arrows in the upper-left panel). Whenthe 3-D volume was scanned and retrieved, the most appropri-ate plane for measuring the CTD and CAD was reached afterrotating the scans in the three orthogonal planes in the upper-left, upper-right, and lower-left panels. The relative position ofthe plane is displayed simultaneously in the lower-right panel.

176 Ultrasound in Medicine and Biology Volume 26, Number 2, 2000

son-Saunders and Trapp 1994; Sladkevicius and Valen-tin 1995). Linear regression and the correlation analysiswere calculated to test the relationship between the in-dependent and dependent variables. With GA as depen-dent variable, polynomial regression analysis was calcu-lated from the first order to the fourth order to find thebest-fit equations to obtain the predicted values of GAusing CTD and CAD as the independent variables. Sim-ilarly, to compare with previous series (Goldstein et al.1987; Smith et al. 1986) in which GA was used as theindependent variable, polynomial regression analysiswas calculated from the first order to the fourth order tofind the best-fit equations to obtain the predicted valuesof CTD and CAD using GA as the independent variable(Chitty et al. 1994a, 1994b, 1994c). Finally, multipleregression analysis was calculated to show the predictingaccuracy using both CTD and CAD together to predictGA. A p value of less than 0.05 was considered statisti-cally significant.

RESULTS

Reproducibility comparison of 2-D and 3-D ultrasoundin the assessment of cerebellar dimensions

Because all the measurements were done by two ofthe authors (C-H. C. and F-M. C.; most cases by C-H. C.and the rest by F-M. C.), the intraobserver error of oneauthor (C-H. C.) and the interobserver error of two

authors (C-H. C. and F-M. C.) were assessed. In thissection, we tested in 30 subjects for reproducibility as-sessment. As shown in Table 1, in the assessment ofCTD, the intraobserver reproducibility of 2-D US (meanof differences20.663; SD of difference 0.776; intraob-server correlation coefficient: 0.996) was less precisethan that of 3-D US (mean of differences20.050; SD ofdifferences 0.283; intraobserver correlation coefficient0.999). In addition, the pairedt-test showed that thedifferences between repeated measurements of CTD by2-D US were different (p , 0.001), but the differencesbetween repeated measurements of CTD by 3-D USwere not. Furthermore, in the assessment of CAD, theintraobserver reproducibility of 2-D US (mean of differ-ences2 3.237; SD of differences 2.291; intraobservercorrelation coefficient 0.880) was also less precise thanthat of 3-D US (mean of differences20.053; SD ofdifferences 0.086; intraobserver correlation coefficient0.999). Although the CAD assessment with 3-D US hadintraobserver difference, yet the minimal SD and SEvalues of intraobserver errors (0.086 and 0.016) andremarkably high intraobserver correlation coefficient(0.999) with 3-D US, in contrast to larger SD and SE ofintraobserver errors (2.291 and 0.148) and lower intraob-server correlation coefficient (0.880) with 2-D US, weresufficient to prove that 3-D is superior to 2-D US in theassessment of intraobserver reproducibility of CAD.

Table 1. Comparison of intraobserver reproducibility for 2-D and 3-D US in the assessment of fetalcerebellar diameters

Items n Mean SD SE ICC P-t p

2-D USTransverse diameter (CTD), mm 30 20.663 0.776 0.142 0.996 24.68 ,0.001Antero-posterior diameter (CAD), mm 30 23.237 2.291 0.148 0.880 27.74 ,0.001

3-D USTransverse diameter (CTD), mm 30 20.050 0.283 0.052 0.999 20.97 NSAntero-posterior diameter (CAD), mm 30 0.053 0.086 0.016 0.999 3.40 ,0.01

n 5 Number of cases; Mean5 mean of intraobserver differences; SD5 standard deviation of intraobserver differences; SE5standard error of intraobserver differences; ICC5 intraobserver correlation coefficient; P-t 5 pairedt value;p 5 two-tailedp value.

Table 2. Comparison of interobserver reproducibility for 2-D and 3-D US in the assessment of fetalcerebellar diameters

Items n Mean SD SE ICC P-t p

2-D USTransverse diameter (CTD), mm 30 1.343 6.323 1.154 0.799 1.16 NSAntero-posterior diameter (CAD), mm 30 22.037 3.004 0.548 0.787 23.71 0.001

3-D USTransverse diameter (CTD), mm 30 0.040 0.393 0.072 0.999 0.56 NSAntero-posterior diameter (CAD), mm 30 0.046 0.642 0.117 0.987 0.40 NS

n 5 Number of cases; Mean5 mean of interobserver differences; SD5 standard deviation of interobserver differences; SE5standard error of interobserver differences; ICC5 interobserver correlation coefficient; P-t 5 pairedt value;p 5 two-tailedp value.

Fetal cerebellar growth by 3-D ultrasound● C-H. CHANG et al. 177

As listed in Table 2, in the assessment of CTD, theinterobserver reproducibility of 2-D US (mean of differ-ences 1.343; SD of differences 6.323; interobserver cor-relation coefficient 0.799) was less precise than that of3-D US (mean of differences 0.040; SD of differences0.393; interobserver correlation coefficient 0.999). Al-though the CTD assessment with 2-D US had no inter-observer difference examined by pairedt-test, the greaterSD and SE values of interobserver errors (6.323 and1.154) and lower interobserver correlation coefficient(0.799) with 2-D US, in contrast to smaller SD and SE ofinterobserver errors (0.393 and 0.072) and remarkablyhigher intraobserver correlation coefficient (0.999) with3-D US, were sufficient to prove that 3-D is superior to2-D US in the assessment of interobserver reproducibil-

ity of CTD. Furthermore, in the assessment of CAD, theinterobserver reproducibility of 2-D US (mean of differ-ences22.037; SD of differences 3.004; interobservercorrelation coefficient 0.787) was also less precise thanthat of 3-D US (mean of differences 0.046; SD of dif-ferences 0.642; interobserver correlation coefficient0.987). Moreover, the pairedt-test showed that the in-terobserver errors of CAD by 2-D US were different(p 5 0.001), but the interobserver errors of CAD by 3-DUS were not. From the above analysis, we concluded that3-D US is superior to 2-D US in both the intraobserverand interobserver reproducibility in the assessment offetal cerebellar dimensions.

GA as the dependent variableIn total, 223 cases were included in this study for

final analysis. With GA as the dependent variable (as thex-axis), Figs. 2 and 3 illustrate the scattergrams of CTD

Fig. 2. With gestational age (GA) as the dependent variable(y-axis), scattergram and correlation between GA and fetal

cerebellar transverse diameter (CTD) are illustrated.

Fig. 3. With gestational age (GA) as the dependent variable(y-axis), scattergram and correlation between GA and fetal

cerebellar antero-posterior diameter (CAD) are depicted.

Table 3. Predicted values of gestational age (GA) usingcerebellar transverse diameter (CTD) as the

independent variable

CTD(mm)

GA, centiles (weeks)

5th 10th 25th 50th 75th 90th 95th

20 17.7 18.4 19.5 20.7 21.9 23.0 23.821 18.3 19.0 20.1 21.3 22.5 23.6 24.322 18.9 19.6 20.7 21.9 23.1 24.2 24.923 19.5 20.2 21.3 22.5 23.7 24.8 25.524 20.1 20.7 21.8 23.1 24.3 25.4 26.125 20.6 21.3 22.4 23.7 24.9 26.0 26.726 21.2 21.9 23.0 24.2 25.4 26.5 27.327 21.8 22.5 23.6 24.8 26.0 27.1 27.928 22.4 23.1 24.2 25.4 26.6 27.7 28.429 23.0 23.7 24.8 26.0 27.2 28.3 29.030 23.6 24.3 25.4 26.6 27.8 28.9 29.631 24.2 24.8 25.9 27.2 28.4 29.5 30.232 24.7 25.4 26.5 27.8 29.0 30.1 30.833 25.3 26.0 27.1 28.3 29.5 30.6 31.434 25.9 26.6 27.7 28.9 30.1 31.2 32.035 26.5 27.2 28.3 29.5 30.7 31.8 32.536 27.1 27.8 28.9 30.1 31.3 32.4 33.137 27.7 28.4 29.4 30.7 31.9 33.0 33.738 28.2 28.9 30.0 31.3 32.5 33.6 34.339 28.8 29.5 30.6 31.9 33.1 34.1 34.940 29.4 30.1 31.2 32.4 33.6 34.7 35.541 30.0 30.7 31.8 33.0 34.2 35.3 36.042 30.6 31.3 32.4 33.6 34.8 35.9 36.643 31.2 31.9 33.0 34.2 35.4 36.5 37.244 31.8 32.5 33.5 34.8 36.0 37.1 37.845 32.3 33.0 34.1 35.4 36.6 37.7 38.446 32.9 33.6 34.7 36.0 37.2 38.2 39.047 33.5 34.2 35.3 36.5 37.7 38.8 39.648 34.1 34.8 35.9 37.1 38.3 39.4 40.149 34.7 35.4 36.5 37.7 38.9 40.0 40.750 35.3 36.0 37.1 38.3 39.5 40.6 41.351 35.9 36.6 37.6 38.9 40.1 41.2 41.952 36.4 37.1 38.2 39.5 40.7 41.8 42.553 37.0 37.7 38.8 40.1 41.3 42.3 43.154 37.6 38.3 39.4 40.6 41.8 42.9 43.755 38.2 38.9 40.0 41.2 42.4 43.5 44.2


and CAD vs. GA, respectively, together with the linearregression equations and correlation coefficients. All thecorrelation coefficients of CTD vs. GA (r5 0.95,p ,0.0001) and CAD vs. GA (r5 0.82,p , 0.0001) werehighly significant. In addition, with GA as the dependentvariable, polynomial regression analysis showed that thebest-fit equations for both CTD vs. GA and CAD vs. GAwere the first order. The best-fit predictive equation ofGA by CTD was GA (week)5 9.02811 0.585333CTD (mm) (r 5 0.95,n 5 223, SE5 1.82 weeks,p ,0.0001), and the best-fit predictive equation of GA byCAD was GA (week)5 10.8551 1.16723 CAD (mm)(r 5 0.82,n 5 223, SE5 3.41 weeks,p , 0.0001). Forclinical relevance, we further used CTD and CAD as theindependent variables and GA as the dependent variableto generate two predicted charts of GA (Tables 3 and 4).

Comparison with previous series using CTD and CAD asthe dependent variables

To compare with previous reports in which GA asthe independent variable was used (Goldstein et al. 1987;Smith et al. 1986), polynomial regression showed thebest-fit equations to predict CTD and CAD were both thefirst-order polynomial equations. The best-fit predictiveequation of CTD by GA was CTD5 210.632 11.54863 GA (r 5 0.95, n 5 223, p , 0.0001). Thebest-fit predictive equation of CAD by GA was CAD520.896461 0.576533 GA (r 5 0.82, n 5 223, p ,0.0001). Accordingly, the predicted values of CTD andCAD using GA as the independent variable are listed inTables 5 and 6.

Tables 7 and 8 are listed to show our predictedvalues of CTD and CAD using 3-D US vs. the data from

the previous series using 2-D US (Goldstein et al. 1987;Smith et al. 1986). In general, our data of CTD and CADusing 3-D US are similar to those reported in other seriesusing 2-D US (Goldstein et al. 1987; Smith et al. 1986).

Table 4. Predicted values of gestational age (GA) usingcerebellar antero-posterior diameter (CAD) as the


CAD(mm)

GA, centiles (weeks)


10 16.9 18.2 20.3 22.5 24.8 26.8 28.211 18.0 19.4 21.4 23.7 25.9 28.0 29.412 19.2 20.6 22.6 24.9 27.1 29.2 30.513 20.4 21.7 23.8 26.0 28.3 30.3 31.714 21.5 22.9 24.9 27.2 29.4 31.5 32.915 22.7 24.1 26.1 28.4 30.6 32.7 34.016 23.9 25.2 27.3 29.5 31.8 33.8 35.217 25.0 26.4 28.4 30.7 32.9 35.0 36.418 26.2 27.6 29.6 31.9 34.1 36.2 37.519 27.4 28.7 30.8 33.0 35.3 37.3 38.720 28.5 29.9 31.9 34.2 36.4 38.5 39.921 29.7 31.1 33.1 35.4 37.6 39.7 41.022 30.9 32.2 34.3 36.5 38.8 40.8 42.223 32.0 33.4 35.5 37.7 40.0 42.0 43.424 33.2 34.6 36.6 38.9 41.1 43.2 44.525 34.4 35.7 37.8 40.0 42.3 44.3 45.7

Table 5. Predicted values of cerebellar transverse diameter(CTD) using gestational age (GA) as the independent

variable

GA(weeks)

CTD, centiles (mm)


20 15.4 16.5 18.3 20.3 22.4 24.1 25.321 17.0 18.1 19.9 21.9 23.9 25.7 26.822 18.5 19.6 21.4 23.4 25.5 27.2 28.423 20.1 21.2 23.0 25.0 27.0 28.8 29.924 21.6 22.7 24.5 26.5 28.6 30.3 31.525 23.2 24.3 26.1 28.1 30.1 31.9 33.026 24.7 25.8 27.6 29.6 31.6 33.4 34.627 26.3 27.4 29.2 31.2 33.2 35.0 36.128 27.8 28.9 30.7 32.7 34.7 36.5 37.729 29.4 30.5 32.3 34.3 36.3 38.1 39.230 30.9 32.0 33.8 35.8 37.8 39.6 40.831 32.4 33.6 35.4 37.4 39.4 41.2 42.332 34.0 35.1 36.9 38.9 40.9 42.7 43.833 35.5 36.7 38.5 40.5 42.5 44.3 45.434 37.1 38.2 40.0 42.0 44.0 45.8 46.935 38.6 39.8 41.6 43.6 45.6 47.4 48.536 40.2 41.3 43.1 45.1 47.1 48.9 50.037 41.7 42.9 44.6 46.7 48.7 50.5 51.638 43.3 44.4 46.2 48.2 50.2 52.0 53.139 44.8 46.0 47.7 49.8 51.8 53.6 54.740 46.4 47.5 49.3 51.3 53.3 55.1 56.2

Table 6. Predicted values of cerebellar antero-posteriordiameter (CAD) using gestational age (GA) as the


GA(weeks)

CAD, centiles (mm)


20 6.7 7.6 9.0 10.6 12.3 13.7 14.621 7.2 8.1 9.6 11.2 12.8 14.3 15.222 7.8 8.7 10.2 11.8 13.4 14.9 15.823 8.4 9.3 10.7 12.4 14.0 15.4 16.324 9.0 9.9 11.3 12.9 14.6 16.0 16.925 9.5 10.4 11.9 13.5 15.1 16.6 17.526 10.1 11.0 12.5 14.1 15.7 17.2 18.127 10.7 11.6 13.0 14.7 16.3 17.7 18.628 11.3 12.2 13.6 15.2 16.9 18.3 19.229 11.8 12.8 14.2 15.8 17.5 18.9 19.830 12.4 13.3 14.8 16.4 18.0 19.5 20.431 13.0 13.9 15.3 17.0 18.6 20.0 21.032 13.6 14.5 15.9 17.6 19.2 20.6 21.533 14.2 15.1 16.5 18.1 19.8 21.2 22.134 14.7 15.6 17.1 18.7 20.3 21.8 22.735 15.3 16.2 17.7 19.3 20.9 22.4 23.336 15.9 16.8 18.2 19.9 21.5 22.9 23.837 16.5 17.4 18.8 20.4 22.1 23.5 24.438 17.0 17.9 19.4 21.0 22.6 24.1 25.039 17.6 18.5 20.0 21.6 23.2 24.7 25.640 18.2 19.1 20.5 22.2 23.8 25.2 26.1


Correlation vs. the common growth indexes of fetal bi-ometry

Figure 4 demonstrates the scattergram and correla-tion of CTD vs. BPD, illustrating the correlation coeffi-cient of CTD vs. BPD was highly significant (r5 0.94,p , 0.0001). Similarly, as shown in Table 9, all thecorrelation coefficients of CTD vs. the other common

fetal growth indexes (i.e., CTD vs. OFD, CTD vs. HC,CTD vs. AC, CTD vs. FL, and CTD vs. EFW) werehighly significant (allp , 0.0001), indicating the growthof CTD was related to the common growth indexes offetal biometry. Furthermore, as depicted in Fig. 5, thecorrelation coefficient of CAD vs. BPD was highly sig-nificant (r 5 0.94, p , 0.0001). Similarly, as listed inTable 9, all the correlation coefficients of CAD vs. theother common fetal growth indexes (i.e., CAD vs. OFD,CAD vs. HC, CAD vs. AC, CAD vs. FL, and CAD vs.EFW) were highly significant (allp , 0.0001), indicat-ing the growth of CAD was related to the commongrowth indexes of fetal biometry.

Using both CTD and CAD to predict GATo answer the question of whether the inclusion of

both CTD and CAD can improve the predicting accuracyof GA, we applied multiple regression analysis to useCTD and CAD together in predicting GA. Our resultsshowed that when only CTD was used as the indepen-dent variable, the correlation coefficient was 0.95, withSE of estimation as 1.82 weeks. When CTD and CADwere used together as the independent variables, thebest-fit equation was GA (week)5 8.45451 0.529183CTD (mm)1 0.157653 CAD (mm), with a multipleRas 0.95 and SE of estimation as 1.79 weeks. Hence, theinclusion of CTD and CAD to predict GA only presentedminimal improvement that was not significant.

DISCUSSION

Accurate estimation of the GA is important in mod-ern obstetrics. The measurement of fetal BPD is one ofthe most useful biometries to estimate the GA (Chitty etal. 1994a; Chang et al. 1996). But, in the followingconditions, the measurements of BPD may be unreliable

Table 7. Comparison of cerebellar transverse diameter (CTD)with other series

GA(weeks)

Goldstein et al.(1987)(mm)

Smith et al.(1986)(mm)

This studyCTD(mm)

20 20.0 20.5 20.321 22.0 21.8 21.922 23.0 23.1 23.423 24.0 24.4 25.024 25.0 25.7 26.525 28.0 27.0 28.126 29.0 28.4 29.627 30.0 29.7 31.228 31.0 31.0 32.729 34.0 32.3 34.330 35.0 33.6 35.831 38.0 34.9 37.432 38.0 36.3 38.933 40.0 37.6 40.534 40.0 38.9 42.035 40.5 40.2 43.636 43.0 41.5 45.137 45.0 42.9 46.738 48.5 44.2 48.239 52.0 45.5 49.840 – 46.8 51.3

Table 8. Comparison of cerebellar antero-posterior diameter(CAD) with other series

GA (weeks)Smith et al. (1986)

CAD (mm)This studyCAD (mm)

20 10.2 10.621 10.9 11.222 11.5 11.823 12.2 12.424 12.9 12.925 13.6 13.526 14.3 14.127 15.0 14.728 15.7 15.229 16.4 15.830 17.1 16.431 17.8 17.032 18.5 17.633 19.2 18.134 19.9 18.735 20.6 19.336 21.3 19.937 22.0 20.438 22.7 21.039 23.4 21.640 24.1 22.2

Fig. 4. Scattergram and correlation between the fetal cerebellartransverse diameter (CTD) and biparietal diameter (BPD).


because of distortion, compression, or crowding of thefetal head, breech presentation, oligohydramnios, or mul-tiple gestations (Chang et al. 1996, 1998). While in thethird trimester, when the fetus is in the vertex presenta-tion and the fetal head is engaged, measurement of theBPD also becomes difficult. In such conditions, the mea-surement of fetal CTD in the estimation of the GA is anaccurate method, even in cases of abnormal shape of thefetal head; for instance, brachycephaly or dolichocephaly(Goldstein et al. 1987; Smith et al. 1986). After 30 weeksof gestation, measurements of the fetal CL by 2-D US aremore difficult. It may be due to the increased density ofthe bones forming the base of the skull. Or, it may be dueto the folia formation and fissuring across the CL in thethird trimester (Goldstein et al. 1987; Smith et al. 1986).Furthermore, determination of the “largest” diameters of

CTD and CAD is subject to substantial error if theaccurate plane of measurement is not obtained by 2-DUS.

Our results indicate that 3-D US is superior to 2-DUS in the reproducibility test of fetal cerebellar dimen-sions, either the intraobserver error or the interobservererror. Thus, in this prospective study, we use 3-D ultra-sonography to determine the normal growth charts offetal CL during gestation without the weakness of 2-Dultrasonography. However, our results by 3-D US aresimilar to those by 2-D US (Goldstein et al. 1987; Smithet al. 1986).

Of interest, Goldstein et al. (1987) reported thatfetal CTD was not affected in fetuses with intrauterinegrowth restriction and, thus, can be used as a goodparameter to assess fetal GA. In contrast, Huang and Liu(1993) compared the growth of cerebellar vermis (ante-rior-posterior distance of vermis) postnatally in new-borns of appropriate for gestational age (AGA) and smallfor gestational age (SGA). They concluded that the non-fetal cause of intrauterine growth-retarded newbornsmay demonstrate growth retardation in the cerebellarvermis at late gestation. However, they had not com-pared the CAD or CTD postnatally in newborns of AGAand SGA (Huang and Liu 1993). Hence, we still cannotconclude whether the fetal CTD and CAD are affected inthe SGA fetuses. Further studies are needed to comparethe CTD and CAD values in fetuses of AGA and SGA.

To date, no data of normal fetal CTD and CADassessed by 3-D US have been reported. Our studyshowed that the normal charts of fetal CTD and CADassessed by 3-D US can be used as a good reference forevaluating the normal fetal growthin utero. Furthermore,the fetal CTD and CAD correlated well with BPD, OFD,

Fig. 5. Scattergram and correlation between the fetal cerebellarantero-posterior diameter (CAD) and biparietal diameter

(BPD).

Table 9. The correlation and regression equations of cerebellar transverse diameter (CTD) and cerebellar antero-posterior diameter (CAD) vs. common fetal growth indexes

Y X Regression equations n r p

CTD (mm) BPD (cm) Y 5 211.261 1 6.2965X 223 0.94 ,0.0001CTD (mm) OFD (cm) Y 5 216.387 1 5.6751X 223 0.90 ,0.0001CTD (mm) HC (cm) Y 5 215.080 1 1.9402X 223 0.93 ,0.0001CTD (mm) AC (cm) Y 5 23.1427 1 1.5483X 223 0.93 ,0.0001CTD (mm) FL (cm) Y 5 27.0679 1 7.8576X 223 0.93 ,0.0001CTD (mm) EFW (g) Y 5 19.280 1 0.00956X 223 0.93 ,0.0001CAD (mm) BPD (cm) Y 5 20.98769 1 2.3150X 223 0.80 ,0.0001CAD (mm) OFD (cm) Y 5 23.8086 1 2.1964X 223 0.81 ,0.0001CAD (mm) HC (cm) Y 5 22.8581 1 0.73395X 223 0.81 ,0.0001CAD (mm) AC (cm) Y 5 1.6555 1 0.58579X 223 0.82 ,0.0001CAD (mm) FL (cm) Y 5 20.13565 1 3.0289X 223 0.83 ,0.0001CAD (mm) EFW (g) Y 5 10.323 1 0.00351X 223 0.79 ,0.0001

Y 5 dependent variable;X 5 indpendent variable;n 5 number of cases; r5 correlation coefficients;p 5 p values of correlationcoefficient; CTD5 cerebellar transverse diameter; CAD5 cerebellar antero-posterior diameter; BPD5 biparietal diameter; OFD5occipito-frontal diameter; HC5 head circumference; AC5 abdominal circumference; FL5 femur length; EFW5 estimated fetalweight.


HC, AC, FL and EFW in normal pregnancies. We mayapply these normal data as a reference for dating the GAor detecting the CNS anomalies in addition to assessingthe fetal CL growth.

Acknowledgements—This study was supported in part by grants fromNational Science Council, Taipei, Taiwan, to C-H. C. (NSC 88-2314-B006-129), to F-M. C. (NSC 87-2314-B006-131, NSC 88-2314-B006-126), and to C-H. Y. (NSC 88-2314-B006-131). The authors aregrateful to Dr. Ren-Ing Liang, and Ms. Wen-Chu Chen, Yeh-ChinCheng and Yi-Jen Wang for their assistance.

REFERENCES

Benacerraf BR, Stryker J, Frigoletto FD. Abnormal US appearance ofthe cerebellum (banana sign): Indirect sign of spina bifida. Radiol-ogy 1989;171:151–153.

Chang CH, Chang FM, Yao BL, et al. Re-analysis of fetal biparietaldiameter during gestation in Taiwanese by Altman’s method. J MedUltrasound 1996;4:162–168.

Chang CH, Chang FM, Yu CH, et al. Fetal head circumference innormal pregnancy: remodeling by Altman’s method. J Med Ultra-sound 1998;6:61–67.

Chang FM, Hsu KF, Ko HC, Yao BL, Chang CH, Yu CH, Liang RI,Chen HY. Fetal heart volume assessment by three-dimensionalultrasound. Ultrasound Obstet Gynecol 1997a;9:42–48.

Chang FM, Hsu KF, Ko HC, et al. Three-dimensional ultrasoundassessment of fetal liver volume in normal pregnancy: a compari-son of reproducibility with two-dimensional ultrasound and asearch for a volume constant. Ultrasound Med Biol 1997b;23:381–389.

Chang FM, Liang RI, Ko HC, Yao BL, Chang CH, Yu CH. Three-dimensional ultrasound-assessed fetal thigh volume in predictingbirth weight. Obstet Gynecol 1997c:90:331–339.

Chitty LS, Altman DG, Henderson A, et al. Charts of fetal size: 2. Headmeasurements. Br J Obstet Gynaecol 1994a:101:35–43.

Chitty LS, Altman DG, Henderson A, et al. Charts of fetal size: 3.Abdominal measurements. Br J Obstet Gynaecol 1994b:101:125–131.

Chitty LS, Altman DG, Henderson A, et al. Charts of fetal size: 4.Femur length. Br J Obstet Gynaecol 1994c;101:132–135.

Dawson-Sauders B, Trapp RG. Basic and clinical biostatistics. 2nd ed.Norwalk, CT: Appleton & Lange, 1994.

Goldstein I, Reece EA, Pilu G, Bovicelli L, Hobbins JC. Cerebellarmeasurements with ultrasonography in the evaluation of fetal growthand development. Am J Obstet Gynecol 1987;156:1065–1069.

Hsieh FJ, Chang FM, Huang HC, Lu CC, Ko TM, Chen HY. Com-puter-assisted analysis for predicting fetal weight by ultrasound. JFormosan Med Assoc 1987;86:957–964.

Huang CC, Liu CC. The differences in growth of cerebellar vermisbetween appropriate-for-gestational-age and small-for-gestational-age newborns. Early Hum Dev 1993;33:9–19.

Kelly IMG, Gardener JE, Lees WR. Three-dimensional fetal ultra-sound. Lancet 1992;339:1062–1064.

Kuo HC, Chang FM, Wu CH, Yao BL, Liu CH. The primary applica-tion of three-dimensional ultrasonography in obstetrics. Am J Ob-stet Gynecol 1992;166:880–886.

Lee A, Deutinger J, Bernaschek G. Voluvision: three-dimensionalultrasonography of fetal malformations. Am J Obstet Gynecol1994;170:1312–1314.

Liang RI, Chang FM, Yao BL, Chang CH, Yu CH, Ko HC. Pre-dicting birth weight by fetal upper arm volume with use ofthree-dimensional ultrasonography. Am J Obstet Gynecol 1997;177:632– 638.

Merz E, Bahlmann F, Weber G. Volume scanning in the evaluation offetal malformations: a new dimension in prenatal diagnosis. Ultra-sound Obstet Gynecol 1995;5:222–227.

Persutte WH, Coury A, Hobbins JC. Correlation of fetal frontal lobeand transcerebellar diameter measurements: the utility of a newprenatal sonographic technique. Ultrasound Obstet Gynecol 1997;10:94–97.

Riccabona M, Nelson TR, Pretorius DH. Three-dimensional ultra-sound: accuracy of distance and volume measurements. UltrasoundObstet Gynecol 1996;7:429–434.

Robins JB, Mason GC, Waters J, et al. Case report: cerebellar hemi-hypoplasia. Prenat Diag 1998;18:173–177.

Sladkevicius P, Valentin L. Reproducibility of Doppler measure-ments of blood flow velocity in the uterine and ovarian arteriresin premenopausal women. Ultrasound Med Biol 1995;21:313–319.

Smith PA, Johansson D, Tzannatos C, et al. Prenatal measurement ofthe fetal cerebellum and cisterna cerebellomedullaris by ultrasound.Prenat Diagn 1986;6:133–141.


Documents

Three-dimensional ultrasound in the assessment of fetal cerebellar transverse and antero-posterior diameters