Fetal volume measurements in the first trimester of pregnancy with three-dimensional ultrasound

Preterm birth and a low birth weight are major complications with significant consequences for families and society. It is expected that these complications are the result of the intra-uterine conditions in the first trimester of pregnancy. If it would be possible to detect the fetus at risk early in pregnancy, then the obstetric care can be adjusted accordingly. Earlier reports suggested that fetal growth in the first trimester of pregnancy is of significant value in assessing these complications in pregnancy, the clinical value of these findings is unknown because of the small differences between normal and abnormal growth (Chapter 1). If the fetal size is measured with routine two-dimensional ultrasound, the differences between normal and abnormal fetal size are small. The extra third dimension with three-dimensional ultrasound is expected to give more information about fetal development. The fetal volume rises seven times faster than the crown-rump-length (routine two-dimensional measurement), so impaired fetal growth will be more obvious. This thesis describes in vitro and in vivo studies in order to analyze the still rather complex volume measurements with three-dimensional ultrasound. Furthermore the predictive value of fetal volume measurements in relation to pregnancy outcome is discussed. In vitro studies Three-dimensional volume measurements are expected to be of diagnostic value in general gynecologic and obstetric practice. Despite that the introduction of volume measurements with VOCAL (Volume Organ Computer Aided AnaLysis) was an advancement, the volume measurements are still rather time consuming and complex, as explained in Chapter 1. The learning curve for volume measurements with three-dimensional ultrasound and VOCAL were analyzed in Chapter 2. There is no significant learning curve for volume measurements with three-dimensional ultrasound. In addition, the measurements from inexperienced sonographers were similar to those of an expert. Chapter 3 describes the relation between the volume of an object and the measurement error in vitro for a range of volumes that are comparable to actual fetal volumes in the first trimester of pregnancy. The results show that the percentage error, i.e. absolute measurement error expressed as a percentage of the actual volume, was smaller for larger objects. One should be aware of the volume-dependent absolute and percentage measurement error when interpreting the measured values. Explorative research in order to develop and verify a more practical semi-automated method for volume calculations with 3DUS images is evaluated in Chapter 4. The results of this study show that mathematical volume calculations are possible with the newly developed semi-automated method. This method was successfully applied on a first trimester fetus, where the points of interest at the contour of the fetal head and body were detected. We also succeeded in detecting voxels in the whole contour, including the limbs, of a first trimester fetus with a gestational age of 12 weeks. In vivo studies The high inter- and intra-observer reliability of abdominal fetal volume measurements with three-dimensional ultrasound measurements of the fetal head and rump, i.e. an inter- and intra-class correlation of 0.934 and 0.994, respectively is discussed in Chapter 5. Because of all these promising results, a prospective cohort study was performed to determine whether it is possible to detect a fetus at risk for preterm birth and/or low birth weight by measuring the fetal volume with three-dimensional ultrasound in the first trimester of pregnancy, of which the study protocol is described in Chapter 6. The results of this prospective cohort study are reported in Chapter 7. The difference in mean percentage error between normal and complicated pregnancies (preterm birth and/or low birth weight) was neither significant nor clinically relevant. The fetal volumes of the neonates born after preterm birth and/or low birth weight are distributed throughout the range of the neonates born a normal birth weight, indicating that it is hard to distinguish the complicated pregnancies from the normal ones by fetal volume alone. Analysis for CRL as a predictor of a low birth weight and the analysis with the individual growth curves showed results similar to the original analysis, i.e. no significant or clinically relevant differences between the normal and complicated group. In conclusion, the measurement of the three-dimensional fetal volume in the first trimester of pregnancy is, by itself, not useful for detecting pregnancies at risk for preterm birth and/or low birth weight. The combination with biochemical markers can be subject of future research. If fetal volume measurements appears to be useful after all, then we know that there is no learning curve for the volume measurements with three-dimensional ultrasound and that the inter- and intraobserver reliability of these measurements are good. Further research concerning automated volume measurements or automated detection of the expected fetal shape might be helpful in pregnancy dating and detection of congenital anomalies