The use of a neural network for the ultrasonographic estimation of fetal weight in the macrosomic fetus

doi:10.1016/0002-9378(92)91621-G

American Journal of Obstetrics and Gynecology

Volume 166, Issue 5, May 1992, Pages 1467-1472

https://doi.org/10.1016/0002-9378(92)91621-G Get rights and content

The error associated with regression analysis methods for the ultrasonographic estimation of fetal weight in the suspected macrosomic fetus, approximately 10%, is clinically unacceptable. This study was undertaken to evaluate the applicability of an emerging technique, biologically simulated intelligence, to this problem. One hundred patients with suspected macrosomic fetuses underwent ultrasonographic measurements of biparietal diameter, head and abdominal circumference, femur length, abdominal subcutaneous tissue, and amniotic fluid index. The biologically simulated intelligence model included gestational age, fundal height, age, gravidity, and height. The model was then compared with results obtained from previously published formulas relying on the abdominal circumference and femur length. The biologically simulated intelligence yielded an average error of 4.7% from actual birth weight, statistically better (p = 0.001) than the results obtained from regression models.

References (11)

FP Hadlock et al.
Estimation of fetal weight with the use of head, body, and femur measurements—a prospective study
Am J Obstet Gynecol
(1985)
MJ Shepard et al.
An evaluation of two equations for predicting fetal weight by ultrasound
AmJ Obstet Gynecol
(1982)
Gl Hirata et al.
Ultrasonographic estimation of fetal weight in the clinically macrosomic fetus
AmJ Obstet Gynecol
(1990)
RK Tamura et al.
Real-time ultrasound estimations of weight in fetuses of diabetic gravid women
Am J Obstet Gynecol
(1985)
HD Modanlou et al.
Macrosomia: maternal, fetal, and neonatal complications
Obstet Gynecol
(1980)

There are more references available in the full text version of this article.

Cited by (57)

Using Akaike information criterion and minimum mean square error mode in compensating for ultrasonographic errors for estimation of fetal weight by new operators
2013, Taiwanese Journal of Obstetrics and Gynecology
Citation Excerpt :
Clinical practice has found that the US-based EFW estimation yielded a 10% error for newborns weighing 2–4 kg; and >15% error was observed for newborn babies weighing <2 kg or >4 kg [6]. Recently, using artificial neural network (ANN) approaches to estimate fetal weight has become popular [7–9], and it is more accurate than the traditional regression methods. Farmer et al [7], Chuang et al [8], and Cheng et al [9] have reported that adopting the ANN model could provide more accurate EFW than before.
The accuracy of ultrasound (US) measurements is operator dependent. In order to decrease the operator-dependent errors in estimated fetal weight (EFW), a model selection analysis was undertaken to select significant compensation weighting factors on ultrasonographic parameters to support artificial neural network (ANN), and thus to enhance the accuracy of fetal weight estimation.
In total, 2127 singletons were examined by prenatal US within 3 days before delivery for ANN development, and another 100 cases were selected from new operators for evaluation. First, correlation analysis was used to analyze the differences between the prenatal and postnatal parameters. Second, Akaike information criterion (AIC) was used to determine the number of database partition and optimal weightings for compensating the input parameters of the ANN model. Finally, minimum mean squared error (MMSE) mode was utilized to determine the optimal EFW.
EFW of the proposed compensation model using AIC and MMSE showed mean absolute percent error of 5.1 ± 3.1% and mean absolute error of 158.9 ± 96.2 g. When comparing the accuracy of EFW, our model using AIC and MMSE was superior to those conventional EFW formulas (all p < 0.05).
We proved that performing the parameter compensation (by AIC) and model compensations (by MMSE) for the ANN model can improve EFW accuracy. Our AIC–MMSE model of EFW will contribute to the improvement of accuracy when adding new US datasets measured by new operators.
Efficient fetal size classification combined with artificial neural network for estimation of fetal weight
2012, Taiwanese Journal of Obstetrics and Gynecology
Citation Excerpt :
Recently, it was shown that estimating fetal weight using ANN models was more accurate as compared with regression methods. Nevertheless, Farmer et al [13] focused mainly on macrosomia and did not cover the entire range of fetal weight. Chuang et al [14] reported another ANN model of EFW, covering 200–4400 g. Although their results showed ANN model could provide more accurate EFW than before, their EFW was less accurate in actual fetal body weights below 2500 g and above 4000 g.
A novel analysis was undertaken to select a significant ultrasonographic parameter (USP) for classifying fetuses to support artificial neural network (ANN), and thus to enhance the accuracy of fetal weight estimation.
In total, 2127 singletons were examined by prenatal ultrasound within 3 days before delivery. First, correlation analysis was used to determine a significant USP for fetal grouping. Second, K-means algorithm was utilized for fetal size classification based on the selected USP. Finally, stepwise regression analysis was used to examine input parameters of the ANN model.
The estimated fetal weight (EFW) of the new model showed mean absolute percent error (MAPE) of 5.26 ± 4.14% and mean absolute error (MAE) of 157.91 ± 119.90 g. Comparison of EFW accuracy showed that the new model significantly outperformed the commonly-used EFW formulas (all p < 0.05).
We proved the importance of choosing a specific grouping parameter for ANN to improve EFW accuracy.
Prediction of risk for cesarean delivery in term nulliparas: a comparison of neural network and multiple logistic regression models
2009, American Journal of Obstetrics and Gynecology
We sought to develop a neural network (NN) to predict the risk for cesarean delivery (CD) in term nulliparas.
Using software (BrainMaker for Windows, Version 3.0; California Scientific Software, Nevada City, CA), we trained an NN with 225 patients obtained by chart review and included for nulliparity, singleton vertex > 36 weeks' gestation, and reassuring fetal heart rate on admission. Training inputs included several maternal and fetal clinical variables. Two logistic regression (LR) models using 225 and 600 patients (LR225 and LR600, respectively) were developed. The NN and LR models were tested for prediction of CD in a set of 100 patients not used for development.
The NN, LR225, and LR600 correctly predicted 53%, 26%, and 32% of the patients with CD and 88%, 95%, and 95% of the patients with vaginal delivery, respectively.
Compared with LRs, the NN was slightly better in predicting CD and was similar for predicting vaginal delivery in nulliparas with term singletons.
Fetal weight estimation based on deep neural network: a retrospective observational study
2023, BMC Pregnancy and Childbirth
A Systematic Review on The Applications of Machine Learning for Fetal Birth Weight Prediction
2023, Research Square
Fetal biometric based abnormality detection during prenatal development using deep learning techniques
2022, Multidimensional Systems and Signal Processing

View all citing articles on Scopus

¹: Presented at the Eleventh Annual Meeting of the Society of Perinatal Obstetricians, San Francisco, California, January 28–February 2, 1991.

^a: From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Southern California School of Medicine, Los Angeles County-University of Southern California Medical Center, Women's Hospital.

View full text

The use of a neural network for the ultrasonographic estimation of fetal weight in the macrosomic fetus1

Am J Obstet Gynecol

AmJ Obstet Gynecol

AmJ Obstet Gynecol

Am J Obstet Gynecol

Macrosomia: maternal, fetal, and neonatal complications

Obstet Gynecol