Cluster-based artificial neural network on ultrasonographic parameters for fetal weight estimation

Accurate estimation of fetal weight and growth measurement is vital to decide the best delivery ways in the prenatal assessment of obstetrics. Regression-based methods using ultrasonographic parameters are generally used in clinic and provide relatively acceptable estimate of fetal weight. Besides, peculiar body figure and macrosomia cause the misjudgment. This study proposed a cluster-based artificial neural network model to improve the accuracy of fetal weight estimation through ultrasonographic parameters. Principal component analysis is adopted to reduce the effects of co-linearity between body figure features. K-means method is used for fetal sizes classification. A cluster based artificial neural network model is proposed for fetal weight estimation. The performance between the proposed model and the regression formula was compared and examined with Friedman test. The results show that the proposed cluster-based ANN model outperformed those of previous models. The results of this study may contribute to a better decision-making over the choices of birth deliveries options, consequently to reduce the possibilities of maternal-fetal morbidity and mortality.