Effects of fetal head shape variation on the second stage of labour.

Fetal head geometry plays an important role in the mechanics of childbirth during the second stage labour. Large heads have been shown to be associated with difficult and prolonged childbirth. However, the relationship between the fetal head geometry and childbirth mechanics has not been quantitatively analysed. To address this, our study used finite element (FE) modelling techniques and biomechanical simulations to analyse the contribution of fetal head shape and size on the mechanics of childbirth. X-ray computed tomography (CT) images from 26 newborn infants (less than 9 days old) without skull abnormalities were used to construct individual-specific FE models of the fetal skull. Simulations of childbirth were conducted using each model of the skull and a customised pelvic floor model based on magnetic resonance imaging (MRI) of a healthy nulliparous woman. The force required for delivery, the maximum principal stresses, and the maximum principal stretch ratios at the left and right pelvic floor muscle-pubic bone interfaces were quantified. Partial least squares regression (PLSR) models for predicting these mechanical indices were constructed using: (i) either the FE geometries of the fetal heads or the biometrical parameters (biparietal diameters and fetal head circumferences) as inputs; and (ii) either a linear or a quadratic function for the inner relation. The predictabilities of the mechanical indices using the PLSR models were quantified using a leave-one-out analysis. Quantitative associations were found between the geometric parameters of the fetal head and the indices of childbirth mechanics. When using the full FE geometries as inputs, the PLSR model using a linear inner relation gave better predictability than the model using a quadratic inner relation. This could be attributed to the quadratic inner relation correlating response to the noise in point-to-point correspondence. When using the biometrical parameters of the skull as inputs, the PLSR model using a quadratic inner relation gave the best overall predictability. Such a model could be implemented in a clinical setting as a predictive model for childbirth planning and as an educational tool for clinical training.