The purpose of this paper is to outline a novel method of using template based cross-correlation to identify and track uterine contractions during labour. A purpose built six-channel Electromyography (EMG) device was used to collect data from consenting women during labour and birth. A range of templates were constructed for the purpose of identifying and tracking uterine activity when cross-correlated with the EMG signal. Peak finding techniques were applied on the cross-correlated result to simplify and automate the identification and tracking of contractions. The EMG data showed a unique pattern when a woman was contracting with key features of the contraction signal remaining consistent and identifiable across subjects. Contraction profiles across subjects were automatically identified using template based cross-correlation. Synthetic templates from a rectangular function with a duration of between 5 and 10 s performed best at identifying and tracking uterine activity across subjects. The successful application of this technique provides opportunity for both simple and accurate real-time analysis of contraction data while enabling investigations into the application of techniques such as machine learning which could enable automated learning from contraction data as part of real-time monitoring and post analysis.
Keywords: Electromyography; Intrapartum monitoring; Parturition; Pregnancy; Uterine contraction; Uterine monitoring.