Enhancing uterine contraction detection in electromyometrial imaging based on spatial continuity of the electric field

Preterm birth remains a critical global health challenge, leading to high mortality rates and long-term morbidity. Uterine contractions are a key factor in the onset of preterm labor, making accurate detection of uterine contractions crucial in managing preterm labor risk. Electromyometrial Imaging (EMMI) is a non-invasive imaging technique that provides high-resolution, three-dimensional maps of electrical activity on the uterine surface, offering both spatial and temporal resolution for monitoring uterine contractions. However, existing methods for analyzing EMMI data are susceptible to environmental noise and motion artifacts. This study introduces a novel approach to process EMMI data, effectively addressing the limitations of existing analysis methods. Our approach integrates motion artifact removal and uterine contraction detection using a kernel change point detection method that leverages global information to estimate the number of uterine contractions and local information to preserve the spatial continuity of the electric field. By leveraging the spatial structure of the uterine surface, our approach enhances robustness and accuracy in identifying uterine contractions. Extensive experiments were conducted to evaluate our approach to existing methods using three distinct metrics. The results highlight the superior performance of our approach, making a significant advancement in uterine contraction detection using EMMI data. This improved methodology has the potential to enhance early identification and management of preterm labor risks, contributing to efforts to reduce preterm birth rates and improve neonatal outcomes.