Contextually-aware Fetal Sensing in Transabdominal Fetal Pulse Oximetry

Transabdominal fetal pulse oximetry (TFO) is a noninvasive technique that can provide physicians with a convenient measure of fetal oxygen saturation. This is accomplished by sending a known light intensity signal towards the mother's abdomen, where it is modified by the maternal and fetal tissues, and observed some distance away. The measured signal, captured by a photodetector, contains a mixture of both maternal and fetal information, where the fetal portion must be extracted to calculate the fetal oxygen saturation. However, the ability to decouple the maternal and fetal components is highly dependent on the physiological and structural parameters of the physical system, making it difficult to robustly extract the fetal signal across patients over a long-period of time. In this work, we propose a contextually-aware sensing approach that utilizes additional information about the physical system (physiological, spatial, and temporal) to extract the fetal signal. It does this by using easily-measurable parameters of the mother's physiology to reduce the maternal impact, incorporating data fusion techniques to combine spatial information from multiple detectors, and utilizing historical data points to improve and validate the fetal signal estimates. The efficacy of the proposed approach is supported by experimental evaluation using in vivo measurements captured on pregnant sheep.