Separation of thoracic and abdominal measurements in neonatal vital sign monitoring using radar vision fusion system

Non-contact, rapid and accurate measurement of respiratory rates (RR) and heart rates (HR) in neonates has significant clinical importance. Existing methods predominantly focus on thoracic respiratory signal measurement. This thoracic-focused approach, when applied to neonates who exhibit predominantly abdominal breathing, leads to a low signal-to-interference ratio (SIR) that compromises the accuracy of RR measurement compared with using abdominal signal. Moreover, neonatal and staff motion in clinical environments pose challenges for the robustness of monitoring systems. In this paper, a method for the separation of thoracic and abdominal measurements based on MIMO radar is proposed to make use of the RR information contained within the abdominal signal, while extracting HR information from thoracic signal, in which case RR can be extracted more precisely from abdominal signal with a high SIR. In order to ensure proper separation of radar beams under the interference from neonatal and staff motion, this paper presents an integrated measurement system that combines a monocular camera and MIMO radar to achieve precise and real-time guidance for the radar beams. Experimental results demonstrate significant improvements in RR measurement accuracy and system robustness. We report maximum root mean square errors of 2.16 Beats Per Minute(BPM) for RR measurements and 3.54 BPM for HR measurements.