Cardiopulmonary monitoring is of critical importance in a variety of clinical and non-clinical applications ranging from monitoring physiological conditions of crew members during space missions to emotion and stress recognition in applications involving human-machine interaction. Current solutions involve attaching gel-based electrodes for electrocardiogram (ECG) monitoring and pulse oximetry sensors connected to fingertips or earlobes for photoplethysmography (PPG) monitoring. Gel-based electrodes require preparation and their application can cause skin irritation. In addition, the use of current contact-based solutions is further complicated by the fact that a relatively large device such as a Holter monitor has to be carried by the subject at all times. Wearable sensors are a step in the right direction, yet the sensor needs to be continuously worn (on the wrist, chest, etc.) by the subject.
We propose to build on our prior research experience in non-invasive remote cardiopulmonary monitoring as well as computer vision and machine learning to develop a non-invasive cardiopulmonary monitoring system and extract clinically important information from multiple subjects in the field of view. Specifically, our proposed sensing framework involves i) an optical camera; ii) a depth-sensing camera, iii) a Doppler radar-based solution; and iv) a sensor fusion component for integration of data received by multiple sensing modalities.