Detecting Irregular Cardiac Activity using Video Plethysmography
Gill R. Tsouri, Rochester Institute of Technology (RIT)
PhotoPlethysmoGraphy (PPG) is used for the assessment of cardiovascular function. The PPG sensor attaches to the skin, shines a light and captures the intensity of light being reflected back or transmitted through it to produce a signal representing the pulsating heart. Very recently, the FDA has approved the use of PPG sensors for identification of irregular cardiac activities in products such as the Apple Watch. Despite its popularity, PPG sensors have limitations when long-term monitoring in required. Devices embedding such sensors are expensive and attaching them to the skin can create discomfort. Furthermore, the monitoring process depends on participation of the subjects being monitored.
Over the past decade, various groups experienced with the feasibility of performing a PPG measurement from ambient light reflected off the skin using a camera as the PPG sensor. This approach is often called remote-PPG or Videoplethysmography (VPG). VPG is an attractive approach to acquiring a pulsatile signal, since it requires no physical contact with the subject being monitored. It can therefore operate without burdening the subject with wearable devices and the need to participate in the monitoring process. An additional consequence is that compliance with long term monitoring is achieved. The increasing proliferation of smart devices with embedded front cameras, such as tablets, smartphones and laptops enables remote-PPG based applications even further. Potential applications range from monitoring various health conditions of subjects to inferring their stress levels based on vagal tone activity. Extracting a reliable pulsatile signal using remote-PPG is challenging. The light reaching the camera is much weaker than the light captured by sensors attached to the skin, thus forcing signal extraction to operate under low signal to ratio conditions. In addition, subject motion and changes in ambient light intensity corrupt the extracted pulsatile signals.
This talk would present recent results stemming from an ongoing project funded by the National Institute of Health exploring the possibility of detecting arrhythmias such as Atrial Fibrillation (AF) using VPG. AF is a cardiac condition currently affecting over 33 Million people in the world with over 2 Million additional people diagnosed every year.