Introducing a New Feature Extraction Method for Non-Contact Blood Pressure Estimating Through iPPG Signals Extracted using G-R Method from Video Images of Different Facial Regions

Abstract

Background and objective: Blood Pressure (BP) is an important physiological parameter of the human body, constant monitoring of which can provide physicians with detailed information on cardiac function. Most methods measure BP using contact-based techniques such as Photoplethysmography (PPG) signal. Recently, a new non-contact PPG signal measurement method has been proposed to extract imaging Photoplethysmography (iPPG) signal through camera and subsequent signal processing algorithms. Methods: This paper proposes a new method for estimating Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) using only the iPPG signal. For this purpose, iPPG signals were first extracted from the right and left cheeks and the forehead through the G-R method to eliminate surface reflections from the skin. The feature vectors were then extracted using the whole-based frequency method to match with the inappropriate shapes of iPPG signals. Finally, linear and nonlinear regression algorithms were employed to estimate BP from the iPPG signal. Results: The proposed algorithm using the G-R method for iPPG signal extraction and the whole-based frequency feature extraction algorithm managed to estimate SBP with a Mean Error (ME) of 0.21 and a Standard Deviation (SD) of 6.77 mmHg. Moreover, the DBP was estimated with a ME of 0.17 and an SD of 5.72 mmHg. Conclusion: The research results show that the proposed method can be used for the continuous and non-contact monitoring of BP. However, more complicated iPPG extraction methods can be employed to obtain BP values from facial videos in motion.