An attempt to construct a general model for resting blood pressure estimation using independent components of facial thermal images

Abstract

The increasing number of people suffering from hypertension has been a problem worldwide. Noncontact measurement for blood pressure is expected to help monitor blood pressure daily and control the disease. The technique of noncontact blood pressure measurement using facial thermal image has been studied. A individual model performance for resting blood pressure estimation from facial thermal images was evaluated in the previous study. The resting blood pressure could be estimated with the root mean square error of 5.73 mmHg between the measured and estimated values using the individual model. A general model of resting blood pressure estimation needs to be constructed for practical use. The objective of this study is a construction of general models for resting blood pressure estimation from facial thermal images. As a result, skin temperature of the nasal region was might be related to the resting blood pressure and a common region among subjects. The accuracy of resting blood pressure estimation using a general model was root mean square error of 7.73 mmHg.