Camera-Based Blood Pressure Monitoring Based on Multisite and Multiwavelength Pulse Transit Time Features
Yukai Huang,Dongmin Huang,Jia Huang,Guowei Wang,Liping Pan,Hongzhou Lu,Min He,Wenjin Wang
DOI: https://doi.org/10.1109/tim.2024.3457944
IF: 5.6
2024-01-01
IEEE Transactions on Instrumentation and Measurement
Abstract:Pulse transit time (PTT) is a promising physiological feature for camera-based blood pressure (BP) monitoring. Existing approaches of PTT-based camera-BP estimation can be categorized into multisite PTT (MS-PTT) based and multiwavelength PTT (MV-PTT) based, the former measures PTT between arteries at different skin locations while the latter measures PTT between arteries and arterioles at different skin layers. These two types of PTT features reflect different vascular properties and thus their fusion can provide richer information for more accurate BP calibration. We therefore propose a novel multidimensional PTT (MD-PTT) feature set that integrates MS-PTT and MV-PTT with spatial anatomical constraints for camera-BP monitoring. An end-to-end video-processing framework is developed for robust MD-PTT feature extraction and BP calibration. The proposed method was validated in both the lab setting and hospital intensive care unit (ICU) setting. The lab experiment shows that MD-PTT reduced the mean absolute error (MAE) of systolic BP (SBP), diastolic BP (DBP), and mean BP (MBP) by at most 1.81, 1.93, and 1.61 mmHg against MV-PTT, and by at most 2.54, 2.46, and 2.92 mmHg against MS-PTT. The clinical trial in ICU shows that the MAE of MD-PTT with generalized modeling on three types of BP were 6.49, 6.82, and 5.98 mmHg, with personalized modeling were 4.78, 5.52, and 4.52 mmHg, respectively. The results were in the acceptance range of the British hypertension society (BHS) standards. The clinical evaluation shows that the novel MD-PTT feature may enable accurate BP estimation for camera-based patient monitoring.