Authors: Aaron M. Zakrzewski, Brian Anthony
Current non-invasive blood pressure estimation technology suffers from many inconveniences; for example, some techniques cut off blood flow, are hard to use, might be inaccurate, and require external calibration. In this poster, an algorithm is developed to estimate blood pressure at the carotid artery in a way that is non-invasive, potentially continuous, easy-to-use, and calibration-free. Inspired by quantitative compression-based elastography methods, this optimization algorithm solves an inverse problem using the Levenberg-Marquardt method. It takes an ultrasound image sequence and applied force as an input and estimates blood pressure, artery hyperelastic stiffness, average background tissue linear stiffness, and artery thickness. By considering the zero pressure condition of the artery, the blood pressure estimation does not require any external calibration. To test the algorithm, a Terason 3000t system is used to acquire ultrasound data at the carotid artery as the applied force slowly increases from 2 N to 8 N. The carotid artery is segmented from the image sequence using an algorithm based on the Kalman Filter. This data is used as input into the blood pressure estimation algorithm, and the resulting percent errors of the algorithm when estimating systole and diastole blood pressures are 6.7 and 7.4 percent, respectively, compared to blood pressure estimated by an automatic blood pressure cuff. The results of the algorithm suggest that ultrasound can be used in a clinical setting to estimate blood pressure.