Arterial blood pressure (BP) and pulmonary arterial pressure (PAP) are fundamental for diagnosis and management of both systemic and pulmonary hypertension and for monitoring of surgical and critically ill patients. Systemic hypertension is the most common modifiable risk factor for cardiovascular disease and the leading contributor to mortality and disability in the world. Pulmonary hypertension is a group of pulmonary vascular disorders leading to increased PAP, right ventricular failure, and death. It is also common in the intensive care unit. Diagnosis of pulmonary hypertension remains challenging.

Arterial BP is most commonly monitored with the non-invasive cuff-based sphygmomanometer, which only outputs peripheral systolic and diastolic BP, differs from the critical central BP, and is prone to errors in the presence of arrhythmias. Instead of providing a continuous measurement, its BP values are averaged over many pulses. Invasive catheterization allows for direct peripheral arterial BP monitoring, but values are still not central and the technique is not risk-free. Noninvasive estimates of PAP from echocardiography, computed tomography (CT) scan and magnetic resonance imaging (MRI), although useful, remain variable and operator-dependent. Therefore, right heart catheterization is still the gold-standard diagnosis for PAP and pulmonary hypertension, despite being highly invasive. Monitoring of PAP at home or primary care is currently infeasible.

The overarching aim of this proposal is to evaluate whether a non-invasive radio-frequency (RF) sensor can retrieve central BP transients accurately and non-invasively. We hypothesize that the near-field coherent sensing (NCS) by RF carriers, which has been benchmarked on heartbeat and respiration waveforms with the gold-standard devices, can also be applied to derive central BP from the vibration characteristics of the aorta and pulmonary arteries in the entire cardiac cycle. The principle of operation is similar to that of the ground penetrating radar and air-puff tonometry, where Hilbert-Huang Transforms and machine learning can be adapted for BP signal processing.

Our effort will start from building a phantom heart model, which will allow direct pumping control and host the NCS and catheter-based pressure sensors. We will also explore the multiple-input-multiple-output (MIMO) NCS to improve the local mapping of vibration characteristics. We will use instrumented and anesthetized pigs as animal models for aortic BP and PAP studies, where similar sensors from the phantom will be deployed. The accuracy of the sensors and their ability to track changes in relation to the gold-standard pressure catheters will be benchmarked. Aortic BP studies will be modified by infusion of inotropes and vasopressors, while PAP by acutely changing the fraction of inspired oxygen (FiO2) and by inducing surfactant depletion followed by lung recruitment. Additionally, electrocardiogram (ECG) and photo-plethysmography (PPG) will be synchronized on the animals to investigate whether noninvasive calibration of the NCS BP readout can be realized.