Sensing Technologies for Ambulatory Blood Pressure Monitoring using Pulse Transit Time
Elevated Blood Pressure (BP) affects more than a quarter of the world’s population and is considered to be a major risk factor for cardiovascular disease. Compared to BP measurements in the presence of a physician, BP values measured over 24 hours are now considered to be more accurate predictors of adverse cardiovascular events such as stroke or end-stage kidney disease. Most devices monitor BP using an occlusive cuff on the left arm. The cuff is obtrusive and can affect the patient’s daily routine, especially his / her sleep. Pulse Arrival Time (PAT) and Pulse Transit Time (PTT) have emerged as alternate surrogates of monitoring BP in a cuffless manner. Typically, the sensor architectures proposed in the literature measure the Electrocardiogram (ECG) as well as the arterial pulse wave at the finger or wrist using photoplethysmography, where the pulse arrival time is the time difference between the ECG R-peak and a fiducial point on the photoplethysmographic signal. The pulse arrival time is mapped onto BP using a non-linear function. The accuracy of an ambulatory blood pressure monitoring system is limited by several challenges, such as changes in vascular tone of the peripheral arteries as well as the Pre-ejection Period (PEP) of the heart, which is affected by factors such as body posture. Alternative locations for the measurement of the pulse wave signal are required in order to increase the BP monitor’s accuracy, while reducing the frequency of re-calibration of PAT / PTT to a cuff-based BP monitor. The core of this thesis is the design, development and experimental validation of new sensor and system architectures for ambulatory blood pressure monitoring and Systolic Time Intervals (STI) measurement. These are validated using in-house electronic designs with commercial off-the shelf components. The data collection is performed on healthy human males. The research includes developing and validating the sensors for arterial pulse measurement at the carotid and subclavian arteries, the aorta and the left ventricle of the heart using continuous wave radar, electrical bio-impedance and piezoelectric sensors. Furthermore, the application of these signals for PAT, PTT and STI estimation is investigated. From ten human subjects on an exercise bike, we have acquired the PAT and PTT from the ECG, the Impedance Cardiogram (ICG) and the carotid arterial pulse at the neck using continuous wave radar. We have found that the correlation coefficient between systolic blood pressure and the carotid PAT was -0.69 (p=0.001), which is in reasonable agreement with finger PAT from previous literature. In a second investigation, we have developed and tested a new architecture for PTT estimation at the carotid and subclavian arteries using ECG, electrical bio-impedance, Continuous Wave (CW) radar from the sternum. The PAT and PTT were measured on six healthy male subjects during exercise on a bicycle ergometer. For all subjects, the Pearson correlation coefficients for PAT-Systolic BP and PTT- systolic BP were -0.66(p=0.001) and -0.48 (p=0.0029). Correlation coefficients for individual subjects ranged from -0.54 to -0.9 and -0.37 to -0.95 respectively. Another contribution of this thesis is a feasibility study on the estimation of STIs using continuous wave radar at three locations of the thorax. Particular attention is paid to the effect of antenna placement as well as the radar signal processing to acquire the motion of the aorta and left ventricle in order to increase the signal reproducibility across a respiration cycle and subjects. In a third investigation, the estimation of STIs was done using CW radar at 2.45GHz with a body- contact antenna. Ten healthy male subjects aged 25-45 were measured at 30 degree incline from the supine position. 60-second recordings were taken without breathing and with paced breathing. Phonocardiogram (PCG), ECG, respiration and ICG were measured simultaneously as reference signals. The radar antennas were placed at locations corresponding to Wilson’s ECG lead positions V1, V4 and V6. The results indicate that the position near Wilson’s lead V1 gives the most reproducible signals within an individual’s respiration cycle. The final contribution of this thesis is the design and development of a frequency-sensing electronic readout circuit for low-voltage signals from a piezoelectric sensor. Applications are the arterial pulse signal, the Ballisto- or Seismocardiogram or the PCG. The proposed circuit consists of a Colpitts oscillator for voltage to frequency conversion, and a commercial Phase Locked Loop for frequency to voltage conversion. For the frequency sensing readout, we show that the noise levels in frequency sensing can be reduced by increasing the oscillation frequency, while maintaining 1% non-linearity. The results of this work are expected to contribute towards low noise analog front end designs for piezoelectric sensors using frequency sensing as an alternate architecture.
Read more
2
