Development of non-invasive, optical methods for central cardiovascular and blood chemistry monitoring.

Abstract

Cardiovascular disease and sepsis are leading causes of mortality, morbidity and high cost in hospitals around the world. Failure of the circulatory system during cardiogenic shock and sepsis both can significantly impair the perfusion of oxygen through organs, resulting in poor patient outcome if not detected and corrected early.

Another common disorder which goes hand-in-hand with cardiovascular disease is Diabetes Mellitus. Diabetes is a metabolic disorder resulting from the inability of the body to regulate the level of glucose in the blood. The prevalence of diabetes worldwide is increasing faster than society’s ability to manage cost effectively, with an estimated 9% of the world population diagnosed with metabolic disease.

The current gold standard measurements for venous oxygen saturation, arterial pulse wave velocity (PWV), and diabetes management through blood glucose concentration monitoring are all invasive. Invasive measurements increase risk of infection and com- plications, are often high cost and disposable, and have a low patient compliance to regular measurements.

The aim of this thesis is to develop non-invasive methods of monitoring these important dynamic physiological variables, including, venous oxygen saturation, pulse wave velocity, and blood glucose concentration. A novel photoplethysmography-based NIR discrete wavelength spectrometer was developed using LEDs to both emit light, and detect the light reflected back through the tissue. Using LEDs to detect light simplifies sensing circuit design, lowering hardware costs, allowing adaptable sensing specific to the needs of the user.

A reflectance pulse oximeter was developed to measure the oxygen saturation at both the external jugular vein, and carotid artery. Measuring the jugular venous pulse (JVP) allows estimation of the venous oxygen saturation through either the JVP, or through breathing induced variation of the JVP. In addition to oxygenation, the de- vice developed is capable of adapting the sensing layout to measure the arterial pulse waveform at multiple sites along a peripheral artery, such as the carotid or radial. The PWV local to the carotid artery, and radial artery can then be measured, providing key information of cardiovascular risk. A novel algorithm for PWV measurement over multiple pulse waveforms was also developed. Expanding the sensor to use multiple different wavelength LEDs allow discrete spectroscopy in pulsatile blood. An absorption model of components in blood at specific wavelengths was created to isolate the spectral fingerprint of glucose. The sensor successfully measured the oxygen saturation at the carotid artery, and external jugular vein across 15 subjects, giving mean oxygen saturations of 92% and 85% respectively, within the expected physiological ranges. Venous oxygen saturation calculated using breathing induced changes to JVP was 3.3% less than when calculated on the JVP alone, with a standard deviation of 5.3%, compared to 6.9%. Thus, future work on the sensor will focus on extraction of the breathing induced venous pulse, rather than measuring from the JVP itself. The PWV on the carotid and radial artery was successfully measured within the ex- pected physiological ranges, with the novel phase difference algorithm proving more robust to noise than the gold standard foot-foot method. The phase difference method returned a mean PWV at the radial artery of 4.7 ±0.6 m s−1, and a mean CoV of 20%, compared to 4.0 ±1.4 m s−1, and a moan CoV of 58% for the foot-foot method. The proof of concept PWV sensor gives promising results, but needs to be calibrated against invasive gold standards, such as aorta and femoral pressure catheters.

A glucose trial involving adult and neonatal subjects provided validation of the NIR non-invasive pulse glucometer. The sensor has an R2 of 0.47, and a mean absolute relative difference (MARD) of 19% compared to gold standard reference measurements.

Clarke error grid analysis returns 85% of measurements in Zone A, 11% in Zone B, and 4% in Zone C. While the sensor is not as accurate as the gold standard invasive measurements, the ability to constantly measure without any pain or discomfort will help increase measurement compliance, improving user quality of life, plus further development may improve this.

Overall, this thesis provided novel contributions in non-invasive venous oxygen saturation, PWV, and glucose concentration monitoring. The adaptability of the sensor shows promise in helping reduce the pain and inconvenience of the current invasive measurements, especially in diabetes management, where the sensor has the most potential for impact.