Power-efficient and high-performance interference resilientpulse oximeter

This thesis presents a novel sigma delta-based pulse oximeter with higher resilience against ambient light interference in comparison to the conventional pulse oximeters available in the market. The pulse oximeter reported in this thesis uses a novel cross-coupled sigma delta modulator to drive the pulse oximeter LEDs with two chaotic, non-overlapping pulse density modulated sequences. The deployment of this modulator not only provides more immunity against ambient light but it also offers more control over the pulse oximeter LEDs power consumption resulting in less power consumption under low ambient light conditions.

Additionally, this thesis suggests a novel trans-impedance amplifier (TIA) topology for better detection of the physiological signal under low perfusion state and weak heart pulses. The TIA increases the physiological signal power by partially removing its DC component and providing enough headroom for the amplification of the AC physiological signal. This AC physiological signal is used to calculate the level of blood oxygenation and its accurate detection has a direct impact on the accuracy of the overall pulse oximeter measurement.
The thesis also offers a detailed description about the computer model of the pulse oximeter developed in Simulink. This model is the first computer model of pulse oximeter and it is a useful tool for design and development of a pulse oximeter. Comparing the hardware measurement results and the Simulink simulation results presented in Chapter 2 and 5 reflects the capability of the model in predicting the behavior of pulse oximeter under different measurement conditions. The thesis also includes a detailed explanation about our flexible PC-based pulse oximeter hardware prototype which was used as a reference to evaluate the performance of our novel sigma delta-based pulse oximeter. This PC-based conventional pulse oximeter prototype performs the entire signal processing tasks within the Matlab environment and therefore it can be individually used as a platform to design and evaluate the new signal processing algorithms designed for pulse oximeter application.
Finally the thesis reports on the standalone FPGA based fixed point implementation of the novel pulse oximeter processing chain on a Xilinx Spartan 3 FPGA Fabric, with real-time measurement results compared and contrasted against the results obtained from the Matlab based processing engine.