Date of Award

January 2016

Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

Kirk Shelley

Abstract

The study of plethysmographic physiology has been limited by a lack of standardized and open plethysmographic hardware in clinical use. Fundamental differences in the processing of output between various devices obfuscate direct comparison, and the role played by physiology versus that by technology in the final viewable plethysmogram (PPG). This study proposes a largely automated, unbiased method for quantitatively comparing the outputs from proprietary pulse oximeter devices along three metrics: temporal delay, amplification, and complexity. It then applies these methods to the deconstruction of Masimo and Nellcor pulse oximeters.

With IRB approval, twelve healthy, awake subjects were studied. Each individual was attached simultaneously to a Nellcor ear probe and a Masimo finger device, and then instructed to perform incentive spirometry, Valsalva, and Mueller breathing maneuvers interspersed with normal breathing. For temporal delay and amplitude comparisons, the raw PPG data were first synchronized, then subsequently filtered into corresponding autonomic, respiratory, and cardiac frequency ranges. To assess the temporal delay, they were processed according to a sliding-window cross-correlation function, and the time shift of maximum correlation for each window was averaged, to determine a representative overall delay for each frequency range. For the amplitude analysis, the absolute value of the filtered data were integrated over a pre-determined time frame chosen at each frequency range, then divided to arrive at a ratio. These data were manually filtered to remove sequences corresponding to the noise artifact. Lastly, to assess pulse complexity, the raw data were converted from time-domain to frequency domain using digital Fast Fourier Transformation (dFFT), and an algorithm programmed to search for fundamental cardiac frequency, as well as the first five harmonic peaks. The dFFTs were then normalized according to fundamental frequency peak, and the ratios of the amplitude of each harmonic peak to the corresponding harmonic peak from the other device were generated.

As outlined in this study, the Nellcor device temporally led the Masimo in the respiratory (p= 0.0338) and autonomic (p=0.0024) frequency ranges. Similarly, the Nellcor device demonstrated greater amplitude representation in those ranges as well (p<0.001). With regards to pulse complexity, however, the Masimo signal was better represented up to the first three harmonics (p<0.04). While the generalization of these results may be limited by the device placement, this study successfully presents a systematic method for comparing commercial hardware devices, paving the way for better understanding of this non-invasive modality.

Comments

This is an Open Access Thesis.

Share

COinS