Possible improvement and extension to other physiological signals 

Get Complete Project Material File(s) Now! »

Factors aecting PPG waveform

PPG waveform could be characterized by several features. However, it should be noted that those features could variate from one subject to another or for the same subject due to health conditions. Arteriosclerosis, hypertension and some dermatoses are some diseases that aect the PPG waveform.
In addition to health conditions, PPG waveform variates with age and gender. In [13], a study analysis how age aects the dicrotic notch and the PTT. In fact, the notch amplitudes are larger for older subjects than younger ones. Also, experiments has shown that PTT is higher in the female group than in the male group. Vasoconstriction and vasodilatation are physiological mechanisms corresponding to the decrease, inversely the increase, in the diameter of the blood vessels. Those phenomena aect imperatively the PPG amplitude. In [14], an example of how PPG amplitude increases after infusion of vasodilator Nipride.

PPG mathematical model

Mathematical modeling of the PPG waveform is the center of interest of many researchers. Modeling helps researchers to assess algorithms for PPG processing. In the literature, there are mainly two approaches for PPG modeling, depending on the study purpose: temporal modeling and shape modeling.

Modeling based on respiratory modulation

As described above, PPG is modulated by three signals induced by respiratory activity. Using this information, PPG can be simulated as a sinusoidal signal shifted by the baseline with a variable amplitude and frequency. x(t) = b(t) + a(t)cos(2fHR(t)) where a(t), b(t) and fHR(t) refer to AM, BW and FM modulation depending on the respiratory rate fRR.

Limits of mathematical models

PPG mathematical modeling is a tool for researchers to assess algorithms. But, the question remains to what extent these models are close to PPG acquired in real conditions. In fact, model based on respiratory modulation does not take PPG pulse shape parameters into account, which limits its use to only assessing RR estimation algorithm.
The studies around the temporal model of the pulse, ie how the parameters vary over time and how they are aected by respiratory activity, remain limited. In [15], authors combine both shape modeling and temporal modeling. A Gaussian model is applied to parametrize pulse shape and autoregressive moving average is applied for modeling temporal behavior of each pulse shape parameter. This method yields good results to synthesize missing segments from PPG signals and to derive probabilistic distributions of pulse shape parameter. However, the relationship between respiratory modulation, temporal pulse shape parameter evolution and patient conditions is still ambiguous. In fact, these models do not allow the tracking of the respiratory activity. PPG model should include other parameters relative to subject conditions such as age and gender and combine both respiration and shape modeling. To our knowledge, such models are not studied yet.

PPG for clinical physiological monitoring

PPG has widespread uses in many clinical settings. The main direct application is the measurement of blood oxygen saturation by pulse oximeter. In the following, we will focus on the monitoring of blood oxygen saturation and other subordinate applications for PPG signal.

READ  Experiments on Remote Sensing Time Series datasets

Monitoring blood oxygen saturation

As described in Section 2.2.1, pulse oximeters are based on absorption dierences between Red (R) and infrared (IR) light waves. The photodetector continuously analyzes the R / IR ratio. From the equivalence Table 2.1 between this R / IR ratio and calibration values of oxygen saturation, the monitor displays the value of the measured oxygen saturation. The correspondence between the values of R/IR and those of SpO2 reported in Table 2.1 are obtained from a calibration algorithm. In fact, the pairs (R/IR,SpO2) with SpO2 measured between 75 and 100% are obtained from experiments on healthy volunteers. For SpO2 below 75%, the displayed values are obtained by extrapolation of the data between 75 and 100% [1].

Table of contents :

1 Introduction 
1.1 Context
1.2 Motivation
1.3 Objectives
1.4 Outline of the thesis
2 The photoplethysmography signal 
2.1 Introduction
2.2 PPG measuring characteristics
2.2.1 Technical facts about PPG
2.2.2 Sites and devices for measuring PPG
2.2.3 Measurement protocol and reproducibility
2.3 PPG waveform characteristics
2.3.1 Pulse characteristics
2.3.2 PPG waveform modulations
2.3.3 Factors aecting PPG waveform
2.3.4 PPG mathematical model
2.4 PPG for clinical physiological monitoring
2.4.1 Monitoring blood oxygen saturation
2.4.2 Monitoring heart activity
2.4.3 Monitoring respiratory activity
2.4.4 Monitoring hypovolemia
2.5 PPG database
2.5.1 Sukor Data
2.5.2 CapnoBase Data
2.5.3 ReaStoc data
2.6 Conclusion
3 Detection of artifacts in PPG signal 
3.1 Introduction
3.2 Artifact causes and impact
3.3 State of the art
3.3.1 Filtering method with PPG restoration
3.3.2 Morphology analysis and artifact detection method
3.4 RDT for artifact detection
3.4.1 Simple artifact detection for short records
3.4.2 Adaptive RDT for artifact detection
3.5 Results and discussion
3.5.1 Simple RDT detection performance
3.5.2 Adaptive RDT performance
3.5.3 Possible improvement and extension to other physiological signals
3.6 Conclusion
4 Respiratory rate estimation from PPG 
4.1 Introduction
4.2 State of art
4.2.1 RR estimation from raw PPG
4.2.2 RR estimation from derived PPG signals
4.2.3 General limits of the existing methods
4.3 Consensus spectrum for RR estimation
4.4 RR from PPG modulations
4.4.1 Extracting PPG modulations
4.4.2 Extraction of respiratory modulations
4.4.3 RR estimation
4.5 Results
4.5.1 Results on Capnobase
4.5.2 Results on Reastoc
4.6 Discussion
4.6.1 Comparison between the proposed algorithms
4.6.2 Age impact on algorithms performance
4.6.3 Comparison with others methods
4.7 Conclusion
5 Optimization of respiratory rate monitoring from PPG 
5.1 Introduction
5.2 Artifact detection impact on respiratory rate estimation
5.2.1 Motivation
5.2.2 Analysis methodology
5.2.3 Results on Capnobase
5.2.4 Results on Reastoc
5.2.5 Limits of artifact detection
5.3 SRQI impact on respiratory rate estimation
5.3.1 SRQI denition
5.3.2 Analysis methodology
5.3.3 Results on Capnobase
5.3.4 Results on Reastoc
5.4 Discussion
5.5 Conclusion
6 Conclusion and perspectives 
6.1 Summary and main contributions
6.2 Perspectives and future works


Related Posts