(An) evaluation of cardiovascular system using photoplethysmography

다국어 초록 (Multilingual Abstract)

This thesis consists of two major aims: signal characteristic analysis of PPG and physiological analysis of PPG applications. The first aim, signal characteristic analysis of PPG, contains three objectives: harmonic components analysis based on spectrum analysis, frequency filtering of PPG using analytic results of harmonic analysis and feature detection using adaptive threshold detection method.

1. In the characteristic analysis of PPG signal, numerous statistical simulations show that the spectral characteristics of PPG coincide with those of blood pressure pulse. It can be seen that the first dominant harmonics exhibits 0.83 of morphological correlation coefficient and consists 33.97 % of total power.

2. Innovative schemes are proposed in PPG analysis by utilizing the characteristics of the combination of standing wave and harmonics. In standing wave analysis, the minimal time duration becomes nearly twice of the average heart beat interval, which enables us to produce more sophisticated analysis and utilize similarities of conventional research outcomes based on blood pressure analysis as well. Moreover, the distinguishable characteristic of PPG spectrum plays an important role in frequency domain filtering, which makes it clear to separate between DC component and respiratory component of PPG waveform.
3. It is proposed that innovative signal processing algorithm and an adaptive threshold peak detection scheme which has detection rate in PPG-peak and PPG-foot by 98.24 % and 98.92 %, respectively. Proposed algorithm improves detection rate in PPG-peak and PPG-foot detection by 15.10 % and 11.40 % compared with the conventional local maxima/minima detection method, respectively.

The second aim of this thesis, physiological investigation based on analytic study of PPG, includes interrelation analysis between blood pressure and PPG, significance validation of PPG variability as an ANS evaluation index, and investigation of respiratory mechanism by comparing respiratory components extracted from ECG and PPG. This thesis employs novel approaches which based on previous blood pressure analytic methods such as pressure-flow relations, pulse transit time, pulse wave velocity, because there are many previous researches in related field. Proposed schemes are applied to analyze blood pressure variation, assess of ANS, and investigate respiratory mechanism.

1. To investigate analytical relations between PPG and BP, the derivatives of PPG are used. Pressure-flow relations and vascular movement model are also designed by analyzing of each functional characteristics and sectional information. Pressure Index (PI), as a normalized pressure-related term, is derived with the results of morphological analysis of PPG and vascular modeling. It is appeared that the proposed PI is significantly (p<0.01) correlated with systolic blood pressure (r=0.950), diastolic blood pressure (r=0.637), pulse pressure (r=0.896), mean arterial pressure (r=0.883), which proves the appropriateness of the proposed approach.

2. In ANS assessment, both ECG-derived HRV and PPG-derived PRV are compared with respect to PWV indices which related to the sympathetic activity. As a result, PRV represents the high correlations (r=0.831, p<0.01), close to the correlation of HRV (r=0.734, p<0.01) with sympathetic index, which indicates that PPG could be an effective surrogate method to assess ANS function as well as HRV.

3. Respiratory drive mechanism is investigated by transfer function analysis using ECG- and PPG-derived respiratory components. Consequently, significantly correlated (magnitude squared coherence estimate, MSCE > 0.5) respiratory components are observed in both derived respiratory components (ECG and PPG), and the phase analysis is the same as that of known as physiological respiratory drive.
The results in this thesis are based on the analysis of offline clinical data from non-invasive measurement, and any invasive measurements or pharmacological tests are excluded throughout this study. This thesis only deals with indirect methods for measurement of physiological signals, and carries out the technological assessment and the physiological investigations. Further study based on the direct measurement and analysis such as continuous, invasive measurement, pharmacological test employing extended subject groups would be meaningful and beneficial to achieve more precise and reliable results. Composite analysis of both multi-site measured PPG and multiple sensor inputs are also expected to improve the accuracy of the PPG analysis and innovative physiological investigation.