본문 바로가기 주메뉴 바로가기
국회도서관 홈으로 정보검색 소장정보 검색

결과 내 검색

동의어 포함

목차보기

Title Page

Abstract

Contents

Abbreviations and Variables 15

Chapter Ⅰ. Introduction 17

Chapter Ⅱ. Technical Challenges 22

2.1. Working Principles of IPG 22

2.2. Blood Pressure Calibration Methods 25

2.2.1. Energy Conservation Equation 25

2.2.2. Moens-Korteweg equation 28

2.2.3. Bramwell-Hill equation 29

2.2.4. Summary 31

Chapter Ⅲ. Background knowledge for designing the Bipolar IPG sensor 32

3.1. Need for the bipolar IPG sensor 32

3.2. Antenna principles for the bipolar IPG sensor 33

3.2.1. Half-wavelength dipole antenna 33

3.2.2. microstrip antenna 34

Chapter Ⅳ. Proposed Research on the Bipolar IPG sensor 36

4.1. Design of the bipolar IPG sensor 36

4.1.1. Sensor size and operating frequency 36

4.1.2. Materials and structures 38

4.1.3. Correlation between the arterial area and impedance 39

4.2. Distensibility Correction Technique 41

4.2.1. Requirement of Pulse Transit Time 41

4.2.2. Principle of the algorithm for updating Pulse Transit Time 41

4.2.3. Correction algorithm for updating Pulse Transit Time 43

Chapter Ⅴ. Verification of the Bipolar IPG sensor and the distensibility correction algorithm 45

5.1. Verification of the sensor: correlation with PPG 45

5.1.1. Protocol 45

5.1.2. Result 45

5.2. Verification of the sensor: accuracy 47

5.2.1. Protocol 47

5.2.2. Result 49

5.3. Verification of the algorithm: persistence 53

5.3.1. Protocol 53

5.3.2. Result 53

5.4. Verification of the algorithm: robustness in various circumstances 54

5.5. Verification of the algorithm: safety and low-power consumption 55

Chapter Ⅵ. Discussion and Conclusion 57

References 59

List of Tables

Table 1. Some frequencies of the ISM (Industrial, Scientific, Medical) frequency band. The proposed sensor, for medical purposes, needs to use frequencies within the ISM band. Half-wavelength is calculated with not free space... 37

Table 2. Table indicating bias error, precision error, and mean absolute difference (MAD) for systolic and diastolic blood pressure in post-exercise measurements. 51

Table 3. Table presenting overall experimental results. The accuracy of the proposed sensor for pre- and post-exercise measurements meets Grade A accuracy according to the AAMI and BIHS standards for the certification... 53

List of Figures

Figure 1. Global Mortality Rates by Cause of Death. Cardiovascular diseases claim the highest number of lives, accounting for 31.8%. 17

Figure 2. Graph depicting awareness and treatment rates of hypertension among individuals categorized by gender. Females exhibit higher rates of awareness and treatment compared to males, with awareness rates for males... 18

Figure 3. Non-invasive blood flow measurement methods. (a) Measurement of vessel thickness using ultrasound [24]. Ultrasound probe generates ultrasound from the skin surface, analyzing the signals reflected from the vessel... 20

Figure 4. Various Debye models. Usually the two-component circuit is too simple to mimic with sufficient precision the frequency dependence of the variables measured. The three-component model combines features of... 22

Figure 5. Electrical resistance model of the artery with its own electrical properties. The artery has a time-varying cross-sectional area and a constant length. Bio-impedance (Z) of the artery is equivalent to the resistance (R) as... 24

Figure 6. Graph illustrating the time-varient representation of vascular information using bio-impedance. The graph shows fluctuations representing vascular information, but there are challenges regarding direct conversion... 24

Figure 7. Schematic diagram of blood pressure calibration using the law of conservation of energy. The left-hand side of the energy conservation equation represents the energy exerted by the heart to propel blood, while the... 26

Figure 8. Illustration explaining PTT, which is inversely proportional to the velocity of blood flow. While the changes in blood flow cannot be directly converted into blood pressure, they manifest as pulse waves in the form... 26

Figure 9. Neckband for continuous blood pressure measurement [58]. (a), (b). The neckband is composed of an IPG and an ECG, designed for utilizing the energy conservation law method. The IPG measures the pulse wave... 27

Figure 10. the method for calibrating blood pressure using the Moens-Korteweg formula. The Moens-Korteweg formula is an equation that relates the characteristics of the blood vessels (thickness, elasticity, internal radius)... 28

Figure 11. Wristband for continuous blood pressure measurement consisting of two IPGs [60]. A wristband composed of two IPGs is used to measure blood pressure using the vascular information of the radial artery in the... 29

Figure 12. The measurement of blood pressure waveform using the Bramwell-Hill equation. The Bramwell-Hill equation expresses the relationship between pulse wave velocity, blood density, blood pressure and the cross-... 30

Figure 13. Continuous blood pressure measurement using information obtained from PPG and IPG sensors and the Bramwell-Hill equation. Real-time changes in the cross-sectional area of the blood vessel provide the... 30

Figure 14. Method of obtaining information close to cross-sectional area using IPG. As IPG typically consists of four electrodes, it has a wide measurement range. Although it can contain volumetric information of the blood... 32

Figure 15. Adopting the working principle of the antenna to narrow the measurement range. The size of the antenna can be adjusted according to the operating frequency, allowing the antenna size to be reduced as desired.... 33

Figure 16. Working principle of a half-wavelength dipole antenna. The size of a half-wavelength dipole antenna is half the wavelength, allowing the electric energy from the source to distribute to the ends of the dipole antenna,... 34

Figure 17. Basic structure of the microstrip antenna (patch antenna) which consists of transmission line, antenna path, ground layer, and dielectric substrate. (a) Top view. (b) Side view. 35

Figure 18. Consideration of the size of the bipolar IPG sensor for application. The proposed sensor will be applied to a wristband to measure changes in the radial artery. Therefore, the lateral dimension of the sensor should be... 37

Figure 19. Utilizing a low-profile antenna type for the bipolar IPG sensor. A low-profile antenna is suitable for wristbands. An outer insulator is added to prevent direct contact between the skin and the metal (patch). 38

Figure 20. (a) Fabrication photo of the bipolar IPG sensor. Copper was used for the metal layer, and FR-4 was applied to the insulator layer. (b) Simulation and experimental results of the bipolar IPG sensor for the operating... 39

Figure 21. Finite element analysis simulation environment to determine the linearity between cross-sectional area and bio-impedance. The cross-sectional area of the blood vessel varies with the cardiac cycle. The simulation... 40

Figure 22. Configuration changes of the cross-sectional area of blood vessels with respect to the cardiac cycle. The blood vessel can be divided into the internal radius and cross-sectional area. In the vasodilatory phase, the... 40

Figure 23. Relationship between the total thickness of the blood vessel and the reciprocal of impedance. As the total thickness of the blood vessel increases, the impedance decreases, resulting in an increase in the reciprocal of... 41

Figure 24. Blood flow reflecting at a branching point. Blood within the radial artery flows distally but reflects at a branching point and returns. 42

Figure 25. Pulse waveforms generated by blood reflecting at a branching point. The blood reflecting and returning creates a second peak. The PPT can be easily obtained by differentiating the waveform. 43

Figure 26. Long-term maintenance with a single calibration using a reference device. Systolic and diastolic pressure can be obtained from the reference device to determine 𝛼 and PPT. 44

Figure 27. Blood pressure sensing performance results with a PPG sensor and a cuff device (n=20). (a) the pictures of the hand wearing wrist cuff device, PPG sensor on the fingertip and the proposed sensor on the wrist. The... 46

Figure 28. Optimal frequency distribution. 47

Figure 29. Experimental process diagram. The effectiveness of the algorithm is tested on 20 patients. After one calibration, no further calibration is required. Pre-exercise experiments are conducted to verify the accuracy of... 48

Figure 30. Blood pressure waveforms obtained before and after exercise. Both blood pressure and heart rate (HR) increased. 48

Figure 31. Filtering for blood pressure calibration. Signal filtering is included to match the obtained bio-impedance values with blood pressure values. Figure shows three cycles of the bio-impedance and a representative... 49

Figure 32. Bland-Altman plots for pre-exercise measurements. The accuracy of the proposed sensor during rest is found to be very high. 50

Figure 33. Bland-Altman plots for post-exercise measurements. The black and red data represent measurements with and without the correction technique, respectively. 51

Figure 34. Bland-Altman plot comparing pre-exercise (black) and post-exercise measurements (red) with the correction technique applied. Both systolic and diastolic blood pressure show high correlation and low errors. 52

Figure 35. Blood pressure measurements for long-term without calibration. Blood pressure waveforms at a time and after 24 hours. The estimated systolic and diastolic blood pressures are compared with the reference blood... 54

Figure 36. Experiment results for the robustness of the proposed sensor. (a) the picture of the sensor with water. (b) signal to noise comparison of the sensor with and without moisture. (c), (d) sensed bio-impedance signals at... 55

Figure 37. Safety comparison with the guidelines of the International Commission of IEEE [69] and ICNIRP. 56

Figure 38. The sensed signal using the powers: (a) 250, (b) 100, and (c) 50 μW, respectively. (d) SNR graphs of the sensor for various power consumptions. 56

초록보기

 Cardiovascular diseases are the leading cause of global death, but the awareness and treatment rates for the conditions remain low. One of the contributing factors for the lack of awareness and treatment is the discrete measurement of blood pressure using cuff-based sphygmomanometers, which can lead measurement errors. The healthcare professionals need continuous blood pressure waveforms noninvasively to make accurate diagnoses of patients' cardiovascular conditions. Additionally, for long-term monitoring, blood pressure measurements should consume low level of electric power. To achieve noninvasive and continuous monitoring, impedance plethysmography is adopted, which allows capturing vascular information using a low-power electromagnetic (EM) field from outside the body. The accurate representation of cardiovascular conditions can be obtained through the blood pressure waveform, which can be derived using the Bramwell-Hill equation. This equation expresses the relationship between blood pressure, vascular cross-sectional area, and pulse wave velocity. Conventional impedance plethysmography methods inject current into the body and measure voltage to obtain vascular impedance, focusing on vascular volume rather than cross-sectional area. Moreover, pulse wave velocity requires measuring blood flow at two different locations, necessitating the attachment of two sensors and resulting in high power consumption.

In this paper, we propose a bipolar impedance plethysmography sensor that obtains vascular impedance in a narrow area and a distensibility correction technique for calculating pulse wave velocity from a single sensor. To replace the electrodes that conventional impedance plethysmography sensors use, which require a large measurement area, we utilize the operating principle of an antenna forming bipolar nodes. The bipolar impedance plethysmography sensor operating at high frequencies allows for very narrow spacing between nodes, enabling the acquisition of vascular cross-sectional area information. The operating frequency of the antenna is set at 5.8 GHz, considering the spacing between nodes, application, and the Industrial, Scientific, and Medical (ISM) frequency band regulation. To apply the designed bipolar impedance plethysmography sensor to a wrist-worn mobile application, we employ a low-profile antenna type and add an insulation layer to prevent direct contact between the metal surface and the skin. We validated whether the designed bipolar impedance plethysmography sensor can accurately convert vascular cross-sectional area changes into impedance using finite element analysis simulations. The relationship between variation of vascular cross-sectional area and changes in impedance exhibits a high linear relationship with a Pearson's correlation coefficient (r-value) of 0.9.

The blood in the radial artery flows in the distal direction and encounters branching point, with some blood continuing in the same direction and some being reflected. These reflected signals form the second peak in the pulse wave. The length between sensor's position and the bifurcation location are fixed, and by calculating the time difference between the first peak formed when the blood initially passes through the sensor and the second peak formed by reflection, pulse wave velocity can be calculated. By utilizing pulse wave velocity calculated using a single sensor, and by measuring vascular cross-sectional area, the blood pressure waveform can be monitored.

The study obtained IRB approval and conducted experiments involving 20 subjects. The experiments began with calibration between a commercial cuff sphygmomanometer and the proposed sensor. The experiments were then divided into pre-exercise and post-exercise phases, with the subjects performing a brief aerobic exercise on an indoor bike for one minute between the phases to raise blood pressure and alter distensibility. The pre-exercise experimental results showed a Pearson's correlation coefficient (r-value) of 0.85 and 0.91 for systolic and diastolic pressure, respectively, with bias errors of -0.05 and 0.26 and precision errors of 3.38 and 2.62. The post-exercise experimental results showed bias errors of -6.31 and -2.22 and precision errors of 4.22 and 4.80 for systolic and diastolic pressure, respectively, indicating lower accuracy. However, by employing distensibility correction techniques, the bias errors for systolic and diastolic blood pressure were reduced to -1.08 and -1.79, respectively, and the precision errors improved to 1.92 and 3.39. As a result, the proposed paper utilizes a single sensor operating at 50uW to obtain blood pressure waveforms, with mean absolute differences for systolic and diastolic blood pressure below 4, meeting the Grade A standard set by the Association for the Advancement of Medical Instrumentation (AAMI) and the British and Irish Hypertension Society (BIHS).