Blood Pressure Estimation Using Photoplethysmography (PPG) 1 Siddhi Sham Karande, BE E&TC, VIIT Pune. 2 Kiran Rajendrasingh Thakur, BE E&TC, VIIT Pune. 3 Sneha Dattatraya Waghmare, BE E&TC, VIIT Pune 4 Mrs. S.K. Habbu,Assistant Professor, E&Tc Engineering, VIIT Pune. Abstract-Blood Pressure is one of the common method used to monitor health. B.P. measurement is useful to continuously observe fluctuations in the systolic & diastolic blood pressure. Non-invasive methods like a mercury sphygmomanometer are used for taking blood pressure measurement, but are not continuous. In this paper, a theoretical explanation of invasive method of estimating systolic blood pressure using PPG is aimed. In this method, wearable, continuous, and cuff less device can be developed and subjects having hypertension can use this device to control their BP levels using PPG signal. This method could be able to provide a new BP measurement system with more convenience and accuracy. Keywords-blood pressure, BP, PPG, invasive, estimation. I. INTRODUCTION Blood Pressure, which indicates force exerted by blood flowing through our blood vessels on arterial walls, is critical parameter used to monitor health. Deviation from its normal level can lead to disease like heart disease, failure of kidney and hypertension. High BP can lead to bursting of blood vessels while very low blood pressure causes dizziness and even loss of consciousness. If left untreated, high BP can lead also lead to stroke, enlarged heart, heart attack and even peripheral vascular disease which lead to lack of blood circulation in legs, cramp-like pain in calves. In addition, it is vital to monitor the blood pressure of patients to ensure they re stable before and after surgery. The oxygen content of blood is another crucial parameter to determine good health. In healthy patients, the oxygen saturation, called SpO 2, is between 98% and 100%. This can measured using non-invasive clip-on device called finger oximeter(fig.1), which measures amount of light of specific frequencies to calculate amount of oxygen present.[1] The graph obtained by plotting the electrical signal generated by the photodiode is called a photoplethysmogram(ppg). A normal PPG signal looks as shown in Fig.2. 102
Fig.1 Finger Oximeter Fig.2 PPG Waveform The motivation behind this paper is to introduce and estimate Blood Pressure using PPG Signal from our finger tips.features like Energy, Amplitudes and phases were extracted from cardiac components. Then, Artificial Neural Network was trained using this features. Our method is convenient and fast so, can be used to analyse pulse wave along with estimation of BP. Our objective is to generate algorithm so that we can estimate accurate BP from signals of PPG. Our objective is that the cuffless, wearable and the most essential feature continuous device can be created and the hypertensive patient can utilize our device to control BP in better way. II. LITERATURE SURVEY SR. NO. METHOD/DEVICE ADVANTAGES DISADVANTAGES 1 Simple Should be handled carefully to protect the MERCURY Durable damage to the mercury containing tube SPHYGMOMANOMETER Accurate Not suitable for household use No need for recalibration. ANEROID Cheaper 2 SPHYGMOMANOMETER More portable Operating this device requires practice. Less expensive Compact 3 AUTOMATIC DIGITAL Portable Device is delicate and requires proper care SPHYGMOMANOMETER Easy operation Repairing device can be complicate Minimum chances of human error Cannot be used in the noisy environment 4 AUSCULTATORY METHOD another. Simple Mechanical error might be introduced Doesn t require much equipment Observation differs from one observer to TABLE 1: VARIOUS MEHODS USED TO MEASURE BLOOD PRESSURE 103
III. SYSTEM ARCHITECTURE SUBJECT PPG DATA Corresponding B.P. Data Filtering Feature Extraction B.P. Readings Non-linear Neural Network MATLAB Estimated B.P. Fig.3 SYSTEM ARCHITECTURE i. Subject: Subject is the patient (men and women) who s B.P. and pulses were taken. ii. Data Acquisition System: PPG and BP readings were taken simultaneously of 150 subjects with sampling rate of 100MHz. BP was measured by automatic digital sphygmomanometer at arm and by using pulse oximeter at finger tips PPG was taken. 18000 PPG sample of each subject were taken for 3 minutes of duration. The PPG data taken was stored as an excel file for each subject.(ref. Fig.3) IV. BLOCK DIAGRAM 104
Fig.4 block diagram of data acquisition system V. METHODOLOGY COLLECTING PULSES Fig.3 FROM Block PULSE Diagram OXIMETER AND CORRESPONDING B.P. CREATING DATABASE Fig.4 BLOCK DIAGRAM EXTRACTING FEATURES IN TIME AND FREQUENCY DOMAIN TRAINING NEURAL NETWORK ESTIMATING BLOOD PRESSURE Fig.5 1) Database of 150 people were created out of which 100 consist of normal blood pressure and 50 having high blood pressure (above 135mmHg). 2) The raw data was filtered using averaging filter and smooth function. 3) Features were extracted from filter data. 105
4) Features extracted include amplitude, rise period, fall period, total period and difference between amplitude in time domain and FFT in frequency domain which gives energy and entropy of PPG signal. Mean, variance and sleence was removed of each feature. These features were given as input to non-linear neural network along with corresponding blood pressure and thus neural network is trained. 5) To train ANN,we used Levenberg -Marquardt algorithm. For training 70% of data was used, for testing 15% and 15% for the validation. 6) For estimation of Systolic BP, we are using trained neural network. VI. APPLICATIONS i. BP Monitors are used to measure blood pressure in an artery. ii. It contains a pressure sensor for sensing arterial wall vibration. iii. Non-invasive method to count systolic blood pressure. iv. Cheaper way for regular monitoring. v. User-friendly operations. vi. Continuous data monitoring. VII. CONCLUSION & FUTURE SCOPE 1. Our study proposes a new way of estimation of Blood Pressure by PPG signal. It is possible to improve the accuracy of signal by measuring PPG at point closer to heart. It is possible that with better methods of measuring the PPG, an improved system can be observed. 2. In this paper, system designed is used to measure systolic blood pressure. Future studies can also used to estimate diastolic blood pressure. 3. In the future, including an accelerometer to sense the motion in the finger oximeter can help compensate for the artifacts introduced in the PPG signal due to movement especially during clinical procedures. 4. Hence, despite the PPG signal being quite variable and dependent on any parameter, part of a complex system, with a few improvements to the device s sensitivity, the arterial BP can be monitored continuously and noninvasively.. VIII. REFERENCES 1) 1) Xiaoman Xing, Mingshan Sun. Optical blood pressure estimation with with photoplethysmography and FFT based neural networks,biomedical Optics Express,2016 106
2) 2)Hyung Ro Yoon, A New Method to estimate Arterial Blood Pressure using Photoplethysmographic Signal, 2006 International Confereevce of the IEEE Engineering in Medicine and Biology society, 08/2006. 3) 3)Forouzanfar, Mohamad, Hilme R. Dajani, Voicu Z Groza, Miodrag Bolic, Sreeraman Rajan, and Izmail Batkin Oscillometric blood Pressure Estimation:Past,Present and Future, IEEE Reviews in Biomedical Emgibeerubgn2015 4) Continuous Non-Invasive Arterial Blood Pressure Measurement Using Photoplethysmography. (escholarship UC item). 5) Wireless wearable photoplethysmography sensor for continuous blood pressure monitoring.(2016,ieee) 6) Cuffless PPG based continuous blood pressure monitoring.(2016,ieee) 7) Continuous blood pressure monitoring during exercise using pulse wave transit time measurement.(2014,ieee) 8) Image source www.google.com 107