DEVELOPMENT OF HEART ACTIVITY OBSERVATION SYSTEM BASED ON ECG SIGNAL ANALYSIS AND ELECTRONIC STETOSCOPE

• Main Author: Oktivasari, Prihatin
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/65049
• Institution: Institut Teknologi Bandung

Heart is the number one cause of death globally. Based on 2016 data from The Institute for Health Metrics and Evaluation (IHME), the United States statistical agency shows that more than 17.7 million people die from cardiovascular disease, representing 32.26% of deaths worldwide, with 63% over the age of 70. year, 29.13% aged 50-69 years, and 7.61% aged 15-49 years. This is expected to continue to increase until 2030 and reach the figure of 23.6 million who died from heart and blood vessel disease. Data in Indonesia in 2016 stated that deaths from heart disease reached 36.3 % of total deaths. This number is quite large considering that the disease is classified as an infectious disease that cannot be controlled. The heart is the engine of life. Therefore, if the heart is problematic, it will be very fatal. The heart has a mechanism of electrical flow that is triggered by itself to contract or pump and relax. Disruption of the heart's electrical flow mechanism, resulting in dysfunction of certain parts of the heart and causing changes in heart wave signals. The importance of reading heart signals for early detection of the heart has resulted in the development of tools to read heart signals. Signals and heartbeats can be observed with electrocardiogram (ECG) technology and auscultation techniques with a stethoscope in the field of clinical cardiology. This technology is non-invasive and can be used to study sensory and cognitive activity in both normal and pathological people. The current ECG is limited to P waves, QRS complexes, and T waves, and it is difficult to detect structural abnormalities of the heart valves and defects characterized by heart murmurs. The use of conventional stethoscopes also tends to be very subjective depending on the sensitivity of the ear, environmental noise, sensitivity, low amplitude, and frequency. The sound pattern that is heard is relatively the same and the sound cannot be saved. One solution to overcome these limitations is to combine the results of the analysis of the sound signal from the stethoscope and the heart signal from the ECG. The correlation of the two signals can strengthen observations on the heart, in terms of the circulatory system and heart electrophysiology. The drawback of stethoscopes using reference ECGs is that the time between electrical and mechanical activity in the cardiac cycle is not constant for all patients, due to various pathological conditions. Therefore, it is necessary to conduct a multimodal study of stethoscope sound signals and ECG signals for various types of heart disorders. Researchers used computational methods to get a high level of accuracy, using Fast Fourier Transform (FFT) and wavelet methods in time and frequency domains for signal characterization and Artificial Neural Network (ANN) algorithm for classifying normal and arrhythmic heart signals. In this research, it is also optimized in the development of electronic stethoscope and ECG hardware, especially in the amplifier circuit, filters for more accurate results in removing noise, and real-time data acquisition. The purpose of this study is to interpret the signal from the recording of the electronic stethoscope system and the heart record system and to observe cardiac activity, with the stages covering instrument design, data collection, feature extraction, signal analysis and signal classification. The signal feature extraction includes HRV parameters,, RR, HR, P onset-offset, QRS onset-offset, T onset-offset, and ST segment. The heart sound signal response test was carried out with normal respondents and in the cardiac poly at the hospital. The result of signal characteristics in normal heart using FFT and wavelet studies, identified the pattern of S1 ranging in frequency between 50-150 Hz, tends to be lower in frequency than S2. While S2 has a slightly higher frequency above 150 Hz (between 150-200 Hz). While the heart with abnormalities (murmur) has an amplitude frequency of up to 2000 Hz. While the normal heart signal pattern PQRST lies in the frequency range of 5-50 Hz. The characteristics generated from the cardiac recording system show the P time interval is around 0.089-0.182 seconds, the PR segment is in the range 0.116-0.201 seconds, the QRS segment is in the 0.027-0.036 second range, and the ST segment is in the 0.091-0.235 second range. The cardiac signal response testing was carried out on normal respondents and respondents in the cardiac poly hospital. The test was carried out with light activity or no activity treatment on normal respondents, with measurements of heart rate variability parameters, average HR and average RR. Cardiac signal analysis was recorded on 20 normal respondents with light activity treatment, resulting in an increase in the average HR value, an increase in NN50 and RMSDD values, and a decrease in the average RR value. Meanwhile, the grouping of respondents based on Body Mass Index (BMI) and age, resulted in a relationship, among others, if the BMI value is large, then HR is large, HR is large and RR is low, while age has no effect on HR and RR values. The results of the power spectrum from the distribution of the LF, HF and VHF frequency bands for normal condition respondents have a higher energy range than the abnormal, generally the highest power is > 0.1 for normal, and < 0.1 for abnormal. Optimization of network architecture in the process of detecting heart defects, including learning rate of 0.02, 10 hidden layer neurons, number of iterations (epoh) 1000, target error 0.001. The network is able to recognize 100% of the trained data. The highest amount of data recognized by the ANN system was 119 data, and, with epochs of 49 iterations in 00:01 seconds, and MSE value of 0.000915. The validation of the electronic stethoscope system was carried out with a commercial electronic stethoscope Littmann 3200, the results did not show a significantly different signal pattern, the recorded signal already showed a S1 and S2 pattern, and the frequency distribution was not much different. While the results of heart signal measurements were compared with the 3-lead ADAS1000 commercial system, and 12-lead ECG at the hospital, it showed that the recorded patterns were similar, although they did not show the exact same results for the time area and frequency distribution, but they did show a pattern. the same up-and-down trend. ECG system experimental results compared with commercial ECG there is a difference in results of about 5-10%. Multimodal signaling was performed using a database with the same sampling at 10 seconds, by recording the PQRST heart signal in the lying position and the heart sound in the sitting position. Experimental results show that the number of R peaks is the same as the number of S1 sound peaks, and the number of T peaks with the number of S2 sound peaks, with an average time difference of 0.016 seconds, so that it can be concluded that there is a regular signal pattern relationship between S1-S2 and the RT wave, namely the relationship at the end of the first peak of the QRS wave. cardiac signal due to ventricular depolarization (ventricular contraction) and the appearance of an S1 heart sound and the association of the end of the next peak of the T wave of cardiac signals indicating ventricular repolarization and the appearance of an S2 heart sound. This is consistent with the fact that electrical events in cardiac activity occur before mechanical events. Based on the study of HRV parameters, heart sound signals can be used to determine HRV parameters, the results show the same number of peaks in normal hearts, while in abnormal hearts there are differences in results because abnormal heart conditions have an erratic rhythmic pattern.