
	Trending ▼ ICSE CBSE 10th ISC CBSE 12th CTET GATE UGC NET Vestibulares ResFinder

Notes PG 2020 : Science (Cochin University of Science & Technology (CUSAT), Kochi)

6 pages, 7 questions, 0 questions with responses, 0 total responses,

Rahul Gnath
Cochin University of Science & Technology (CUSAT), Kochi
b tech eee

+Fave Message

Profile Timeline Uploads

Home > rahulsctimst > F Also featured on: School Page

Formatting page ...

Proceedings of the 8th World Congress on Intelligent Control and Automation July 6-9 2010, Jinan, China A New ECG-based Automated External Defibrillator System Wenguang Han1,Yongjun Li1,Rui Zhang1,Chao Hu1,2 Max Q.-H. Meng1,2 1.Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences 1.Department of Electronic Engineering The Chinese University of Hong Kong 2.Shenzhen Institute of Advanced Integration Technology, Chinese Academy of Science Shatin, N.T. HongKong {chao.hu@siat.ac.cn,max@ee.cuhk.edu.hk} 2.The Chinese University of Hong Kong Shenzhen, China wg.han@siat.ac.cn Abstract Automated External Defibrillators (AED) are basic portable defibrillators that are designed for minimallytrained or untrained non-medical personnel. A microprocessor inside the defibrillation automatically analyzes the patient s heart rhythm and advises the operator whether a shock is needed. Audible and visual prompts will guide the user through the process. AED would advise a shock only to ventricular fibrillation. In this paper, we will present a framework of AED system. There are two crucial components should be elaborately designed, the hardware system and the algorithm of detecting ventricular fibrillation. In this paper, we will present the design of hardware system and a new algorithm which could discriminate ventricular fibrillation (VF) from other unshockable rhythm through the measurement of Sample Entropy (SampEn). We compared the sensitivity, specificity, positive prediction, accuracy of the new algorithm with several earlier VF detection algorithms. The experimental results prove that the new algorithm can be well suited for short data sequences analysis, and reaches an elegant balance between detection time and accuracy. Index Terms Automated External Defibrillator, Sudden Cardiac Arrest, ECG Collection, Algorithm of Defibrillation, Ventricular Fibrillation, Sample Entropy, Bi-phase waveform I. INTRODUCTION Sudden Cardiac Arrest (SCA) has been one of the leading causes that strike people to death without any forebode, while the key cause of SCA is the ventricular fibrillation (VF). In VF, the heart is in an uncoordinated and invalid state. Many victims of SCA could survive if bystanders can apply first aid correctly and immediately while victims have not lost ventricular fibrillation. Electrical defibrillation can recover victim rhythm to normal state, and is well proved as the most effective therapy for cardiac arrest caused by VF or ventricular tachycardia (VT) [1]. A lot of research papers have shown that the delay from collapse to delivery of the first shock is the most important factor of the survival. The possibility of successful defibrillation declines at a rate of 7-10% with each minute of delay [2]. During the defibrillation, a rescue device is required and it is called defibrillation. The defibrillator was invented in 1946, and at first it used 978-1-4244-6712-9/10/$26.00 2010 IEEE 2204 the alternating current (AC) method. In 1950s, a new method of defibrillation with direct current (DC) method was discovered by ZOLL, and this DC method substitutes the old AC method quickly. With the improvement of the defibrillation technology, a major breakthrough came to the introduction of Automated External Defibrillators (AED). In the early days, single-phase waveform defibrillation is the major way for SCA treatment. However, with the progress of defibrillator technology, biphasic waveform [3] defibrillation has become more and more important due to its low power requirement, perfect effect and high livability. Traditional AEDs have lots of disadvantages such as bulky, low velocity of charge and low sensitivity of VF detection. So we need to develop a high stable and efficient AED system. In this paper, we present a framework of AED system in both hardware and algorithm. Firstly, we built an ECG signal collection system. Here, we focus on the design of amplifier and acquisition circuit, which is a key part of the hardware system and even the whole defibrillation system. Secondly, the charging circuit and energy selection circuit are presented, which store some appropriate power in two energy-storey capacitors. Finally, the discharge circuit was designed to release the high voltage waveform to the objective. The detection algorithm is another pivotal component in the whole system. It is used to distinguish the ventricular fibrillation (VF) correctly and promptly from the non VF. If a normal sinus rhythm is misinterpreted as VF that might lead to unnecessary shock delivery, a fatal damage to the patient s heart might occur. Therefore, an appropriate VF detection method must be found for any AED. The sample entropy (SampEn) [4] is used in our algorithm to serve as a descriptor of VF detection. The sample entropy is well suitable for analyzing short and noisy datasets. Compared with other conventional time and frequency domain approaches, as well as nonlinear dynamic methods, this presented algorithm can reach an elegant tradeoff between detection time and accuracy. The simulation and experimental results show that the new algorithm can reach 69.1% sensitivity, 93.6% specificity, 72.9% positive prediction and 88.7% accuracy. The following part of this paper is organized as follows. In section II we describe the hardware system briefly. In section III we analyze acquisition circuit, charge circuit and discharge circuit in detail. In section IV, we depict the sample entropy s application in ECG signal analysis, and the paper is concluded in section V. II. HARDWARE SYSTEM OVERVIEW The hardware system is made of control board, high speed and performance charge board and bi-phase wave discharge board. The control board collects ECG signals and analyzes the cardiac rhythm with microprocessor. When VF is detected, the controller would tell the charge board charge up the capacitors. If the capacitors reach a certain voltage and be stopped, the discharge board is ready for a shock to human. Every board mentioned above has a series of circuits, which are introduced as follows: Fig.1 agriculture diagram of S3C2440 [5] 1) The control board is composed of ECG collection circuit, chest impedance detection circuit and microprocessor. ECG collection circuit is a basic and important part in the system. It can collect the ECG signals from human body and send out different voltage signals. Chest impedance detection circuit was designed for getting the chest impedance of human body. With the value of this impedance, we can select some energy for defibrillating. The ARM controller we used is Samsung S3C2440, which is the industry s fastest ARM-based Application Processor. S3C2440 is a cost-effective, lowpower microcontroller solution in a small form factor. It features an ARM920T core, a 16/32-bit RISC microprocessor core [5]. It manages all components, executes software system, disposes ECG signal, select the energy channels and other operation. The agriculture diagram shows in Fig.1. Fig.2 Principle diagram of defibrillator III. COMPOSITION OF DEFIBRILLATION SYSTEM A. ECG collection ECG collection board has four important parts, which are shown in Fig.3. The active snubber circuit works to get stable signals from electrode. Then preamplifier circuit will make a ten magnification times of signals that get from the snubber. At last signals get though the active band-pass filter and active trap filter, a band of 0.05~100Hz (except 50Hz) would be captured. The details of circuits above would be expounded below. 1) The active snubber (shown in Fig.4) is carried out by voltage follower. We realized it with high precision amplifier OPA4277, whose offset voltage 10uV and bias current 1nA [6]. By applying the active snubber, we can get a stable signal at the output pole. 2) For preamplifier (as shown in Fig.5), we use AD620 that is a low drift and precision instrumentation amplifier to carry out a differential amplification of 10 times gain. AD620 has a good performance with its excellent parameters: input offset voltage 50uV, Common Mode Rejection Ratio 120dB, amplifier gain range 1~10000 times [7]. 3) Since the frequency band of ECG signals are from 0.05Hz to 100Hz, we set a band-pass filter with a bandwidth at 0.05~100Hz. After this we set a two-order trap filter to remove the 50Hz power-line interference. In the two parts of the filters above, we adopted a low power JFET 2) The charge board is made up of energy selection circuit, oscillation circuit and over voltage detection circuit. Because there are more than ten selection channels, an analog multiplexer is used to switch the different reference voltage corresponding to different energy. Oscillation circuit is supported by microprocessor, and it offers a PWM waveform to the boost circuit. Over voltage detection is based on a voltage comparator, which reference voltage is supplied by multiplexer s output. 3) The discharge board is composed of external discharge circuit, self discharge circuit and switch drive circuit. External discharge circuit and switch drive circuit are combined to release the bi-phase waveform, and the self discharge circuit could consume the remained energy. In addition, power supply is necessary. The system s block diagram is shown in Fig.2. 2205 quad operational amplifier TL064. TL064 has features as flows: a low power consumption 200uA; high slew rate; low input bias and offset currents; low offset voltage temperature coefficient [8]. multiplexer and switches one of sixteen inputs to a common output as determined by the 4-bit binary address lines A0, A1, A2 and A3, which is shown in Table I. An EN input on the devices is used to enable or disable the device. The DG506 provides low power dissipation and gives low on resistance. These features are suitable for our system. The schematic diagram of multiplexer is shown in Fig. 6. Connect output of multiplexer (Dout) to the negative pole of voltage comparator as a reference voltage. The voltage which is a feedback of transformer is connected to positive pole of voltage comparator. When the voltage of TR is higher than the reference voltage Dout, the voltage comparator would putout a high level so as to telling you the capacitor voltage has reached a certain standard, and the process of charge has been over. Table I Corresponding relation of energy setting and logical signal Fig.3 Flow chart of ECG collection system . Fig.4 circuit schematic of snubber (a) (b) Fig.6 schematic diagram of energy selection circuit C. High voltage charge and bi-phase wave discharge circuit In order to get a high voltage above 1500V, we used flyback converter to implement a convert of 15V to 1500V. The principle is as follows: when the MOSFET is closed, DC power charge up the transformer quickly. While the MOSFET is opened, the transformer would give it energy to capacitors C1 and C2. The principle diagram is shown in Fig.7. We made a PWM signal control the MOSFET close or not. The discharge board is composed of there parts, and they are two high voltage capacitors, external discharge circuit and self discharge circuit. Electrifying circuit charge the two capacitors though transformer to a certain voltage. The processor could control the switch tube and relay close or not so as to release biphasic waveform to human body according the algorithm of defibrillation. Fig.8 is the principle chart of discharge board. Fig.5 schematic diagram of preamplifier circuit B. Energy selection According to American Heart Association s suggestion, we set 10 grade selection channel, shown as in Table I. In order to reduce the number of amplifier channels, a key technique is to use multiplexers. Because there are 10 signal channels, analog multiplexers are needed to choose different channel. In this system, we used one DG506 [9] to realize this function. The DG506 is a 16-channel high performance analog 2206 and the division of the two clusters is obvious (Fig. 11). Here we choose for the threshold d0 = 0.25 . Fig.7 Principle diagram of boost circuit (a) (b) Fig.9 Typical non VF episode in the ECG signal cu01 from CUDB and corresponding sample entropy values. Fig.8 principle chart of discharge board IV .DETECTING VENTRICULAR FIBRILLATION ALGORITHM Sample entropy is a measure of the rate of information production, which was proposed by Richman and Moorman [10]. The algorithm of sample entropy is operated as mentioned in [14]. In order to get a lower computational burden, we apply a rapid algorithm to compute sample entropy based on dual-value matrix, which was first proposed by Bo Hong [10]. Fig. 9 and Fig. 10 show typical non VF and VF signals from the CU database and the corresponding sample entropy values. When computed sample entropy, we first aligned the end of the window to the data point 1250 (i.e., 5 s) of the recording and then consecutively shifted the window 1 point forward for the next computation, so the total number of sample entropy values is 751 ( 8 250 1250 + 1 = 751 ).We can see clearly that the sample entropy values of non VF change slowly and distribute among 0.04~0.07, while those of VF change greatly and distribute all above 0.37. In order to get the threshold, we first selected 100 records from CU database according to the beat and rhythm annotations, including 50 records of non VF and 50 records of VF. Then we calculated the maximum, minimum, mean, and standard deviation of sample entropy through statistical method. At last a reasonable threshold would be captured by which we can distinguish VF from non VF. The sample entropy values of non VF and VF signals distribute among 0.05~0.22 and 0.26~0.52 respectively without superposition (a) (b) Fig.10. Typical VF episode in the ECG signal cu01 from CUDB and corresponding entropy values. 2207 V. CONCLUSION An efficient hardware system is proposed for automated external defibrillator system. Considering the ECG signal is pretty weak and has a large variation range, we emphasize on the circuit design of collection board. With adopting the techniques of preamplifier, active band-pass filter and active trap filter, the proposed system gets high quality and accuracy ECG signals. The boost circuit has also got a successful performance in the practical experiment. We utilize MULTISIM to simulate chest impedance detection circuit, and the simulation has got wonderful results in accuracy and efficiency. Certainly, a further work should be done on controlling the discharge circuit with SCR or IGBT to scale the bi-phase waveform better. In addition, we found in our algorithm of detecting VF that the entropy value is very sensitive. It is very easy to detect whether VF episode happens. But as time went by, with the VF signal changes irregularly, the corresponding entropy value may change greatly. The threshold was decided by a limited samples, so further analysis is needed on a wider variety of signals. Fig.11 Distribution of sample entropy values for non VF and VF episodes The performance of our algorithm was assessed by sensitivity (Se), specificity (Sp), positive prediction (Pp) and accuracy (Ac). Define as followings: TP (14) Se = . TP + FN TN (15) SP = . TN + FP TP (16) PP = . TP + FP TP + TN (17) AC = . TP + FP + TN + FN Where, TP was true positive, the VF case being correctly recognized as VF. FN was false negative, the VF case being wrongly recognized as non VF. TN was true negative, the non VF being correctly recognized as non VF. FP was false positive, the non VF being wrongly recognized as VF. Table II shows the sampling sets of non VF (2670) and VF (663), as well as the number of TP, FN, FP and TN. Table III shows the values of the sensitivity, the specificity, the positive prediction and the accuracy of the new algorithm (SampEn) and the corresponding values of some other algorithms investigated in [11][12]. ACKNOWLEDGMENT This work is supported by Chinese National High Technology Research (863) Fund (2007AA01Z308) National Natural Science Foundation of China (60904031), the Knowledge Innovation Engineering Funds of Chinese Academy of Science, Sc.&Tech. Research Funds of Guangdong, Shenzhen and Nanshan Government, and also sponsored by SRF for ROCS, SEM, awarded to Chao Hu. REFERENCES Resuscitation Council (UK), The use of Automated External Defibrillators, [online]. http://www.resus.org.uk/pages/AtoZindx.htm pp.21-25 [2] WL Lim, CC Hang & KB Neo, Discontinuous Innovations Framework: A Review of Automatic External Defibrillators in the Healthcare Industry, IEEE ICMIT, Sept. 21-24, 2008, pp.356-361. [3] Martens PR, Russell JK, Wolcke B, et al. Optimal response to cardiac arrest study: defibrillation waveform effects [J]. Resuscitation, vol.49, no.3, 2001, pp.233-243. [4] Joshua S. Richman, J. Randall Moorman, Physiological time-series analysis using approximate entropy and sample entropy, Am J Physio Heart Circ Physio, 2000, 278: H2039-H2049. [5] Samsung s S3C2440 Application [online]. www.usa.samsungsemi.com [6] High Precision Operational Amplifiers OPA4277 [online]. Available: www.ti.com/ww/analog/index.html [7] Low Cost, Low Power Instrumentation Amplifier AD620 [online]. Available:http://www.analog.com/zh/otherproducts/militaryaerospace/ad620/products/product.html [8] Low Power JFET Quad Operational Amplifiers [oneline]. Available: http://www.21icsearch.com/pdfdetil_50020423F18AC8E8.html [9] 16 Channel Analog Multiplexers DG506 [oneline]. Available: http://www.21icsearch.com/pdfdetil_C83EBBED4028A1D4.html [10] Hong Bo, Tang Qmgyu, Yang Fusheng, ApEn and Cross-ApEn Property, Fast Algorithm and Preliminary Application to the Study of EEG and Cognition, SIGNAL PROCESSING, 1999,VOl.15 No.2:100-108. [11] A. Amann, R. Tratnig, and K. Unterkofler, A new ventricular fibrillation detection algorithm for automated external defibrillators, IEEE Transaction Computers in Cardiology, vol. 32, 2005, pp. 559 562. [1] TABLE II The results of evaluating the VF detection algorithm: the number of TP, FP, FN, TN. Classification Results VF(663) TP(458) FP(171) Non VF (2670) FN(205) TN(2499) TABLE III The performances of the new algorithm and other algorithms Type of Se(%) Sp(%) Pp(%) Ac(%) algorithm TCI 71.0 70.5 38.9 70.6 VF-filter 30.8 99.5 94.5 85.2 SPEC 29.0 99.3 92.0 84.6 CPLX PSR HILB SampEn 56.4 70.2 74.7 69.1 86.6 89.3 85.4 93.6 52.7 65.0 59.1 72.9 80.3 85.1 83.0 88.7 2208 [12] A. Amann, R. Tratnig, and K. Unterkofler, Detecting ventricular fibrillation by time-delay methods, IEEE Transaction on Biomedical Engineering, vol. 54, no. 1, 2007, pp.174:177 [13] Massachusetts Institute of Technology, CU database [Online]. Available: http://www.physionet.org/physiobank/database/cudb database/cudb. [14] Dongmei Bai, The Extraction And Analysis of ECG Features, DaLian University of Technology, 2005, pp. 37:40 2209

Formatting page ...

Related ResPapers
Notes PG 2020 : Science : Defibrillator
by rahulsctimst

Notes PG 2020
by charvi007

Notes PG 2020 : Science (Cochin University of Science & Technology (CUSAT), Kochi)
by rahulsctimst

Formatting page ...

rahulsctimst chat

Horizontal lines at:
Guest Horizontal lines at:
AutoRM Data:
Box geometries: