Study On Acquisition And Analysis Of ECG Signal Based On LabVIEW

1. IntroductionThe electrocardiogram (ECG) has played a very important role in the clinicdiagnoses for heart diseases which is able to provide external standard for correctanalysis, treatment and inspection. In these days, the process is one of the researchsubjects in the biomedicine engineering. Its analysis and classification can releasethe expert burden and increase diagnoses speed.QRS detection provides the fundamentals for almost all automated ECGanalysis algorithm. It is the most important basis not only for diagnosingarrhythmia but also for analyzing other details in ECG. It is very difficult to detectthe exact QRS because of the complexity of ECG, the presence of noise e.g.baseline drift and circadian variability.Within the last decade many new approaches to QRS detection haveproposed and wavelet-based QRS detection algorithm is much better in resistingdisturb. Wavelet has the function of "changing focus" and is a "microscope" inhigh frequency which is very useful in singularity detection. The algorithm isfeasible for QRS detection by making use of different scales. By the informationin different scales, the characteristic shape of ECG can be described better anddiffer QRS from high P wave, high T wave, noise and baseline drift.We use LabVIEW for programming which is a kind of graphicalprogramming language---G language. It has so many uses like C and Basiclanguage. The difference is that G language is graphics mode but sentence mode.It consists of debugging tools of traditional language. The user may set breakpointand execute highlighting. Animates execution flow by drawing bubbles along thewires so it is very convenient to debug program.2. The research conceptThere are tow parts of ECG analysis in my paper: signal collection and QRSdetection. The former is the design of hardware and the latter is the one ofsoftware. The signals collected are transmitted to the computer by the USBconnecter and the DLL transferred by LabVIEW. The research concept is largelyconcerned with:1. Hardware design of ECG signals collection and data read byLabVIEWWe will design a static data-collection circuit of 18 leads while others are 12leads. There are three parts in our design in terms of the features of ECG and thefactual requirements: multiplexer, analog to digital converter and microprocessor.1. Multiplexer: We have used three 74VHC4051s which can perform tocollect data of 18 leads. The 74VHC4051 is 8-channel analog multiplexer.2. Analog to digital converter: We have used low-power 12-bit samplingCOMS analog to digital converter ADS7806. This chip has only 8 data outputsthat can be 4 least significant bits or 8 most significant bits in the terms of the pinBYTE. The data will be transferred to the microprocessor through a buffer, at thesame time the microprocessor will order when to enable the chip, when to enableA/D converter and when to transmit data.3. Microprocesser: We have used a high-quality, low-power AVR8 bitsmicroprocessor ATmega128.2. R detection of ECGIn this part, R detection algorithms are all based on wavelet transform and wehave made improved research.The first method is based on the quadratic B spline wavelet. The R iscorresponding to the zero-crossing of the transformed signals for the fixedscale s = 2j. We will firstly find the zero-crossings, and then find out the locationsof R peaks by correcting the time delay. By using this method, a R detection rateis 99.65% with 15 records of the MIT/BIH database.From the above method, we know the R waves correspond to thezero-crossing of the transformed signals based on the quadratic B spline wavelet.But we find out that the index of the zero-crossing need to be computeredindirectly, which is complex, and it may be disturbed easily, so we can not asurethe exact location of zero-crossing. The local max approach has importantadvantages with the R analysis based on Marr wavelet. An improved method hasbeen proposed based on the first derivative of Gussian function so that the Rwaves exibit obviously, which is the R detection based on the transform of thefirst derivative of Gussian function and Hilbert transform. In this part, the meanLipschitz exponent α of every level has been computed and the suitable α issaved. So we can detect the locations of R in this level by correcting the timedelay. By using this improved method, a R detection rate is 99.74% with 15records of the MIT/BIH database.3. ConclusionThere are two parts of ECG analysis in my paper: signal collection and QRSdetection based on LabVIEW. The signals collected are transmitted to thecomputer by the USB connecter and the DLL transferred by LabVIEW. The Rdetection has been based on the quadratic B spline wavelet and the first derivativeof Gussian function and Hilbert transform which is our improved method. With 15records of the MIT/BIH database, we have made a conclusion that the R detectionrate by the improved approach is higher, its running time is shorter. At the sametime, this will be the basis for the detection of other ECG waves and clinicdiagnoses.