Study and Implementation Of Automated Cardiac Arrythmia Diagnosis System Using Adaptive Cardiac Profiling And Statistical Features

Abstract

Automated arrhythmia-diagnosis systems for inter and intra-patient variation cases that newlinecan provide high-classification accuracy rates are still an active area for research. newlineMethods for classification such as detection of any symbol and repeated pattern, network newlinebased on neurons, vector machines, learning methods based on training data, combination newlineof neuron and fuzzy using statistical and dynamic features and classification of the ECG newlinesignal by using cross wavelet transform has been proposed by number of authors. newlineHowever, such an approach suffers several pit falls. Therefore the design for patientadaptable newlineclassifier is needed to overcome this pit falls. newlineThe design of efficient Automated cardiac arrhythmia-diagnosis systems for newlineheartbeat detection and classification is the active research area in this field. The newlineproposed design will be very suitable for monitoring cardiac health in remote areas where newlinea cardiac specialist is not easily available and treatment is extremely costly. newlineWe proposed a novel method for pre-processing of ECG signal, QRS beat newlinedetection and features extraction of ECG signal, abnormality detection in ECG waveform newlineand classification of ECG beats by plotting a profiling curve. newlineNoises are removed by conducting a mathematical method based on varying newlinewindow length as according to the distance from the adjacent Rpeak. High frequency newlinenoise (power line interference, electromyography noise) is removed with the help of the newlinevarying window mean procedure. The low frequency noise (baseline wandering, motion newlineartifact) is removed with the help of Fast Fourier Transform. newlineThe QRS beat is the most striking waveform within the electrocardiogram (ECG). newlineQRS beat detection provides the fundamentals for almost all automated ECG analysis newlinealgorithms. For detection of QRS beat in ECG signals, we proposed a method which is newlinenormally used for Intrusion Detection in network. This Network Intrusion Detection newlinetechnique is used for detection of worm in networking applications and for finding the newlinefrequently repeated strings. Our technique relies more on the data stream corresponding newlineto ECG beats than any particular feature. Here the concept of shared counters is used to newlineminimize the memory requirement through shifting suspicious strings. The algorithm newline