Research On Feature Extraction And ST Segment Recognition Algorithm For ECG Signal

Cardiovascular disease is one of the major diseases which threatent human health,and electrocardiogram(ECG) can be regarded as a reflection of electrical activity of theheart in the surface. In the clinical, ECG is important guiding significance in detectingand diagnosing heart diseases. In reality, the human ECG signal is easy to be influencedby various random signal, which makes it difficult to read the effective data andinformation from ECG.Therefore, research the ECG signal’s detection and processingalgorithm has important theoretical and practical application value. This paper mainlyrealizes the detection algorithm and optimization of ECG’s denoising、R wave、T waveand ST segment, and make a tentative optimization for part of the algorithm.We used coif4wavelet through8layer to decomposit the ECG signal for denoising simulation validation,we used B-spline wavelet transformations throughMallat algorithm to detect the peak point of R wave.When the peak point of Rwave is initially detected,if it present missed or mistaken checking,we try to adjust the threshold in the detection of modulus maximum use binary search algorithm.Do a summary of detection of double R wave、duplex R wave and inverted Rwave.We can try to use inquiry algorithm of arc approximation to determine the T wave’speak point after we have detected the peak point of R wave. Detected the T wave by thecircular arc approximation algorithm, by contrasting the peak point of T wave marked indatabase and ECG data we have detected, we found that the approximation algorithm isbetter than the detection algorithm of the wavelet modulus maxima in accuracy. At thesame time, the error between the peak value of T wave detected by approximationalgorithm and the real T wave peak point is smaller, arc approximation algorithm hasbetter detection effect, but we also need to further reduce the calculation, and needfurther improved in the work following.The segment from the QRS complex endpoint to the T wave starting point is STsegment. The accurate identification of ST segment has very important reference valueto myocardial ischemia, coronary heart disease and other symptoms. We choose the STsegment’s offset level, type curve, slope and the bump degrees as the index to identifythe form of ST segment, based on that we research the ST segment pattern recognitionclassification algorithm. Improved the criteria of offset detection,using average amplitude as a test standard; at the same time using polynomial fitting method toidentify the slope and smoothness of ST segment.This paper takes6beats in No.108data files of MIT-BIH ECG database to test the identification of ST segment,finish thepreliminary identification the of ST segment form.Further improvement of the accuracyis still needed in the future, and need to further improve the adaptive of algorithm.