Research And Implementation Of Moving ECG Signal Processing And Waveform Automatic Recognition

This dissertation studied pretreatment technology of ECG and waveform automatic identification technology, which is part of ECG signal processing and waveform Automatic Identification System.This dissertation summarizes the development of ECG signal preprocessing and automatic identification, summarizes the basics of electrocardiogram, PTB diagnostic ECG database, and designs preprocessing of signal system and ECG waveform recognition system overall. On this basis, the baseline drift removal algorithm and automatic recognition algorithm are designed.ECG baseline drift noise greatly influences the accuracy of automatic identification system. According to analyzing the ECG signal baseline drift characteristics, the dissertation designed "baseline drift removal algorithm" based on the first difference and cubic spline curve to filter out ECG baseline drift noise. The work lays the foundation for the automatic identification of ECG waveform. Finally simulation results of the algorithm are verified by the PTB diagnostic ECG database.This dissertation makes an exploratory research on ECG waveform automatical recognition technology and proposes "the waveform recognition algorithm based on the geometric characteristics". The algorithm is divided into two stages:in the first stage, extracting geometric features of the waveform. There are two main waveform geometric features:the motion trends of point and the optimal slope, we design movement trend determination algorithm and optimal slope algorithm to extract each of the two geometric characteristics respectively. According to the principle of determining trends, movement trend determination algorithm predict future movement trends of each point by calculating slopes of approaching five different points of each point. The slope optimization algorithm obtains the optimal slope with the mean value theorem on the basis of the trend determination algorithm. Second stage, waveform recognition. Waveform recognition stage is divided into five parts:P-wave recognition, Q-wave recognition, R-wave recognition, S-wave recognition, T-wave recognition. Firstly, every recognition limits the range of waveform characteristic points by using two geometric characteristics values obtained in the first phase of the geometry, threshold of voltage and threshold of time secondly uses the modulus maxima to detect the feature points of the waveform.Then, we use the VS2010, TeeChart chart control and other development tools to develop the ECG signal processing and waveform automatic identification system based on algorithms proposed in this dissertation. The system realizes the function as follows:the ECG signal preprocessing, waveform recognition, feature parameter detection and issuing initial diagnostic report.Finally, diagnostic ECG database in the Physikalisch-Technische Bundesanstalt is used to evaluate the performance of the algorithm. All the algorithms designed in this dissertation are proven to achieve more satisfactory results, high reliability and strong practicability.