Research Of ECG Signal Quality Evaluation And QRS Complex Detection Techniques

Recently, with the rapid development of medical information technology, the intelligent medical equipment are changing quickly. And the automatic processing for ECG signal is one of the most important diagnostic methods to prevent and diagnose of cardiovascular disease, which acts a great role in practical applications.This thesis detailedly introduces the ECG signal quality evaluation and QRS wave detection algorithm based on the current state of ECG signal automatic processing. Using higher order statistics and wavelet transform, three kinds of algorithms for evaluating ECG signal quality are proposed, which contains kurtosis coefficient, band coefficient and mixing signal quality index. In principle, the variance is proportional to the kurtosis coefficient, but inversely proportional to the band coefficient. The complementarity of two coefficients makes it possible to combine them as mixing signal quality index to assess the ECG signal quality. Then this thesis states three kinds of QRS wave detection algorithms with the traditional threshold difference, higher order statistics and fusion two-lead ECG signal respectively, which is proposed as novel algorithm. For the novel algorithm, the original signal is detected by each lead with single-lead method, and selected to first-lead examination, second-lead examination or two-lead decision method by two-lead decision algorithm, which contains two-lead fusion method and lead judge rule.Three kinds of assessment ECG signal quality are evaluated by using ECG records from the MIT-BIH Noise Stress Test Database, which indicates the validity of these models and the mixing signal quality index has superior estimated value than others. The fusion two-lead ECG signal algorithm are testified by using48sets of ECG records from the MIT-BIH arrhythmia database, and the achieved scores indicate high performance:99.87%sensitivity and99.81%specificity for QRS wave detection, And the false negative and false position in this algorithm are reduced by23.26%and18.27%respectively compared to the first lead, which is reduced by88.21%and95.11%respectively relative to the second lead. The experiments show that the algorithm with two-lead ECG signal can enhance the accuracy of single-lead QRS wave detection. Therefore, in this thesis, those algorithms have great research value in the field of ECG signal automatic processing and analysis.