| J wave is the abnormal change of electric potential at the junction of QRS wave and ST segment on the electrocardiogram,forming a hump-shaped,spike-shaped,or spike-like waveform.The appearance of J wave often indicates a malignant arrhythmia or other serious cardiovascular diseases and has become an important leading factor in disability or death.At present,the diagnosis of J-wave related diseases is based on the doctor’s clinical experience and has certain risks.Therefore,designing a classification model that can detect J wave accurately and quickly to provide reliable basis for clinical diagnosis is of great significance to reduce the fatality rate of these diseases.In this paper,from the perspective of signal processing,based on the idea of feature optimization,a J-wave identification optimization algorithm is proposed.The main research contents are as follows:(1)An optimal algorithm for J wave recognition based on feature selection is proposed.In this algorithm,firstly,after preprocessing the established database signals,that is,denoising and feature point detection,the features of the signal are extracted from time domain and time frequency domainrespectively.The morphological features of the time domain,the features of the wavelet packet decomposition coefficients and their statistical features are obtained to form the feature set.Secondly,the genetic algorithm is used to select the extracted features.The goal is to identify the main features and eliminate the redundant features,so as to promote the establishment of the J wave identification model with the best performance.After that,the mutual independence of features is proved by the correlation calculation of features.Finally,the selected features are used to train the support vector machine and verify the effectiveness of the algorithm.(2)A J wave recognition optimization algorithm based on genetic algorithm and principal component analysis is proposed.Based on the establishment of the J-wave model based on feature selection,the high-dimensional features in the selected features will increase the complexity of the J-wave identification model and may cause dimensional disasters.Therefore,the proposed algorithm uses principal component analysis technology to reduce the dimension of high dimensional cumulant features further,fully reduce the complexity of the model and simplify the computational complexity,so as to improve the classification accuracy and efficiency.By simulation,the proposed method for J wave detection based on feature optimization can achieve 97.5% accuracy,98.9% sensitivity,98.4% specificity and 98.9% positive predictive value.The classification performance is obviously better than other algorithms.At the same time,the classification time can bereduced to 2.3s by using the optimized features to detect the signals,which can achieve higher classification efficiency. |
