T Wave Alternans Computer Detection Research

Cardiopathy, as a type of sudden and chance diseases, has always beenone of the main threats to the health of human being. Sudden cardiacdeath caused by kinds of cardiac disease is an especially large threat.T-wave alternans has been fully recognized as an independent andstatistically significant index for clinical forecasting SCD. TWA is a veryfeeble electrophysiologic phenomenon, which is easily interfered byheart-rate, noise and ectopic beat and accordingly detected difficultly.Onthe basis of summarizing and analyzing the progress and status quo ofTWA research, this dissertation focused on the the detection of TWA. Themajor contributions include the following aspects: 1. Preprocess of electrocardiosignal. A simple method was introducedto remove baseline drift for the preprocessing of electrocardiosignal. 2. Characteristic points location of electrocardiosignal. R wave wasdetected by wavelet transform method. Then T-wave detection intervalwas determined with a simple method and the T-wave apex was detectedin this interval through wavelet transform. 3. Detection of TWA. TWA detection was realized by fusing singularvalue decomposition and enhanced spectral method after T-wave matrixwas constructed. The detection results were assorted and analyzed tovalidate the algorithm. Compared with previous methods, the present T-wave detectionmethod shows advantages as follows. T-wave detection interval wasdelineated to avoid interfernce from heart-rate and P-wave. Slope wasintroduced as a new criterion on the basis of threshold criterion to give amore accurate T-wave apex detection. For TWA detection, the singularvalue decomposition is able to remove noise and give cleanelectrocardiosignal. Enhanced spectral method is suitable for thedetection of non-stationary signal. Fusing these two methods improvedenormously the accuracy of TWA detection.â– ...