POLYNOMIAL APPROXIMATION FOR T WAVE PARAMETER RECOGNITION IN ECG PROCESSING

To succesfully implement a telemedical system for diagnostic purposes it is necessary to verify the diagnostic value of the decision algorithms used to detect life threatening situations. ECG analysis is a useful tool for obtaining information about the overall patient condition, especially for the circulatory system. Proper recognition cannot be performed without creation of proper models, The first step is signal filtration and data preparation, followed by parameter extraction, comparison with the model and diagnosis presentation. Each of these steps reqires a certain approach to minimize the error. Proper filtration needs to be performed. Then, the QRS complex is detected and rythm is calculated. Afterwards, the remaining waves are detected. To be able to perform valuable time dependencies it is necessary to exactly mark the beginnings and ends of intervals. The proposed method is based on opproximating the signal around the wave with a polynomial of a certain degree. This allows detection of inflection points corresponding to the borders of the wave. The method was applied to a set of ECG signals recorced during rest and activity, the results are presented and discussed.