Classification of normal and widened QRS complexes in ambulatory ECG

The adaption of an ECG pattern recognition method for biomedicai device control is presented. It is optimized by using medical experience at a high degree. The real time processing is performed by the continuous evaluation of the Walsh spectrum of overlapping ECG signal sections. Therewith QRS complexes are identified and widened complexes are seperated from normal ones. A specifity of 993 p. c. was yielded. We conclude that high expenditure in developing algorithms does pay if shape classification is required by physicians.