An implementation of the wavelet transform for ECG analysis

An approach to the wavelet transform has been developed using LabView, a block diagram programming language, to detect the different waves of the ECG for the detection of the ECG characteristic waves. This method decomposes the ECG originally into 5 transformed signals at 5 different scales and makes use of scales 2 to 4 of the wavelet transform. In this method only scales 2 to 4 are used to characterise the ECG signal as these prove to give optimum results. A derived criterion is applied on a combination of such scales to determine the location of the peaks of the QRST waves, in addition to the start (onset) of the Q wave and end (offset) of the T wave for the posterior determination of the QT interval. The method has been tested against the well known MIT-BLH database and a QRS sensitivity above 99.95% has been achieved for some records.