Compression of ECG signals exploiting correlation between ECG cycles

In this work a novel technique is presented for the compression of the ECG signals. Experimentally it is observed that there exists a strong correlation between different cycles of the ECG signals of a particular person. To utilize this feature with the aim of compressing the ECG signals, a sample cycle is chosen from the ECG record which has to be compressed. Then differences between the sample cycle and the other cycles are calculated and, it is, therefore, to be expected that these difference signals have low energy due to the strong correlation exists between the different cycles of the ECG signals. For the compression of ECG signals Discrete Cosine Transform (DCT) is applied to the sample cycle and the difference signals and, the number of DCT coefficients are adaptively calculated in order to achieve a good compression ratio (CR) as well as a good percent root mean square difference (PRD). It is also possible to obtain low correlation between the sample cycle and a particular cycle where the patient exhibits different ECG pattern than a usual ones. The proposed method also deals with these abnormalities to ensure high quality of the reconstructed ECG signals. Finally, our presented method is compared with a recently published technique using some records from the MIT-BIH database and the proposed method shows promising improvement in CR allowing a very low distortion.