TC-ADPCM coding for ECG signals

Data compression of electrocardiogram (ECG) signals is necessary for the storage and transmission of digitised ECGs. A new coding algorithm termed the TC-ADPCM (Transform Coding-Adaptive Differential Pulse Code Modulation) algorithm is proposed in this paper for ECG data compression. This algorithm combines the benefits of two well-known techniques used for data compression, namely, the discrete cosine transform (DCT) transform and prediction coding. The proposed algorithm exploits the correlation between successive samples as well as the correlation between the DCT coefficients of successive beats of ECG. The correlation between the successive samples is removed by applying the DCT and the correlation between the DCT coefficients is removed by using an adaptive prediction coding technique. The results obtained by using the proposed algorithm are compared with those obtained by using the DCT and adaptive prediction coding methods independently. All the algorithms are tested by using the ECG signals obtained from the MIT-BIH ECG compression database. The comparison is done by computing the compression ratio and percent RMS error, and also by the visual inspection of the reconstructed signals. It is shown that higher compression ratios can be achieved by using the proposed algorithm with a clinically acceptable quality of reconstruction. The new algorithm also compares well with compression schemes based on methods other than DCT and prediction coding. The proposed method is shown to be reliable for ECG signals digitised at lower sampling frequencies.