ECG data compression with wavelet and discrete cosine transforms

When applying transform techniques in data compression, an efficient approximation of the original signal using fewer transform coefficients is desired. The Discrete Cosine Transform (DCT) has been an effective tool in such applications. It decomposes a signal into a set of sinusoidal waveforms that are global in time. The DCT is not as efficient for signals with only local variations. The Wavelet Transform (WT) is a new technique that can decompose a signal into a set of small waveforms called wavelets. These wavelets possess local supports in the time domain, which makes the WT suitable for representing signals with local variations. Even though the ECG is dominated by low frequencies, its QRS complex exhibits strong localized characteristics. In this study, the effectiveness of the two transforms in compressing ECG data is investigated. Having noted the weakness of each transform, the two techniques are combined in two ways to compress the ECG data. It is observed that with the new techniques, better visualization quality can be achieved with the same total number of transform coefficients.