ECG data compression using the discrete cosine transform (DCT)

Orthogonal transforms provide alternate signal representations that can be useful for electrocardiogram (ECG) data compression. The goal is to select as small a subset of the transform coefficients as possible which contain the most information about the signal, without introducing objectionable error after reconstruction. With a sampling rate of 1 kHz, more than 99% of the power in the DCT is contained within the first 20% of the coefficients. Despite this result a 5:1 compression ratio cannot be obtained by merely substituting zero for the remaining 80%. The coefficients after the first 20%, although of relatively small magnitude, preserve the DC integrity of the signal. Approximating these components as zero leads to introduction of spurious sinusoidal terms in the reconstructed signal. A quantization technique to better approximate the higher coefficients has been used to obtain an accurate representation of the signal. The tradeoffs between accuracy, speed, and compression ratio are discussed