Compressão de sinais eletrocardiográficos baseada na transformada cosseno discreta.

Abstract

The extensive use of digital electrocardiograms produces large amounts of data and thus efficient techniques to compress electrocardiographic signals are important in order to reduce the amount of data to be recorded or transmitted. Eventual distortions introduced by the compression process must be carefully controlled, so that they do not lead to erroneous diagnostics. The present work proposes some solutions to the problem of electrocardiographic signal compression, presenting a new electrocardiogram compressor based on optimum dead-zone quantization of discrete cosine transform coefficients. In the proposed method, the signal to be compressed is partitioned in blocks of fixed size, which are then transformed. Each DCT block is quantized using a quantization vector and a threshold vector specifically determined for each signal. These vectors are defined, via Lagrange multipliers, so that the estimated entropy is minimized for a given distortion in the reconstructed signal. The proposed optimization strategy is an adaptation for electrocardiograms of a technique previously applied for image compression. The algorithm ends up with an entropy coding stage, for which five variants are described. Three of theses variants use arithmetic coding and the others use Golomb-Rice coding. The Percent Root-Mean-Square Difference (PRD) was adopted to measure the distortion in the reconstructed signals. To assess the performance of the proposed compressor, 2-minute sections of all 96 records of the MIT-BIH Arrhythmia Database were compressed at different PRD values, and the corresponding compression ratios (CR) were computed. Traces of original and processed signals are also shown, allowing visual inspection of the quality of the reconstruction. Furthermore, the effect of block size on the compromise between CR and PRD was empirically tested. An average CR of 11.3:1 was achieved for PRD equal to 2.5%, with the best-performance entropy coding variant. Experiments with records from the MIT-BIH Arrhythmia Database used in the literature allowed direct comparisons with various compressors. The comparisons revealed that the CR-PRD compromise of the proposed method is equivalent or superior to that of the best electrocardiogram compressors known.