An ECG Signal Compression Technique based on DCT and Evolutionary Algorithms

In this manuscript, a new ECG signal compression method is proposed with adaptive reconstruction quality. It is based on discrete cosine transformation (DCT) and evolutionary techniques, where DCT is used for the decomposition of original ECG signal. The transformed coefficients are then thresholded using the optimized threshold level. As reconstruction is an essential step in the development of a compression algorithm, it is made adaptive using target-based reconstruction error. Different optimization techniques are utilized for optimizing the threshold level. The exploited techniques in this work are particle swarm optimization algorithm (PSO), flower pollination algorithm (FPA), artificial bee colony algorithm (ABC), and cuckoo search algorithm (CS). The encoding of thresholded coefficients is performed by applying the modified run-length encoding (MRLE) technique. The compressed signal is decoded using modified run-length decoding (MRLD), followed by inverse discrete cosine transformation (IDCT). The performance measures used to test the efficiency of this algorithm are compression ratio (CR), percentage root-mean-square difference (PRD), signal-to-noise ratio (SNR), and quality score (QS). This algorithm focuses on achieving an excellent reconstruction quality along with a good compression ratio. The results show that the ECG signal is compressed effectively with PRD ranging from 1.02 % to 4.95 %, indicating excellent reconstruction.