The Discrete wavelet transform has great capability to analyse the temporal and spectral properties of non stationary signal like ECG. In this paper, we have developed and evaluated a robust algorithm using multiresolution analysis based on the discrete wavelet transform (DWT) for twelve-lead electrocardiogram (ECG) temporal feature extraction. In the first step, ECG was denoised considerably by employing kernel density estimation on subband coefficients then QRS complexes were detected. Further, by selecting appropriate coefficients and applying wave segmentation strategy P and T wave peaks were detected. Finally, the determination of P and T wave onsets and ends was performed. The novelty of this approach lies in detection of different morphologies in ECG wave with few decision rules. We have evaluated the algorithm on normal and abnormal beats from various manually annotated databases from physiobank having different sampling frequencies. The QRS detector obtained a sensitivity of 99.5% and a positive predictivity of 98.9% over the first lead of the MIT-BIH Arrhythmia Database.

Efficient compression reduces memory requirement in long term recording and reduces power and time requirement in transmission. A new compression algorithm combining Linear Predictive coding (LPC) and Discrete Wavelet transform is proposed in this study. Our coding algorithm offers compression ratio above 85% for records of MIT-BIH compression database. The performance of algorithm is quantified by computing distortion measures like percentage root mean square difference (PRD), wavelet-based weighted PRD (WWPRD) and Wavelet energy based diagnostic distortion (WEDD). The PRD is found to be below 6 %, values of WWPRD and WEDD are less than 0.03. Classification of decompressed signals, by employing fuzzy c means method, is achieved with accuracy of 97%.