Fast parallelized algorithm for ECG analysis M RIZZI, M D’ALOIA and B CASTAGNOLO Dipartimento di Elettrotecnica ed Elettronica, Politecnico di Bari via E. Orabona, 4 – 70125 Bari ITALY rizzi@deemail.poliba.it Abstract: - A new approach based on the adoption of wavelet transforms is presented for the R point localization in ECG signals. The conceived real time signal processing technique, which uses a fast parallelized algorithm, has been evaluated adopting the standard MIT-BIH Arrhythmia database which includes specially selected holter recordings with anomalous but clinically important phenomena. In the procedure a soft thresholding technique is applied to dyadic scales in which the ECG signal is decomposed. Therefore, noise contribution is reduced and then signal is easily reconstructed in the time domain for further processing. Moreover, the tool analyzes the signal on different level wavelet representation at the same time showing a great parallelism degree and an enhancement in processing time. To evaluate the algorithm noise immunity, the MIT-BIH Noise Stress Test Database has been adopted containing baseline wander, muscle artifacts and electrode motion artifacts as noise sources. The obtained performance shows the method validity in terms of algorithm speed up and characteristic parameter values. In fact, sensitivity and positive predictivity values of about 99.8% are obtained with a detection error rate of about 0.4%. Moreover, the conceived procedure gives satisfactory results also for ECG signals heavily corrupted by noise Key-Words: - ECG, QRS, wavelet transform, parallel filter bank, signal processing, parallel computing 1 Introduction Electrocardiography is an important tool in diagnosing the heart condition and consequently in discovering many cardiac diseases The Electrocardiogram (ECG) is a non invasive graphic record representing direction and magnitude of the heart electrical activity that is generated by depolarization and repolarization of the atria and the ventricles. Therefore it provides valuable information about the functional aspects of the heart and of the cardiovascular system and consequently is widely used for cardiac disease diagnostic and for urgent treatments of ill patients [1] In fact, an early detection of heart abnormalities can prolong life and enhance its quality adopting suitable cures. Most of the clinically useful informations for cardiac state health are indicated by the ECG shape such as intervals and amplitudes of the signal. The QRS detection is the most important task in ECG signal analysis systems. In fact, after the QRS identification, the heart rate may be calculated and other parameters can be examined to avoid serious pathologies such as ischemia. For example, accuracy of RR intervals is crucial for reliable heart rate variability (HRV) analysis, which is widely considered to provide a simple non-invasive and quantitative assessment of cardiac-autonomic function in health and in disease states [2]. HRV analysis has been increasingly recognized as a useful tool for understanding autonomic regulation during sleep as well as patient screening in obstructive sleep apnea syndrome, congestive heart failure and other disorders [3], [4], [5]. Due to the non-stationary behaviour of biological signals, disease symptoms may not show up all the time but would manifest at certain irregular intervals during the day. Therefore, the study of ECG pattern by analysts may have to be carried out over several hours (such as nigh-time data or 24 hours holter monitoring) with an high probability of missing vital informations. Therefore, computers based analysis is advisable. The implementation of a procedure for detection of P wave, QRS complex and T wave is a difficult task due to the time varying behaviour of the human body and consequently all processing methods should change their state during measurement. Moreover, noise contamination, due to baseline drifts changes, motion artifacts and muscular noise, is frequently encountered. Classical QRS detectors are composed of a preprocessor stage for QRS complex emphasizing and a decisional stage for QRS enhanced signal thresholding. The ECG signal is first band-pass WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE M. Rizzi, M. D'Aloia, B. Castagnolo ISSN: 1109-9518 210 Issue 8, Volume 5, August 2008