A transform-based ECG compression using discrete tchebichef moments with global and local error measures as control

This paper presents an alternative compression technique for Electrocardiogram (ECG) data compression using discrete Tchebichef moments instead of conventional methods such as Discrete Cosine Transform and Wavelet Transform. It is desired to have an efficient compression without significantly distorting the clinical features of the signal. However, assessment of reconstructed signal quality remains an open problem today. Therefore, this paper proposed a technique to ensure that the reconstructed signal quality remains diagnostically usable by implementing both local and global error measures as the controller, guaranteeing signal quality. Furthermore, detection of QRS beat of the input signal is not needed in this technique. The performance of this technique was assessed from simulations done on the records of the MIT-BIH Arrhythmia Database. Results show that the proposed technique was able to achieve good compression ratio (CR) while preserving clinical information with low distortion on reconstructed signals with an outcome of CR of 6:1 with percent root mean square difference (PRD) of 0.4% and Maximum Amplitude Error (MAX) of 20. This paper shows that the proposed technique is able to produce a reconstructed signal where its clinical diagnostic information remains intact, regardless of different types of ECG leads.