CSD Optimized DWT Filter for ECG Denoising

The biomedical ECG signal represents the electrical event of the contractile activity of the heart. The alteration in the regular wave shape of ECG is used to identify life-threatening cardiovascular diseases. The origins of artifacts in biomedical signals are enormous, including the motion of the subject itself. Moreover, the potential of these artifacts in deteriorating the performance of signal processing algorithms is also severe. The removal of low-frequency artifacts, such as baseline drift and motion artifact, is focused in this work as these artifacts may lead to false predictions of the heart diseases. In this work, the baseline and motion artifacts cancellation are attempted with the various discrete wavelet transform (DWT) based methods. The MIT-BIH and noise stress databases are used for various wavelets based methods evaluation, and SNR comparisons with the existing methods. Moreover, the canonic sign digit (CSD) implementation of DWT filter banks is carried out for hardware optimization and compared with the existing implementation. The proposed CSD implementation proves that it takes less hardware and gives higher operating frequency than the direct form implementation.