Noise reduction using anisotropic diffusion filter in inverse electrocardiology

Filtering has been widely used in biomedical signal processing and image processing applications to cancel noise effects in signals recorded from the body. However, it is important to keep the desired characteristics of the physiological signal of interest while suppressing the noise characteristics. In this study, we used anisotropic diffusion filter (ADF) to cancel the noise on the body surface potentials measurements (BSPM) with the goal of improving the corresponding solutions of the inverse problem of electrocardiology (ECG). ADFs have been applied to image processing and they have the advantage of preserving sharp edges while rejecting the noise, thus we have chosen ADFs instead of more conventional filtering techniques. We used unfiltered and filtered BSPMs to estimate the epicardial potential distributions. We compared Tikhonov regularization results when the data included measurement noise and geometric errors. In both cases, filtering of BSPMs using the ADF improved our solutions.