Partition of Medical Signals with a Two Band Adaptive Filter Bank

Cetin Tas I., ARICA S.

26th IEEE Signal Processing and Communications Applications Conference (SIU), İzmir, Türkiye, 2 - 05 Mayıs 2018 identifier identifier


Medical signal processing is often used to detect diseases. Segmentation, filtering, categorization are basic operations in this field. In this study; a two-band quadrature mirror orthogonal filter bank structure is used to emphasize and extract signals like QRS wave from electrocardiography, evoked potential from electroencephalography signal. One branch of the filter bank is adopted to filter out a target signal. This is done by minimizing error energy at the output of the filter bank system. The approach is run and tested for a QRS signal and a synthetic evoked potential signal. The common orthogonal wavelet filters (Daubechies, Symlet, Coiflet) and designed filter are compared in terms of mean squared error (MSE). The designed filter attains the target signal with the minimum MSE and outperforms common classical wavelet filters.