Comparative Design and Performance Evaluation of IIR Digital Filters for Biomedical Signal Processing Using MATLAB and Python

Signals encountered in biomedical applications are often corrupted by noise such as baseline drift, power-line interference and random disturbances, which can distort important signal characteristics. To investigate noise-reduction performance, this study presents a comparative design and implementation of Infinite Impulse Response (IIR) digital filters for biomedical signal processing in both MATLAB and Python. A synthetic composite signal comprising multiple sinusoidal components (50 Hz and 150 Hz) combined with additive white Gaussian noise (AWGN) was generated and used as the test input. Several IIR filters, namely Chebyshev Type I, Chebyshev Type II, Elliptic, and Bessel, were designed and implemented on both computational platforms. The performance of these filters was evaluated using key metrics, including signal-to-noise ratio (SNR), waveform preservation, computational efficiency, and ease of implementation. Results show that all filters improved signal quality to varying extents, with each exhibiting distinct trade-offs between selectivity and signal distortion. MATLAB provided superior visualization and built-in signal-processing capabilities, facilitating rapid implementation and analysis, while Python offered greater flexibility, scalability, and the advantages of an open-source ecosystem. The Elliptic filter exhibited the sharpest transition band among all filters evaluated with stopband attenuation -90 for MATHLAB and -85 for Python, making it the most effective in terms of frequency selectivity and noise reduction but with ripples in both the passband and stopband but introduce substantial distortions in the biomedical signal waveform. while, Bessel filter demonstrated the most consistent performance in preserving the signal's morphological characteristics with phase deviation 4-6 degrees across 0.5-40 Hz for MATLAB and 3-5 degrees for Python. Although this study is motivated by applications in Electrocardiogram (ECG) signal processing, the evaluation was conducted using a synthetic test signal. Therefore, validation using real ECG data is recommended for future work to confirm clinical applicability. Overall, the findings highlight the effectiveness of IIR filters in noise reduction and their relevance to biomedical signal processing tasks.