A Comparative Study of Heart Sound Denoising Using Adaptive Line Enhancer with LMF Algorithm

This paper presents a comparative study of heart sound denoising techniques by employing the Adaptive Line Enhancer (ALE) with the Least Mean Fourth (LMF) algorithm. Three variations of the ALE using LMF are developed and evaluated. The first approach replaces the conventional Least Mean Square (LMS) algorithm with LMF in the ALE structure. The second approach adopts a transform domain framework, and a Hadamard domain LMF is developed and used for ALE. Since the transform domain LMF requires the FIR filter order to match the dimension of the transformation, a novel structure of the transform domain ALE is developed, allowing the FIR filter order to be independent of the transform dimension. The study evaluates the performance of the ALE with LMF algorithm in both domains, alongside the proposed new structure. Simulated tests, focusing on steady-state behavior as the key performance index, demonstrate the superior performance of the newly developed transform domain ALE, surpassing both the time domain ALE with LMS and LMF algorithm and the conventional transform domain approach.