使用高階統計徑向基底類神經網路於腸鳴音信號強化

Abstract

胃腸道疾病常見的診斷方式是以非侵入式的腸音聽診作為診斷的依據，但腸音的發生沒有規律性且振幅不明顯，因此容易受到背景環境的干擾，提高診斷困難度及降低準確性。此外，腸音的頻帶範圍也容易與背景環境雜訊重疊，導致數位濾波器無法達到良好的腸音信號強化。本論文提出使用高階統計徑向基底類神經網路來達到腸鳴音信號強化的功能，高階統計具有抑制高斯雜訊及對稱分布之非高斯雜訊的特性，能有效保留具有非高斯且非對稱分佈特性的腸鳴音，將高斯雜訊及對稱分布之非高斯雜訊抑制，達到腸鳴音信號強化的目的。本論文的模擬與實驗結果顯示，高階統計徑向基底類神經網路可在穩態與非穩態的高斯雜訊中有效強化受干擾的腸音，因此高階統計徑向基底類神經網路可作為強化腸鳴音的有效技術。

The auscultation of bowel sounds provides a non-invasive method to diagnosis gastrointestinal motility diseases. However, bowel sounds can be easily contaminated by background noises, and the frequency band of bowel sounds is easily overlapped with background noise. Therefore, it is difficult to enhance the noisy bowel sounds by using precise digital filters. In this study, higher-order-statistics (HOS)-based radial basis function (RBF) network was proposed to enhance noisy bowel sounds. Higher order statistics technique provides the ability to suppress Gaussian noises and symmetrically distributed non-Gaussian noises due to their natural tolerance. Therefore, the influence of additional noises on the HOS-based learning algorithm can be reduced effectively. The simulated and experimental results show that HOS-based RBF can exactly provide better performance for enhancing bowel sounds under stationary and non-stationary Gaussian noises. Therefore, HOS-based RBF can be considered as a good approach for enhancing noisy bowel sounds.