Open Access

ARTICLE

An Improved Lung Sound De-noising Method by Wavelet Packet Transform with Pso-Based Threshold Selection

Qing-Hua He^a, Bin Yu^b, Xin Hong^a, Bo Lv^a, Tao Liu^b, Jian Ran^b, Yu-Tian Bi^a

^a State Key Laboratory of Trauma, Burns and Combined Injury, Daping Hospital, Surgery Institute of the Third Military Medical University, Chongqing 400042, China;
^b College of Communication Engineering, Chongqing University, Chongqing 400044, China

* Corresponding Author: Yu-Tian Bi,

Intelligent Automation & Soft Computing 2018, 24(2), 223-230. https://doi.org/10.1080/10798587.2016.1261957

Abstract

Lung abnormalities and respiratory diseases increase with the development of urban life. Lung sound analysis provides vital information of the present condition of the pulmonary. But lung sounds are easily interfered by noises in the transmission and record process, then it cannot be used for diagnosis of diseases. So the noised sound should be processed to reduce noises and to enhance the quality of signals received. On the basis of analyzing wavelet packet transform theory and the characteristics of traditional wavelet threshold de-noising method, we proposed a modified threshold selection method based on Particle Swarm Optimization (PSO) and support vector machine (SVM) to improve the quality of the signal, which has been polluted by noises. Experimental results show that the recognition accuracy of de-noised lung sounds by the improved de-noising method is 90.03%, which is much higher than by the other traditional de-noising methods. Meanwhile, the lung sound processed by the proposed method sounds better than by other methods. All results make it clear the modified threshold selection can obtain a better threshold vector and improve the quality of lung sounds.

---

Keywords

---