基于肺音谱图Hough变换的喘鸣音识别方法

发布时间：2018-05-08 07:45

本文选题：喘鸣音 + Hough变换　；参考：《东北大学学报(自然科学版)》2017年11期

【摘要】：提出了一种基于Hough变换从肺音的STFT光谱图中检测喘鸣音的方法.这一方法先对采集的数字肺音数据的STFT谱图进行ROI区域的截取,再利用Canny算子进行图像边缘检测,最后基于Hough变换数据的分析来自动识别喘鸣音.临床分析的数据包括临床采集的肺音和国际上共享的肺音文件.Hough变换检测方法在60例喘鸣音的检测中达到了87%的准确率,70例正常呼吸音的识别率达到74%.
[Abstract]:A method for detecting wheezing sounds from STFT spectra of lung sounds based on Hough transform is proposed. This method firstly intercepts the STFT spectrum of the collected digital lung sound data, then uses the Canny operator to detect the edge of the image. Finally, based on the analysis of the Hough transform data, the panting sound is automatically recognized. The data of clinical analysis include lung sound collected by clinic and lung sound file shared internationally. Hough transform method has achieved 87% accuracy in detecting wheezing sound in 60 cases and the recognition rate of normal respiratory sound in 70 cases has reached 74%.
【作者单位】：东北大学中荷生物医学与信息工程学院;辽宁中医药大学;辽宁中医药大学附属医院;
【基金】：国家自然科学基金资助项目(81273800)
【分类号】：R443.6;TN911.7

【相似文献】