Methodology Article
Research on Feature Extraction and Recognition of CHD Heart Sound Signal Based on S Transform
Zeng Zheng,
Pan Jiahua,
Cai Guanghui,
Yang Hongbo,
Wang Weilian
Issue:
Volume 1, Issue 1, March 2019
Pages:
1-7
Received:
28 October 2018
Accepted:
30 November 2018
Published:
20 December 2018
Abstract: Auscultation is the main means in the early diagnosis of congenital heart disease. The research on analysis and classification of CHD heart sound has important significant and can be used in clinical diagnosis of CHD. It will be helpful for machine auxiliary diagnosis. In this work, a feature extraction and recognition algorithm based on S transform was put forward, including the heart sound signal preprocessing, feature extraction and classification recognition. In heart sound preprocessing, denoising, envelope extracting, and segmenting were done to obtain the each cycle of the heart sound. Some of time-frequency analysis methods such as STFT, Wigner-Ville, wavelet transform, and S transform were discussed and analyzed. Then S transform and wavelet transform were used for feature extraction of each cycle. Finally, the BP neural network was used as classifier to recognize the normal and the abnormal heart sound signal. All cases of CHD heart sound used in this experiment came from heart sound data base sampled in clinic at Yunnan Fuwai Cardiovascular Disease Hospital. 361cases heart sounds including CHD and healthy heart sound were selected randomly for analysis. The result showed that recognition rates of S transform method and wavelet transform method were 80.4% and 76% respectively. S transform has a better recognition than wavelet transform.
Abstract: Auscultation is the main means in the early diagnosis of congenital heart disease. The research on analysis and classification of CHD heart sound has important significant and can be used in clinical diagnosis of CHD. It will be helpful for machine auxiliary diagnosis. In this work, a feature extraction and recognition algorithm based on S transfor...
Show More
Fitting a Cardiac Cycle of Left Ventricular Blood Pressure Curve
Sun Bing,
Tian Feng,
Liu Li,
Zhang Yan,
Qi Jingai
Issue:
Volume 1, Issue 1, March 2019
Pages:
8-14
Received:
10 December 2018
Accepted:
16 January 2019
Published:
20 February 2019
Abstract: The left ventricular blood pressure curve of one cardiac cycle of an actual heart is known, and its blood pressure curve is fitted by the fuzzy control of three elements. According to the characteristics at the time points on the curve of blood pressure, after divided into five sections, it could be fitted. When using a three-element-controlled fuzzy fit, the curve is expressed as y(x)=c2x2+c1x+c0, where ci is the constant to be solved, and because of the three unknown c2, c1, c0, there must be 3 constraints . If we know the three coordinate points on the distribution curve and slope m1 on the initial segment, When y (x)=c2x2+c1x+c0 is expressed in a matrix, then the inverse of it can be obtained. Thus, we can get the expression y=f(x)=f(x1, y1, x2, y2, m1). In the Maple software environment, entering the expression f(x1, y1, x2, y2, m1) and x1, x2, y1, y2, m1, then could get the y = f(x) expressions and plot their graphs. The fitting results of y= f(X) expressions for the 5 periods are shown in table 3 and after observed fitting effect of the three control elements, the visual effect on the curve is still well as possible. The maximum and the minimum blood pressure in the left ventricle are 120 and 1.5 mm Hg, respectively, and the resulting expression y = f(x) for each segment can be put into practical use for electrostatic excitation of aorta electrostatic pumps.
Abstract: The left ventricular blood pressure curve of one cardiac cycle of an actual heart is known, and its blood pressure curve is fitted by the fuzzy control of three elements. According to the characteristics at the time points on the curve of blood pressure, after divided into five sections, it could be fitted. When using a three-element-controlled fuz...
Show More
