Introduction: To realize noninvasive diagnosis and early diagnosis of coronary heart disease, the study proposes a new time-frequency method for analyzing heart sound signals. This method is based on Choi-Williams Distribution (CWD). Methods: CWD distribution is developed and modified from Wigner Ville distribution (WVD). To solve the problem of cross-term interference existing in WVD there is an improved version of WVD, called Choi-Williams Distribution (CWD), which introduces the smoothing window as the kernel function and deals with the time-frequency analysis of heart sound signal. Results: The improved method has good performance and can be implemented simply without much increase of operation complexity. Conclusion: In this paper, 21 cases of heart sound signals are acquired from the outpatients and hospitalized patients with coronary heart diseases. The research results of 21 cases show that the CWD method can be used to analyze heart sounds. It accurately identifies the 9 cases of heart sounds of health people and 12 cases of heart sounds of patients with coronary heart disease. Besides, the CWD displays obvious differences between heart sounds of healthy people and abnormal heart sounds. The contour line of heart sounds from healthy people shows the following characteristics: concise, columnar and non-divergence; while the contour line of abnormal heart sounds is divergent and has many columnar links. These research shows that CWD method can effectively distinguish heart sounds between healthy people and patients with coronary heart disease.
in this paper, near infrared spectroscopy techniques are used for the detection of meat in the process of corruption time, studied the feasibility of pork freshness level. And the qualitative analysis model is established based on the software OPUS. During the model establishment process, the kinds of the class of TVBN values are re-divided 5 from 3 using the SOM network clustering to better reflect level of freshness of meat. And to increase the accuracy of prognostication, the principal component analysis is used to reduce dimension except choosing the pretreatment method of the 13-point first derivative smoothing, and the result is that the rate of correct promote and the number of which of bias of predictive class is decreased.
With the gradually mature of hyper spectral image technology, the application of the meat nondestructive detection and recognition has become one of the current research focuses. This paper for the study of marine and freshwater fish by the pre-processing and feature extraction of the collected spectral curve data, combined with BP network structure and LVQ network structure, a predictive model of hyper spectral image data of marine and freshwater fish has been initially established and finally realized the qualitative analysis and identification of marine and freshwater fish quality. The results of this study show that hyper spectral imaging technology combined with the BP and LVQ Artificial Neural Network Model can be used for the identification of marine and freshwater fish detection. Hyper-spectral data acquisition can be carried out without any pretreatment of the samples, thus hyper-spectral imaging technique is the lossless, high-accuracy and rapid detection method for quality of fish. In this study, only 30 samples are used for the exploratory qualitative identification of research, although the ideal study results are achieved, we will further increase the sample capacity to take the analysis of quantitative identification and verify the feasibility of this theory.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.