Signal Process And Expert Diagnostic System Of Doppler Ultrasonic Signal

Transcranial Doppler is a non-invasive cerebrovascular disease inspection methods which uses ultrasonic Doppler effect to detect the intracranial brain artery at the end of the main hemodynamic and blood physiological parameters. It was introduced to our country in 1988, and was widely used. For TCD can penetrate skull without hurting it, and its simple operation and well reproducibility is excellent. More importantly, it can provide important hemodynamic information that other image technology cannot detect, therefore it has important significance in the evaluation of cerebrovascular disease as well as the disease diagnosis.At present, commonly used transcranial Doppler diagnostic methods are based on inspect results of transcranial Doppler instrument, and by the inspection results doctors can analyse and judge the disease by right of their understanding of the pathology knowledge and experience accumulated over the years. However, this TCD diagnostic method is influenced by subjective factors, and the diagnostic results using doctor's medical standards are closely related, which has more visible performance at small hospital and remote areas of hospital. It leads to high misdiagnosis rate of cerebrovascular disease. It has been hoped that there is an intelligent diagnosis methods which can get rid of human factors.This methods can get accurate and objective diagnosis results.The funding of this project was from Technology Office in Shandong Province. In this thesis, transcranial Doppler audio signal is used as the research object, transcranial Doppler signal characteristics are considered, the wavelet analysis theory and support vector machine are used as the theory base. How to receive Transcranial Doppler spectrum parameters, feature extraction from transcranial Doppler signal and diagnostic expert system are researched and implemented in this thesis.First of all, hardware platform of transcranial Doppler is given in this thesis, and transcranial Doppler audio signal can be got through it. The spectrum parameters are obtained from the maximum frequency curve which is extracted from collected audio signals. These sound spectrum parameters and feature data extracted from the maximum frequency curve are used as input data of classifier to carry out taxonomic research. BP neural network classifier and least squares support vector machine are discussed in this thesis, and their classification results are compared. Finally, on the basis of the above, Doppler diagnostic expert system is designed. At the same time, the existing Doppler signal analytic methods are set up at the foundation of spectrum, so it is easy to lose some characteristics of the signal. Based on this fact, the Doppler audio signals are processed directly to extract useful information in this thesis, and it has received, and satisfactory results are got. In addition, considering of the required long distance transmission of the doppler audio signals and doppler maximum frequency curve, a few compression technologies are compared in this thesis and high compression ratio can be got on the basic of compression effect.The integrated diagnostic platform of transcranial Doppler results is implemented in this thesis. Accurate and objective diagnosis to patients can be got. Design and implementation of Doppler diagnostic expert system can provide a useful basis for prevention and treatment of cerebrovascular disease.