Minging And Classification Of Photoacoustic Signals For Blood Pressure

Photoacoustic imaging technology is a noninvasive and nonionized biomedical imaging method developed in the last 20 years.The photoacoustic signal produced by irradiating an absorber with a short pulse laser,provides both a high contrast of pure optical imaging and a high resolution of pure acoustic imaging,in which an accurate and noninvasive functional imaging of biological tissue is obtained.Photoacoustic imaging technology can be used to study the morphological structure of biological tissue and analyze the related pathological and physiological characteristics.In this dissertation,according to the structural characteristics of human cervical artery,the light absorption distribution of tissue,the phantom experiment of artery and the method of continuous noninvasive intelligent diagnosis of blood pressure based on photoacoustic measurement are presented systematically.The photoacoustic signals of human blood pressure were mined and calibrated.The main contents and innovations are mainly reflected as follows:Firstly,an optical model was constructed according to the tissue structure.The Monte Carlo simulation method was used to employ the fluence rate distribution of different depths with different wavelengths of exciting light source,in order to guide the optimal light source in photoacoustic blood pressure measurement system.Combined with experimental study,the light absorption differences of several absorbers at different depths are presented for improving a detection accuracy and determining an applicable imaging depth.Secondly,based on the principle of photoacoustic imaging,a continuous photoacoustic blood pressure measurement system was designed.The hardware and related software design of the system had been completed.And the feasibility of the system had been verified by experiments,including the spatial resolution of the system,which provides a reference basis for optional experimental parameters.Thirdly,according to the time-domain model of photoacoustic signal,the relationship between arterial pressure and the photoacoustic signal was explored.The correlation between blood pressure-signal peak-peak interval,blood pressure-tube diameter,blood pressure-signal strength and the spatial distribution of elastic properties in tissue were analyzed.According to the real-time acquisition of vascular diameter information,the morphological changes of arterial vessels can be tracked quickly,effectively and accurately,which provides a theoretical basis for noninvasive continuous measurement of blood pressure.At last,a deep convolution neural network classifier for blood pressure photoacoustic signal was constructed.By using the intelligent diagnosis method,a photoacoustic signal established by Hilbert transform and a two-dimensional image of photoacoustic array were used as the network input.The residual network was used as a classifier to realize the blood pressure photoacoustic signal classification.A high classification accuracy and the photoacoustic signals of blood vessels under different pressures were accurately identified.