Research On The Measurement Of Biological Tissue Optical Property Parameters By LSSVM

Biological tissue optical properties parameters can reflect the physiological state and pathology structure of biological tissue, thus has great significance on clinical diagnosis and treatment of diseases. Such as biological tissue optical parameters in tissue type identification, qualitative analysis of tumor and surgical navigation and other areas of research have important application value. In recent years, the use of laser technology, non-destructive measurement of biological tissue parameters have been well studied and widely used. For the current study to measure the optical properties of monolayer tissue parameters have been more in-depth, scholars have done a lot of work. However, the organizational structure of biological was complexity and diversity, research limited to single-depth cannot explore its internal structure, and multi-organizational model may be more realistic. The purpose of this study is the use of laser technology combined with LS-SVM algorithm to make theory exploration and experimental verification for the non-destructive measurement of double biological tissue optical parameters, and got some conclusions.For how photons propagate in the double biological tissue, we use Monte Carlo Simulation algorithm for the propagation of a detailed analysis and presentation, and through Monte Carlo algorithm to simulate the diffuse reflectance distribution on the surface of double layer of biological tissue, study the relationship between four optical parameters of double layer biological tissues and diffuse reflectance distribution on the surface of double layer of biological tissue. Using least squares support vector machine(LSSVM) algorithm, the diffuse reflectance distribution simulated by Monte Carlo algorithm as its input, the corresponding optical parameters as its output, without considering the specifics of how the photon propagation in double tissue, through training or fitting establish a nonlinear relationship between optical parameters and diffuse reflectance distribution, that is least squares support vector machine regression model. Putting the diffuse reflectance distribution intoregression model, we can fast and accurate measurement of the optical parameters.The impact of the number of samples to train the model and the way of the model to find optimizing parameters on the LSSVM generalization ability are discussed, and finally build the best least squares support vector machine predictive models of double biological tissues. The use of He-Ne laser emits a laser beam of 632.8nm to irradiate double biological tissue model, and CCD camera measure diffuse reflectance distribution on its surface. And thus experimentally study the measurement of optical parameters of the double biological tissue, diffuse reflectance distribution on the surface of body forearm also measured in the experiment.The results show that the generalization ability of least squares support vector machine increase with the number of training samples, and different parameters optimization method has some effect on the predicted result of least squares support vector machine. Simulation results show that LSSVM can achieve accurate measurement of the double biological tissue optical parameters, and prediction accuracy will be higher by sequence measurement of optical parameters of double biological tissues. The experimental measurement of the optical parameters of the two-layer model of biological tissue, and reached certain prediction accuracy. And the measurement results of the optical parameters of the human forearm are in good agreement with the actual values.