Mueller polarization technology-based biological tissue structure classification system

Abstract

The invention provides a Mueller polarization technology-based biological tissue structure classification system, which comprises the following steps of: firstly, obtaining a Mueller matrix corresponding to a biological tissue to be classified and a Mueller polarization parameter combination of the Mueller matrix, and constructing a polarization characteristic matrix according to the statistic of the Mueller polarization parameter combination; and finally, performing joint evaluation on a polarization characteristic matrix which is obtained based on the Mueller polarization technology and is formed by a plurality of polarization characteristic quantities by a support vector machine to obtain the category of the biological tissue. According to the system, a plurality of scalar polarization parameters (scalar phase delay, scalar polarization degree, scalar depolarization, scalar bidirectional attenuation and the like) and a plurality of vector polarization parameters (vector phase delay, vector polarization degree, vector bidirectional attenuation and the like) are combined as key features to classify biological tissues; comprehensive qualitative analysis and accurate quantitative identification on the biological tissue microstructure are realized, the influence of random factors can be greatly reduced, and the classification accuracy is improved.

Description

technical field [0001] The invention belongs to the technical field of polarization measurement, in particular to a biological tissue structure classification system based on Muller polarization technology. Background technique [0002] Scattering occurs when light interacts with a medium, and the change in the polarization state of photons during the scattering process is closely related to the microstructure of the scattering medium. Most biological tissues are highly scattering media, but the polarization information originally carried by light will be lost after multiple scattering, which will affect the contrast and resolution of imaging. Muller polarization technology can suppress multiple scattering and lose the contribution of "diffused photons" to the image by properly screening the polarization state of photons, and improve the "ballistic photon" and "snake photon" with few scattering that maintain the original polarization. The role of photons, thereby improving ...

AI Technical Summary

Problems solved by technology

[0005] First, the polar decomposition method is traditionally used to obtain scalar polarization images, such as scalar phase delay, depolarization, bidirectional attenuation, linear phase delay, circular delay, linear depolarization, circular depolarization images, etc., and samples are obtained from a single scalar polarization parameter image Structural analysis can only observe the variation trend of the polarization parameter with the size of the tissue lesion and the approximate distribution of the tissue structure that affects the polarization parameter. For example, through the scalar phase delay image, it can only be observed qualitatively that the phase delay value varies with the degree of tissue lesion Different trends, or approximate distributions of collagen fibril structures that affect the phase delay can be observed, but the scalar phase delay map will reflect the vertically distributed fiber structure reflected by the horizontal phase delay parameter, and the fibers distributed along the 135° direction reflected by the 45° phase delay parameter. The structure and the helical structure of the fibers reflected by the circular delay parameters are mixed together, and the arrangement details of the respective fiber structures cannot be observed from the images, and the structural information provided is not comprehensive enough.

[0006] Second, the current quantitative analysis methods combined with Mueller polarization imaging technology include: 1. Calculate the statistical parameters (mean, median, standard deviation, deviation) of the intensity map (or spatial spectrum map) of the Mueller matrix and decomposition parameters degree, kurtosis, etc.), and then carry out statistical analysis. This method can achieve quantitative classification from a certain feature quantity under a single perspective, but it is greatly affected by random factors and cannot realize multi-parameter joint characterization. The classification accuracy rate depends on the sample size. 2. Combined with the neural network method based on machine learning to realize modeling and classification of tissue samples with different lesion degrees, this method can jointly represent multiple parameters, which is less affected by random factors and achieves a certain degree of quantification Classification, but also faces the limitation that the accuracy rate is determined by the sample size

Images