Development Of Near Infrared Diffusion Correlation Tomography For Breast Tumor Detection

In recent years,a new clinical imaging diagnosis and treatment technology called near-infrared diffuse light correlation tomography（DCT）has developed rapidly.The principle of this technology is microvascular blood flow imaging through light field time autocorrelation function.In this paper,DCT technology is developed and applied to breast tumor detection,and the hardware equipment suitable for breast examination is designed.Through statistical method and support vector machine（SVM）classification method,the characteristics of blood flow values of different populations（sick and healthy）are analyzed,and the clinical effectiveness of DCT technology is evaluated in many aspects.The main work of this paper is as follows:In terms of hardware equipment,this paper designs an array optical switch to share the light source and detector time-sharing,optimizes the distribution of the light source detector optical fiber on the surface of the measured tissue,and makes a bowl shaped special probe suitable for breast tissue,which can realize large field blood flow imaging while reducing the cost.In terms of extracting blood flow value,this paper extracts it through the n-order linear algorithm proposed by our research group,which can make full use of autocorrelation data and is not affected by tissue volume and geometry.In addition,the n-order linear algorithm is essentially realized by iterative linear regression,which also determines the real-time and robustness of the algorithm.In terms of statistical analysis,based on the clinical data of breast tumors extracted by DCT technology,this paper analyzes and obtains the blood flow value characteristics of the four groups between benign and malignant affected sides,between benign and malignant opposite sides,between malignant affected sides and opposite sides,and between benign affected sides and opposite sides,so as to evaluate the accuracy of DCT technology in measuring tumor blood flow.The DCT imaging system and algorithm developed in this study can detect the changes of blood flow caused by breast tumor and calcification.Specifically,the blood flow value of breast containing tumor is higher than that of the opposite side,and the blood flow of calcified tissue is lower than that of the opposite side.When the blood flow of the left and right breast is seriously asymmetric,it indicates that the probability of breast disease increases.In addition,cysts,tumors or calcifications in biological tissues can be located by DCT technology.Clinical trials also confirmed that DCT technology has the potential to be used in clinical breast cancer screening.At present,there is no report on DCT blood flow analysis of data analysis for different populations（healthy people and patients）.This paper fills this gap.In this paper,SVM trains the breast blood flow data of 55 subjects,and the accuracy of the training model is more than 90%,and the classification accuracy of the test set is more than 80%.It provides a case for combining machine learning method and DCT technology.