Research And Application Of Medical Image Assisted Decision Based On Information Fusion

Breast cancer is a high-risk cancer in women,and the World Health Organization GLOBOCAN has released a 2018 cancer data report that estimates the morbidity and mortality of 36 cancers in 185 countries.The data indicates that there will be approximately 18.1 million new cancer cases and 9.6 million cancer deaths by 2018.Breast cancer morbidity and mortality in the female population were 24.2% and 15.0%,respectively,and breast cancer has become the most common type of cancer in women.This thesis studies the ultrasound-assisted diagnosis of breast ultrasound with reference to the data of breast ultrasound,and proposes a method for fusion of ultrasound image features and a method for decision fusion of multi-classifier classification results.At the same time,a breast ultrasound image-assisted diagnosis system is designed.It can be used to assist the diagnosis and treatment of breast ultrasound images,improve the reading speed of doctors,and relieve the tension between doctors and patients.The main research contents of this thesis include the following three parts:(1)Based on the image of the region of interest in the breast ultrasound image,a method is proposed to extract the morphological features and texture features and to fuse the features based on this.The method pre-processes the breast ultrasound image to extract the morphological features and texture features of the image;the extracted features are fused by the feature fusion method,in order to verify that the features obtained after fusion have better diagnosis for breast cancer.The diagnostic effect is classified and predicted by the classification algorithm for the unfused features and the merged features.(2)Based on the characteristics obtained after feature fusion,the information fusion method based on decision layer is proposed based on multiple classification methods.This paper adopts naive Bayes algorithm,KNN algorithm,SVM algorithm and decision tree algorithm.After the classification algorithm is classified,the information fusion based on the decision layer is performed,and the final classification result is output.And the comparative analysis is designed to verify the classification effect of the decision and the classification effect of the four classification algorithms.(3)Design and implement a breast cancer image-assisted diagnosis and treatment system The system is mainly divided into five modules: The login module,the account setting module,the administrator module,the role management module and the image management module enable the doctor and the patient to view and analyze the breast ultrasound image,and provide reference value for breast cancer diagnosis,and improve the doctor's The speed of reading can effectively alleviate the increasingly tense doctor-patient relationship.