Design And Implementation Of Thyroid Nodule Classification System Based On Circular Convolutional Neural Network

In the whole country and even in the world,thyroid diseases cause serious harm to human health.To address this problem,early detection and early diagnosis of thyroid cancer is an effective means to reduce the incidence and mortality of thyroid cancer.Accurate classification of thyroid nodules has become the focus of current research.In 2018,the latest cancer data of the National Cancer Center showed that the number of patients with thyroid disease in major hospitals increased year by year,so improving the diagnostic efficiency is an effective way to cope with the increasing number of patients.In the diagnosis process,the diagnosis of images often depends on the doctor's knowledge level and clinical experience,subjective factors,and it takes a lot of time.Therefore,the introduction of computer-aided diagnosis technology provides objective assistant decision-making for clinical diagnosis,which is conducive to reducing the diagnosis errors caused by doctor's personal subjective factors in medical diagnosis and helping doctors to obtain accurate diagnosis results.In this paper,the thyroid ultrasound image is taken as the research object,and the construction of thyroid ultrasound image classification model based on circular convolution neural network and the development of thyroid nodule classification system are realized.The system is divided into login module,data management module,image preprocessing module and image classification.Module.The main work of the thesis is as follows:(1)Overview of classification of thyroid ultrasound images.Analyze and introduce the source and treatment of clinical thyroid data,and briefly describe the current status of thyroid ultrasound image classification research,and study the importance of thyroid ultrasound image classification,and classify thyroid ultrasound images according to TI-RADS grading standards and doctors' guidance.A brief analysis of commonly used classification methods:association rules,support vector machines,and deep neural networks.(2)Construct a thyroid ultrasound image classification model based on circular convolutional neural network.The convolutional neural network and the circulatory neural network are introduced.Combining the advantages of convolutional neural network and circulatory neural network,the classification model is improved,the circular convolutional neural network classification model is constructed and the classification model is optimized.The softmax algorithm and L2 regularization are introduced to prevent over-fitting.(3)Development of a thyroid nodule classification system based on circulating neural network.This paper introduces the technologies and tools needed for the development of classification systems,the system development feasibility analysis,requirements analysis and system design.The functions of the thyroid nodule classification system and related tests such as system testing are described.Based on the above research,the thyroid ultrasound image acquired by the cooperative hospital was used as the data set,and the classification and training of the thyroid nodule classification system was realized by the classification and training of the circulatory convolutional neural network classification model.The experimental results show that the classification system has high classification accuracy.