Diagnosis And Analysis System Of Massive B-ultrasound Images Of Thyroid Nodules

Thyroid nodule is a common disease in clinic and has a high incidence in the population.Malignant nodules occur in 5%to 15%of patients.Malignant thyroid nodule is one of the fastest growing malignant tumors in China in the past 30 years.The use of b-ultrasound to detect thyroid nodules is highly recognized by the medical community and patients due to its low price,no trauma,no radioactivity and other advantages.However,due to the poor imaging quality of b-ultrasound images,the diagnostic accuracy of thyroid nodules largely depends on the professional level and clinical experience of doctors,and is greatly affected by subjective factors of doctors,so the misdiagnosis rate is high.In addition,the input of doctors'diagnosis information in the process of diagnosis and treatment is tedious,resulting in a large number of medical documents of b-ultrasound pictures.It is inconvenient to organize and maintain,and difficult to make data statistics.This paper studies the data management and auxiliary diagnosis of thyroid nodule b-ultrasound images,studies the architecture of medical data management system,designs and realizes the diagnosis and analysis system of massive thyroid nodule b-ultrasound images,and studies the calibration and storage management of b-ultrasound images.The auxiliary diagnosis model of thyroid nodule based on machine learning was studied to realize the auxiliary diagnosis of benign and malignant thyroid nodule.A diagnosis and analysis system with functions of information management,file management,picture data annotation and export,and auxiliary diagnosis has been established.The main work and achievements of this paper are as follows:1.Analyzed the functional requirements of the diagnostic analysis system of massive b-mode ultrasound images of thyroid nodules,and designed the diagnostic analysis functional model and data model.The system is divided into basic function module,basic information management module,patient information management module,auxiliary diagnosis module and data statistics module.2.Designed the diagnosis and analysis system architecture of massive b-ultrasound images of thyroid nodules.The key technologies such as permission design,database model design and auxiliary diagnosis analysis are studied.The diagnostic options of the diagnostic page can be customized,and the features can be labeled and stored in detail.The export function of b-ultrasound image and calibration metadata is designed.The exported image and calibration data can be directly used to assist the training of diagnostic model.And integrate the auxiliary diagnostic technique to distinguish the benign and malignant thyroid nodules.3.The machine learn-based thyroid nodule auxiliary diagnosis model was studied,and the deep learning algorithm was used to set up the thyroid nodule auxiliary diagnosis model based on SSD(Single Shot MultiBox Detector).The average accuracy rate of thyroid nodule region recognition reached 87.5%,and the average accuracy rate of benign and malignant diagnosis reached 82.1%,which was close to the average clinical level.It provides a new idea for the auxiliary diagnosis of thyroid nodule.4.Studied the containerized system deployment and system monitoring technology,adopted docker container to deploy the project on Rancher platform,simplified the system deployment process,and strengthened the system monitoring.