基于深度学习的肝包虫病超声影像诊断方法研究  

Hepatic echinococcosis ultrasonic imaging diagnosis based on deep learning

作　　者：马国祥 严传波[1] 杨凌菲 张志豪 MA Guoxiang;YAN Chuanbo;YANG Lingfei;ZHANG Zhihao(College of Medical Engineering Technology,Xinjiang Medical University,Urumqi 830000,China;Ultrasonography Department,The First Affiliated Hospital of Xinjiang Medical University,Urumqi 830011,China)

机构地区：[1]新疆医科大学医学工程技术学院,新疆乌鲁木齐830000 [2]新疆医科大学第一附属医院超声科,新疆乌鲁木齐830011

出　　处：《现代电子技术》2023年第8期66-72,共7页Modern Electronics Technique

基　　金：国家自然科学基金项目(81760315)。

摘　　要：肝包虫病是一种严重的区域性寄生虫病,病变的诊断和分类主要依靠临床医生对超声图像的主观判断,但在医疗条件薄弱的地区,这类病变的筛查和诊断易出现误判、错判。为提高肝包虫病的诊断效率和诊断精度,分析肝包虫病超声图像的影像学特点,文中提出一种智能化辅助诊断方案,将深度学习算法应用于肝包虫病的超声影像诊断任务中。首先,构建新疆地区肝包虫病超声影像数据集,并使用基于ROI的预处理方法使数据集规范化;然后在模型构建方面,与常见的三种深度学习基准模型的肝包虫超声影像分类效果进行对比。实验结果表明,文中诊断方法能够取得较好的分类效果,模型准确率最高达到0.85。此外,根据实验结果,分析不同网络结构下各类型肝包虫病的影像学特点,为临床诊断提供参考依据。Hepatic echinococcosis is a serious regional parasitic disease.The diagnosis and classification of the lesions mainly rely on the subjective judgment of clinicians on ultrasound images.In areas with weak medical conditions,the screening and diagnosis of diseases are prone to misjudgment and incorrect judgment.In order to improve the diagnostic efficiency and accuracy of hepatic echinococcosis and analyze the imaging characteristics of hepatic echinococcosis ultrasound images,an intelligent auxiliary diagnosis scheme is proposed,and the deep learning algorithm is applied to the ultrasonic imaging diagnosis task of hepatic echinococcosis.The ultrasound imaging dataset of hepatic echinococcosis in Xinjiang is constructed,and the ROI⁃based preprocessing method is used to normalize the dataset.In terms of model construction,the effect of hepatic echinococcosis ultrasonic image with this constructed model is compared with those of three common deep learning benchmark models.The experimental results show that the diagnosis method in this paper can achieve good classification results,and the maximum accuracy of the model is 0.85.According to the experimental results,the imaging characteristics of various types of hepatic echinococcosis under different network structures were analyzed to provide reference for clinical diagnosis.

关 键 词：肝包虫病 超声影像 深度学习 辅助诊断 寄生虫病 图像分类 卷积神经网络 

分 类 号：TN919-34[电子电信—通信与信息系统] TP311[电子电信—信息与通信工程]