摘要
胰腺癌是一种致死率比较高的恶性肿瘤,早发现、早治疗可以降低死亡率。CT图像能够准确的定位病灶的位置,是医生检测胰腺癌常用的医疗设备,但大量的阅片工作量容易使医生产生视觉疲劳,造成误诊和漏诊等现象。本文提出通过计算机自动检测胰腺癌的方案,该方案可以辅助医生检测胰腺癌,提高检测率。具体研究内容包括:胰腺区域数据集的收集、图像预处理、感兴趣区域的提取、卷积神经网络胰腺癌分类模型的构建、卷积神经网络模型的训练及测试。并实验验证利用本方案检测胰腺癌的准确率达到了96.8%,高于已有的两种方法,能够实现更高的检测率,可以为医生提供有价值的参考意见。
Pancreatic cancer is a malignant tumor with high mortality.Early detection and treatment can reduce the mortality.CT images can accurately locate the location of lesions,which is a common medical device for doctors to detect pancreatic cancer.However,a large amount of workload of reading films tends to cause visual fatigue among doctors,resulting in misdiagnosis and missed diagnosis.This paper proposes a program of automatic detection of pancreatic cancer by computer,which can help doctors detect pancreatic cancer and improve the detection rate.Research contents include:data collection of pancreatic region,image pre-processing,extraction of interested region,construction of classification model of pancreatic cancer in convolutional neural network,training and testing of convolutional neural network model.In addition,the experiment verified that the accuracy rate of detecting pancreatic cancer by this scheme reached 96.8%,which was higher than the two existing methods and could achieve a higher detection rate and provide valuable reference for doctors.
作者
唐思源
杨敏
刘燕茹
TANG Si-yuan;YANG Min;LIU Yan-ru(Baotou medical college,Inner Mongolia university of science and technology Department of computer science and technology;Baotou medical college,Inner Mongolia university of science and technology Department of medical imaging technology,Baotou,Inner Mongolia,014060)
出处
《软件》
2020年第5期87-90,共4页
Software
基金
包头市医药卫生科技计划项目(项目编号:Wsjj2019040)
2019年包头医学院大学生创新创业训练计划项目(项目编号:BYDCXL-201922)
2019年包头医学院花蕾计划(大学生创新训练)项目(项目编号:2019BYJJ-HL-11)。
关键词
胰腺癌
CT图像
感兴趣区域
卷积神经网络模型
Pancreatic cancer
CT images
Areas of interest
Convolutional neural network model
作者简介
唐思源(1981-),女,内蒙古包头市人,硕士学位,现为内蒙古科技大学包头医学院计算机科学与技术系教研室主任,主要研究方向:医学图像处理;刘燕茹(1982-),女,内蒙古包头市人,硕士学位,主要研究方向:医学图像处理;通讯作者:杨敏(1984-),女,内蒙古包头市人,硕士学位,主要研究方向:医学图像处理。
