摘要
脑动脉瘤是一种严重的脑血管疾病,其破裂后致死率、致残率高。早期脑动脉瘤的及时检出对于临床诊疗具有重要意义。计算机断层血管造影(CTA)已成为脑动脉瘤诊断的关键技术,然而传统的人工读片方式不仅效率低下,而且容易因为人为因素导致误诊和漏诊的情况。为提高脑动脉瘤的诊断效率和精度,将ResNet网络应用于脑动脉瘤自动检测任务中。分别采用ResNet34、ResNet50和ResNet101三种不同深度的ResNet网络在自建头颅CTA图像数据集上进行实验,并与MobileNet、AlexNet及VGG16三种经典深度学习方法进行对比分析。实验结果表明,基于ResNet的脑动脉瘤自动检测模型能够表现出优于其他深度学习方法的检测性能。
Cerebral aneurysm is a serious cerebrovascular disease with high mortality and disability rates after rupture.The timely detection of early cerebral aneurysms is of great significance for clinical diagnosis and treatment.Computed tomography angiography(CTA)has become a key technology for the diagnosis of cerebral aneurysms.However,traditional manual interpretation methods are not only inefficient,but also prone to misdiagnosis and missed diagnosis due to human factors.To improve the diagnostic efficiency and accuracy of cerebral aneurysms,ResNet network is applied to the automatic detection task of cerebral aneurysms.Three different depths of ResNet networks,ResNet 34,ResNet 50,and ResNet 101,were used to conduct experiments on a self built head CTA image dataset,and compared and analyzed with three classic deep learning methods,MobileNet,AlexNet,and VGG16.The experimental results show that the ResNet based automatic detection model for cerebral aneurysms can exhibit better detection performance than other deep learning methods.
作者
张雪原
叶明全
陈璇
王家琦
吴爱萍
殷鹏展
ZHANG Xueyuan;YE Mingquan;CHEN Xuan;WANG Jiaqi;WU Aiping;YIN Pengzhan(School of Medical Information,Wannan Medical College,Wuhu 241002,China;Institute of Artificial Intelligence,Hefei Comprehensive National Science Center,Hefei 230088,China;Radiology Department,The First Affiliated Hospital of Wannan Medical College,Wuhu 241001,China)
出处
《宿州学院学报》
2024年第12期24-28,共5页
Journal of Suzhou University
基金
安徽省重点研究与开发计划项目(2022a05020011)
安徽省高校协同创新项目(GXXT-2022-044)
安徽省高校优秀科研创新团队项目(2022AH010075)
安徽省高校学科(专业)拔尖人才学术资助项目(gxbjZD2022042)。
关键词
深度学习
卷积神经网络
脑动脉瘤
图像检测
Deep learning
Convolutional neural networks
Cerebral aneurysm
Image detection
作者简介
张雪原(1997—),女,安徽广德人,在读硕士,研究方向:医学图像处理与分析。;通信作者:叶明全(1973—),安徽当涂人,博士,教授,研究方向:数据挖掘与机器学习、医疗大数据、医学图像处理与分析。
