检索规则说明:AND代表“并且”;OR代表“或者”;NOT代表“不包含”;(注意必须大写,运算符两边需空一格)
检 索 范 例 :范例一: (K=图书馆学 OR K=情报学) AND A=范并思 范例二:J=计算机应用与软件 AND (U=C++ OR U=Basic) NOT M=Visual
作 者:罗锦文[1] 朱光斌 关玉宝[1] 周斯琴 何月明[2] LUO Jinwen;ZHU Guangbin;GUAN Yubao;ZHOU Siqin;HE Yueming(Department of Medical Imaging,The Fifth Affiliated Hospital,Guangzhou Medical University,Guangzhou 510700,China;Department of Gynecology,The Fifth Affiliated Hospital,Guangzhou Medical University,Guangzhou 510700,China)
机构地区:[1]广州医科大学附属第五医院影像科,广州510700 [2]广州医科大学附属第五医院妇科,广州510700
出 处:《中国医学计算机成像杂志》2022年第4期390-396,共7页Chinese Computed Medical Imaging
基 金:广东高校生物靶向诊治与康复重点实验室(2021KSY009)。
摘 要:目的:探讨基于宫颈癌MR图像纹理特征,建立用于FIGOⅡ期宫颈癌诊断宫旁浸润的RBF神经网络分类预测模型。方法:纳入本院90例经手术或活检病理证实的FIGOⅡ期宫颈癌MR影像资料,其中FIGOⅡB期宫旁浸润组45例,FIGOⅡA期非宫旁浸润组45例。用Mazda软件提取宫颈癌MR图像的794种纹理参数,经过特征选择降维得到10种纹理参数特征,对其中具有统计学差异参数作为自变量,采用SPSS软件进行RBF神经网络预测模型的建立,并构建ROC曲线分析RBF预测模型的诊断效能。结果:成功建立了能够判断宫旁浸润与非宫旁浸润的RBF神经网络预测模型,经过10次反复随机法训练后验证后建立的RBF神经网络分类预测最佳模型的培训整体正确率为84.5%,训练整体正确率为84.4%,ROC模型曲线下面积分别为0.877。结论:基于MR图像纹理特征RBF神经网络预测模型对FIGOⅡ期宫颈癌能提高MR亚分期的准确性,有助于临床医师的临床决策。Purpose:To explore the image texture features based on cervical cancer MR,and to build RBF neural network prediction model for diagnosis of FIGOⅡstage cervical cancer with parauterine infiltration.Methods:In our hospital,90 cases of FIGOⅡstage cervical cancer confirmed by surgery or biopsy pathology results were enrolled,including FIGOⅡB stage with parauterine infiltration group 45 cases,FIGOⅡA stage with no parauterine infiltration group 45 cases.MaZda software was used to extract 794 texture parameters of cervical cancer MR image,and 10 texture parameters were obtained after feature selection and dimension-reduction,among which the parameters with statistical differences were taken as independent variables.SPSS software was used to establish the RBF neural network prediction model,and the ROC curve was constructed to analyze the diagnostic efficacy of the RBF prediction model.Results:The prediction model of RBF neural network was successfully established to judge the parauterine infiltration and non-parauterine infiltration.The overall accuracy of RBF neural network classification prediction model was 84.5%,and 84.4%after 10 times of random training,and the area under the ROC model curve was 0.877.Conclusion:RBF neural network prediction model based on MR image texture feature of FIGOⅡstage cervical cancer can improve the accuracy of MR staging,and help clinicians making clinical decision.
关 键 词:磁共振成像 纹理分析 径向基函数 神经网络 临床分期
分 类 号:R445.2[医药卫生—影像医学与核医学]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在链接到云南高校图书馆文献保障联盟下载...
云南高校图书馆联盟文献共享服务平台 版权所有©
您的IP:216.73.216.200