肝癌CT影像人工神经网络辅助诊断模型的评价

目的：评价人工神经网络辅助诊断模型应用于肝癌CT影像诊断的真实性和可靠性.方法：以临床各期肝癌和其它肝占位疾病的CT影像特征数据为训练样本集,建立肝癌CT影像人工神经网络辅助诊断模型,并确定该神经网络应用于肝癌CT影像诊断的截断点;同时,将另外收集的肝癌和其它肝占位疾病CT影像特征数据通过神经网络进行仿真运算,得到网络诊断输出,计算该神经网络模型应用于肝癌CT影像诊断的真实性和可靠性.结果：神经网络用于肝癌CT影像诊断的灵敏度和特异度分别为：98.0%和96.0%,诊断准确度为97.0%;进一步计算该神经网络模型诊断可靠性,其诊断符合率为91.0%,卡帕值为0.82（Z=4.8,P＜0.05）.结论：人工神经网络模型可以应用于肝癌CT影像的临床诊断.

Evaluation of assistant diagnosis model based on artificial neural network in hepatocellular carcinoma CT imaging

Objective: To evaluate the validity and reliability of assistant diagnosis model based on artificial neural network(ANN) in hepatocellular carcinoma(HCC) CT imaging.Methods: The CT imaging of HCC(n=100) and other liver occupational diseases(n=100) were collected and their features were identified for the training database.The ANN for diagnosis of HCC CT imaging was constructed with a back-propagation training algorithm. Based on the emulational operation by ANN,a network diagnosis was made.The validity and reliability of ANN for HCC diagnosis were calculated.Results: The sensitivity,specificity and accuracy of ANN for diagnosis of HCC CT imaging were 98%,96% and 97%,respectively,the diagnostic consilient rate was 91% and ksppa value was 0.82(Z=4.8.P<0.01).Conclusion: It is suggested that the ANN may be used for clinical diagnosis of HCC.Abstract:Objective:To evaluate the validity and reliability of assistant diagnosis model based on artificial neural network(ANN) in hepatocellular carcinoma(HCC) CT imaging.Methods:The CT imaging of HCC(n=100) and other liver occupational diseases(n=100) were collected and their features were identified for the training database.The ANN for diagnosis of HCC CT imaging was constructed with a back-propagation training algorithm.Based on the emulational operation by ANN,a network diagnosis was made.The validity and reliability of ANN for HCC diagnosis were calculated.Results:The sensitivity,specificity and accuracy of ANN for diagnosis of HCC CT imaging were 98%,96% and 97%,respectively,the diagnostic consilient rate was 91% and ksppa value was 0.82(Z=4.8.P <0.01).Conclusion:It is suggested that the ANN may be used for clinical diagnosis of HCC.