人工神经网络在超声诊断附件包块良恶性中的应用  

Prediction of malignancy in adnexal masses using an artificial neural network

作　　者：王黎明[1] 付庆诏[1] 冯天瑾[2] 刘洪波[2] 车艳辞[1] 周晓彬[3] 

机构地区：[1]山东大学齐鲁医院超声科 [2]中国海洋大学信息学院 [3]青岛大学医学院卫生学教研室

出　　处：《中华超声影像学杂志》2004年第8期597-600,共4页Chinese Journal of Ultrasonography

摘　　要：目的　研究超声技术中应用人工神经网络对附件包块良恶性的诊断性能。方法　将 180例附件包块患者随机分为训练组和测试组 ,利用训练组筛选出对诊断良恶性有意义的单项参数指标 ,并建立统计学方法和人工神经网络方法各自的诊断模型 ,比较两个模型在测试组的诊断性能。结果　 17项诊断参数中单因素分析有 13项对判断附件包块的良恶性有意义 (P <0 .0 5 )。由统计学多因素分析选出的参数 (年龄、包块内壁突起、腹水及血流平均速度 )作为输入层参数的三层BP神经网络模型较统计学模型诊断性能高 (P =0 .0 2 71)。结论　超声技术中应用人工神经网络对附件包块良恶性有良好诊断性能 。Objective To evaluate the performance of artificial neural network(ANN) models for predicting ovarian malignancy in the patients with adnexal masses by using B-mode and color Doppler ultrasonography.Methods The data of 180 patients with adnexal masses were randomly divided into training and testing subsets.The training subsets were used to screen out significant single parameters and to compute the optimum statistic equations and to train the ANN.The testing subsets were used to estimate the performance of each of the two models in predicting ovarian malignancy.Results Thirteen of seventeen parameters were significant to diagnose malignancy in adnexal masses by single analysis(P< 0.05).Three-layer back-propagation network,based on the same input variables selected by using multivariate analysis(women′s ages,papillary projection,ascites and time-averaged mean velocity) had a significantly higher diagnostic rate than statistic model did(P= 0.0271).Conclusions Artificial neural network used in sonographic prediction of malignancy in adnexal masses has better diagnostic capability.There is a need for further investigation.

关 键 词：人工神经网络 超声检查 诊断 良性附件包块 恶性附件包块 

分 类 号：R445.1[医药卫生—影像医学与核医学]