放射性肝病的神经网络模型预测研究  

Prediction on radiation-induced liver disease with an artificial neural network model

作　　者：朱骥[1] 朱小东[2] 梁世雄[2] 徐志勇[1] 赵建东[1] 傅小龙[1] 蒋国梁[1] 

机构地区：[1]复旦大学附属肿瘤医院放射治疗科,上海200031 [2]广西医科大学附属肿瘤医院放射治疗科,南宁530021

出　　处：《癌症进展》2006年第4期314-318,共5页Oncology Progress

摘　　要：目的评价人工神经网络在放射性肝病预测中的价值。方法93例肝硬化Child-PughA级患者接受三维适形放疗,其中8例发生放射性肝病。93例被随机分为训练集和验证集进行模型拟合。结果ROC曲线下面积为0.8897,灵敏度、特异度、准确度、阳性预测值、阴性预测值分别为0.875,0.882,0.882,0.412和0.987。结论预测因子涵盖物理和临床指标,人工神经网络模型在放射性肝病的预测中获得了较高的准确度。Objective To evaluate the efficiency of predicting radiation - induced liver disease （R1LD） with an artificial neural network （ANN） model. Methods From August 2000 to November 2004, a total of 93 primary liver carcinoma （PLC） patients with single lesion and associated with hepatic cirrhosis of Child - Pugh grade A, were treated with hypofractionated 3 - dimensional confonnal radiotherapy （3DCRT）. Eight out of 93 patients were diagnosed RILD. Ninety - three patients were randomly divided into two subsets （training set and verification set）. Results The area under receiver - operating characteristic （ROC） curves was 0. 8897. Sensitivity, specificity, accuracy, positive prediction value （PPV）, and negative prediction value （NPV） were 0.875, 0.882, 0.882, 0.412 and 0.987 respectively. Conclusion ANN was proved high accuracy for prediction of RILD. It could be used together with other models and dosimetric parameters to evaluate hepatic irradiation plans.

关 键 词：放射性肝病 神经网络 预测 

分 类 号：R735.7[医药卫生—肿瘤] R730.55[医药卫生—临床医学]