基于遗传神经网络和舌根癌Radiomics特征的生存期预测  

Survival Prediction Based on Genetic Neural Network and Radiological Characteristics of Tongue Root Cancer

作　　者：潘晓英[1] 杨清萍 PAN Xiaoying;YANG Qingping(Xi'an University of Posts and Telecommunications,Xi'an 710061)

机构地区：[1]西安邮电大学

出　　处：《计算机与数字工程》2019年第10期2509-2512,2545,共5页Computer & Digital Engineering

基　　金：西安邮电大学创新基金项目(编号：114-602080144)资助

摘　　要：在全球范围内癌症严重危害人类健康,正确评估癌症患者的预后状况对于个性化治疗意义重大。影像组学(Radiomics)数据中存在着大量反映肿瘤形状、强度、纹理和小波特性的定量特征,这些特征与其预后效果间存在着不可忽视的关联关系。论文针对舌根癌的Radiomics特征数据,提出了一种基于遗传神经网络(GA-BP神经网络)的生存期预测模型。仿真实验结果表明论文所提模型具有较强的预测能力,与BP神经网络预测模型的比较结果表明,该模型具有更强的预测精度。In the global context,cancer seriously endangers human health and correctly assess the prognosis of cancer patients is of great significance for personalized treatment.There are a large number of quantitative features in Radiomics data that reflect the shape,intensity,texture and wavelet characteristics of tumors,and these characteristics are not negligible associated with their prognosis.In this paper,based on Radiomics characteristic data of tongue root cancer,a survival prediction model based on genetic neural network(GA-BP neural network)is proposed.Simulation results show that the model proposed in this paper has strong predictive ability.Compared with the BP neural network prediction model,the results show that the proposed model has stronger prediction accuracy.

关 键 词：预后 影像组学 BP神经网络 遗传神经网络 生存期 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] R445[自动化与计算机技术—控制科学与工程]