Computer Engineering and Applications ›› 2018, Vol. 54 ›› Issue (14): 138-142.DOI: 10.3778/j.issn.1002-8331.1702-0110

Previous Articles     Next Articles

Research on approach for robust lung cancer survival prediction based on deep learning

PAN Hao1, WANG Zhao2, YAO Jiawen3   

  1. 1.College of Economics and Management, Beijing Institute of Petro-Chemical Technology, Beijing 102600, China
    2.College of Economics and Management, Beijing University of Chemical Technology, Beijing 100029, China
    3.Department of Computer Science and Engineering, University of Texas at Arlington, Arlington, TX 76019, USA
  • Online:2018-07-15 Published:2018-08-06


潘  浩1,王  昭2,姚佳文3   

  1. 1.北京石油化工学院 经济管理学院,北京 102600
    2.北京化工大学 经济管理学院,北京 100029
    3.美国德州大学阿灵顿分校 计算机科学与工程学院,美国 阿灵顿 76019

Abstract: Lung cancer is one of serious diseases causing death for humans. Improving survival prediction performance is meaningful for making the treatment plans and improving the survival rates of lung cancer patients. In this paper, the survival prediction framework is based on histopathology images. In the proposed framework, it firstly detects the lung cancer cells automatically using deep learning method and then extracts features from the detected cells. The topological features are employed to describe the distribution of the cells and the topological features are used as the prediction factors for the survival prediction. Finally, the survival prediction is done by applying cox proportional hazards model with Lasso method. Experimental results show that the proposed method can improve both the efficiency and accuracy of cells detection and the power of lung cancer survival prediction model.

Key words: deep learning, topological features, survival prediction

摘要: 肺癌是一种严重威胁患者生命的恶性肿瘤。通过对肺癌病人进行生存预测分析并制定针对性治疗方案,对提高病人生存率具有重要意义。提出一种基于病理学图像的肺癌患者生存预测分析方法。首先采用深度学习方法对病理学图片进行肺癌细胞自动检测,并对检测出的肺癌细胞进行特征提取。在特征选取中,引入了反映肺癌细胞间关系和分布特性的拓扑特征的提取方法,将提取的拓扑特征作为生存分析的预测因素。最后采用Cox-Lasso方法对肺癌患者进行生存预测分析。实验结果表明,该方法能够提高细胞检测的效率和准确性,并具有较高的肺癌患者生存预测分析能力。

关键词: 深度学习, 拓扑特征, 生存预测