Lung Cancer Classification System for CT Images using Deep Convolutional Neural Network

Abstract

AbstractLung cancer is one of the fatal cancer types, and its identification at early stages might increase the patients’ survival rate up to 60–70%. Predicting patients’ lung cancer survivability has become a trendy research topic by scholars from medicine and computer science domains. In this article, a novel approach to the deep convolutional neural network (DCNN) is proposed for the precise and automatic classification of lung cancer. Specifically, the auto weight dilated convolutional unit utilized multi-scale convolutional feature maps to acquire lung cancer features at different scales and employed a learnable set of parameters to fuse convolutional feature maps in encoding layers. The AD unit is an effective architecture for feature extraction in the encoding stage. We used the advantages of the U-Net network for deep and shallow features, combined with AD units to multimodal image classification. In this model, the four channel model inputs correspond to the CT images of four modes, respectively. The main body of the network is composed of auto-weight dilated (AD) unit, Residual (Res) unit, linear upsampling, and the first and last convolution units.. The network that applied Block-R3 had higher segmentation performance than the networks of Block-R1 and Block-R2 experimental results indicated that DCN had outperformed all the six classical machine learning algorithms in predicting the survival period of lung cancer patients with an accuracy of 88.58%. The results are believed to support healthcare professionals to manage costs and provide treatment at the appropriate time. KeywordsLung cancerCTSegmentationDeep CNNClassification