Convolutional Neural Network for Cancer Treatment Response Prediction

Recently, efforts are exerted on cancer treatment prediction based on the biomarkers related to the tumor. Many biomarkers are used to discriminate cancer but there is no accurate evidence which is more accurate. To save resources and due to the enormous computational power, the analysis of huge sequences such as DNA (DesoxyriboNucleic Acid) becomes a valid process. Machine learning and deep learning algorithms have been used to predict drug response. The indicators show that the performance of deep learning models is better than the performance of machine learning based one. In this paper, we introduce the use of Convolutional Neural Network (CNN) to predict the response of 265 different drugs using the gene expression profile and mutation profile of 19702 genes. DeepInsight algorithm used to convert the input data to images to be more suitable as input to the CNN. We proposed eight CNNs with different architectures. At first, the whole data input passes through the DeepInsight model which converts the gene expression data and mutation data to images. Dimension reduction is then applied using the helper technique inside the DeepInsignt. The CNN are trained with different settings such as image size, no of nodes in layers…etc. Comparative analysis shows that the proposed architectures improve the prediction accuracy in a range between 16% and 24% as a reduction in mean squared error (MSE).