DATA SECURITY IN GENE MUTATION ANALYSIS: VISUALIZATION AND GAN FOR CANCER CLASSIFICATION

Abstract

In this paper, we investigate and propose a novel deep learning-based approach for detecting gene mutations associated with several common cancer types, while ensuring information security during the analysis process. The method begins by visualizing gene mutation data as grayscale images, a crucial step in safeguarding patients sensitive information. Following this, deep learning models are employed to more effectively extract latent features from gene data, while ensuring that personal data remains protected. After visualization, Generative Adversarial Networks are applied to enhance data diversity, generating new image samples from the original gene data without compromising information security. This process not only highlights key features of gene mutations but also improves the generalization apabilities of model while maintaining patient privacy. The critical features learned by the Discriminator  are used as input for a Convolutional Neural Network to classify 12 common cancer types. Experimental results demonstrate that the proposed method achieves superior performance in detecting and classifying cancer gene mutations, while ensuring that personal genetic data is safeguarded. This research not only introduces a novel deep learning approach for gene mutation analysis but also ensures information security, effectively supporting cancer diagnosis and treatment.