Imaging Biomarkers and Gene Expression Data Correlation Framework for Lung Cancer Radiogenomics Analysis Based on Deep Learning

Dong Sui,Maozu Guo,Xiaoxuan Ma,Julian Baptiste,Lei Zhang

DOI: https://doi.org/10.1109/access.2021.3071466

IF: 3.9

2021-01-01

IEEE Access

Abstract:Precision medicine, a popular treatment strategy, has become increasingly important to the development of targeted therapy. To correlate medical imaging with prognostic and genomic data, researches in radiomics and radiogenomics have provided many pre-defined image features to describe image information quantitatively or qualitatively. However, in previous researches, there are only statistical results which prove high correlation among multi-source medical data, but those can't give intuitive and visual result. In this paper, a deep learning based radiogenomic framework is provided to construct the linkage from lung tumor images to genomic data and implement generation process in turn, which form a bi-direction framework to map multi-source medical data. The imaging features are extracted from autoencoder under the condition of genomic data. It can obtain much more relevant features than traditional radiogenomic methods. Finally, we use generative adversarial network to transform genomic data onto tumor images, which gives a cogent result to explain the linkage between them. As a result, our framework provides a deep learning method to do radiogenomic researches more functionally and intuitively.

computer science, information systems,telecommunications,engineering, electrical & electronic

What problem does this paper attempt to address?

The main problem this paper attempts to address is to correlate lung tumor imaging features with gene expression data by constructing a deep learning-based radiogenomics framework and achieving bidirectional mapping. Specifically, the study aims to:

1. **Extract Imaging Features**: Extract imaging features from lung tumor CT images that are highly correlated with gene expression data. These features are richer and more effective than those extracted by traditional methods.
2. **Generate Tumor Images**: Use Generative Adversarial Networks (GAN) to convert gene expression data into tumor images, thereby providing intuitive visual results that explain the relationship between gene changes and tumor imaging.
3. **Improve Diagnosis and Prognosis Prediction**: Improve the accuracy and functionality of lung cancer diagnosis and prognosis prediction through deep learning methods.

The background section of the paper points out that although existing radiomics and radiogenomics studies have demonstrated high correlations between multi-source medical data, these studies usually only provide statistical results and lack intuitive and visual explanations. Therefore, this study proposes a new deep learning framework to overcome these limitations and provide a more intuitive and functional radiogenomics research method.