Lung cancer is currently the leading cause of cancer-related mortality worldwide, and the dominant histopathology is non-small cell lung cancer (NSCLC). Although many new targeted and immunomodulation therapies have emerged, not all patients are responsive to novel therapeutics. A more reliable and accurate risk stratification model to predict the treatment response and survival outcomes are still lacking. The 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) derived radiomics can be used to interrogate tumor biologies such as glycolytic activity and heterogeneity. It can, therefore, be used to predict treatment response and survival outcomes. Cancer genomics derived from gene sequencing can evaluate cancer's genetic alterations. It can be used to feature the genotype of the tumor. However, both tools have drawbacks; combining these two modalities may enable a more robust predictive model for more precise clinical decisions. During the investigator's former study project, the investigators published four Science Citation Index journal papers using the investigators' research results, which found that 18F-FDG PET radiomics can independently predict regional lymph node metastasis in NSCLC and cancer survival by stage. The preliminary findings of the investigator's former research project also disclosed an association between 18F-FDG PET-derived molecular radiomics with genomic heterogeneity and mutation of specific glucose metabolic genes. This time, the investigators plan to include deep radiomics in addition to traditional handcrafted radiomics. The investigators aim to investigate the radiogenomic patterns in different driver gene mutation statuses and clinical scenarios. Finally, the investigators seek to use radiogenomics as a prognostic stratification tool in patients with NSCLC.