The article "Radiomics: The Process and the Challenges" by Virendra Kumar et al. discusses the extraction and analysis of advanced quantitative imaging features from medical images, primarily CT, PET, and MRI. The core hypothesis is that these features can provide valuable diagnostic, prognostic, and predictive information. The process of radiomics is divided into five main steps: image acquisition and reconstruction, image segmentation and rendering, feature extraction and qualification, database and data sharing, and ad hoc informatic analyses. Each step poses unique challenges, such as optimizing image acquisition protocols, ensuring robust and minimal operator input in segmentations, generating features that reflect the complexity of individual volumes, and developing robust statistical approaches to analyze the data. The article focuses on non-small cell lung cancer (NSCLC) and highlights the importance of large image datasets for radiomic analysis. It also addresses the need for standardized protocols and the challenges in segmenting lung tumors, particularly in cases with high-intensity tumors attached to the pleural wall or mediastinum. The article further discusses feature extraction, including tumor intensity histograms, shape-based features, and texture-based features, and the importance of feature qualification to ensure reproducibility and non-redundancy. Finally, it emphasizes the role of integrated databases and data sharing in linking image features to clinical and molecular data, and the importance of statistical and bioinformatics approaches in validating radiomic findings.The article "Radiomics: The Process and the Challenges" by Virendra Kumar et al. discusses the extraction and analysis of advanced quantitative imaging features from medical images, primarily CT, PET, and MRI. The core hypothesis is that these features can provide valuable diagnostic, prognostic, and predictive information. The process of radiomics is divided into five main steps: image acquisition and reconstruction, image segmentation and rendering, feature extraction and qualification, database and data sharing, and ad hoc informatic analyses. Each step poses unique challenges, such as optimizing image acquisition protocols, ensuring robust and minimal operator input in segmentations, generating features that reflect the complexity of individual volumes, and developing robust statistical approaches to analyze the data. The article focuses on non-small cell lung cancer (NSCLC) and highlights the importance of large image datasets for radiomic analysis. It also addresses the need for standardized protocols and the challenges in segmenting lung tumors, particularly in cases with high-intensity tumors attached to the pleural wall or mediastinum. The article further discusses feature extraction, including tumor intensity histograms, shape-based features, and texture-based features, and the importance of feature qualification to ensure reproducibility and non-redundancy. Finally, it emphasizes the role of integrated databases and data sharing in linking image features to clinical and molecular data, and the importance of statistical and bioinformatics approaches in validating radiomic findings.