The article discusses "Radiomics," a field that involves extracting and analyzing large amounts of quantitative imaging features from medical images obtained with CT, PET, or MRI. These features are designed to be extracted from standard-of-care images, allowing for large subject pools and the potential to build descriptive and predictive models relating image features to phenotypes or gene-protein signatures. The core hypothesis is that these models can provide diagnostic, prognostic, or predictive information. The radiomics process includes five key steps: image acquisition and reconstruction, image segmentation and rendering, feature extraction and qualification, databases and data sharing, and ad hoc informatic analyses. Each step poses unique challenges, such as identifying optimal protocols for image acquisition, ensuring robust and minimal operator input for segmentation, generating features that reflect tumor complexity without being overly complex, and developing informatics databases that incorporate image features and medical data. The article also discusses the challenges in image acquisition and reconstruction, segmentation, feature extraction, and the need for large image datasets. It highlights the importance of de-identification of data for privacy, the development of an integrated radiomics database, and the statistical and informatics analysis of radiomic data. The article concludes with the importance of addressing sample size issues and incorporating clinical and risk factor data into radiomics to improve the accuracy and relevance of predictive models.The article discusses "Radiomics," a field that involves extracting and analyzing large amounts of quantitative imaging features from medical images obtained with CT, PET, or MRI. These features are designed to be extracted from standard-of-care images, allowing for large subject pools and the potential to build descriptive and predictive models relating image features to phenotypes or gene-protein signatures. The core hypothesis is that these models can provide diagnostic, prognostic, or predictive information. The radiomics process includes five key steps: image acquisition and reconstruction, image segmentation and rendering, feature extraction and qualification, databases and data sharing, and ad hoc informatic analyses. Each step poses unique challenges, such as identifying optimal protocols for image acquisition, ensuring robust and minimal operator input for segmentation, generating features that reflect tumor complexity without being overly complex, and developing informatics databases that incorporate image features and medical data. The article also discusses the challenges in image acquisition and reconstruction, segmentation, feature extraction, and the need for large image datasets. It highlights the importance of de-identification of data for privacy, the development of an integrated radiomics database, and the statistical and informatics analysis of radiomic data. The article concludes with the importance of addressing sample size issues and incorporating clinical and risk factor data into radiomics to improve the accuracy and relevance of predictive models.