From Handcrafted to Deep-Learning-Based Cancer Radiomics: Challenges and Opportunities

Recent advancements in signal processing and machine learning coupled with\ndevelopments of electronic medical record keeping in hospitals and the\navailability of extensive set of medical images through internal/external\ncommunication systems, have resulted in a recent surge of significant interest\nin "Radiomics". Radiomics is an emerging and relatively new research field,\nwhich refers to extracting semi-quantitative and/or quantitative features from\nmedical images with the goal of developing predictive and/or prognostic models,\nand is expected to become a critical component for integration of image-derived\ninformation for personalized treatment in the near future. The conventional\nRadiomics workflow is typically based on extracting pre-designed features (also\nreferred to as hand-crafted or engineered features) from a segmented region of\ninterest. Nevertheless, recent advancements in deep learning have caused trends\ntowards deep learning-based Radiomics (also referred to as discovery\nRadiomics). Considering the advantages of these two approaches, there are also\nhybrid solutions developed to exploit the potentials of multiple data sources.\nConsidering the variety of approaches to Radiomics, further improvements\nrequire a comprehensive and integrated sketch, which is the goal of this\narticle. This manuscript provides a unique interdisciplinary perspective on\nRadiomics by discussing state-of-the-art signal processing solutions in the\ncontext of Radiomics.\n