Chapter 4 - Modular design, image biomarkers, and radiomics

Abstract

This chapter introduces themes related to modular software design, bioimage processing, computer vision, and image-annotation, which we will return to in subsequent chapters. Our goal here is to conduct a fairly broad-ranging survey of image-processing and radiomics technologies in contexts such as cardiac care and immuno-oncology, so as to present an empirical background as a precursor to discussing image-annotations and radiomics in a more theoretical vein. We also consider a few examples of computational simulations applied to disease and prognostic models (such as tumor growth) and outline how cellular or tumor-scale simulations can be double-checked via cross-reference against image and clinical data. In general, we explore the digital logistics of integrating bioimages, simulations, and clinical data, examining software-development practices in fields such as systems biology, which often unify these distinct forms of data into multi-disciplinary models. We conclude by arguing for a specific form of modular software design as a useful paradigm for implementing computer code appropriate for these sorts of interdisciplinary syntheses.