Alex Zwanenburg and Steffen Löck
Medical imaging provides non-invasive anatomical and functional visualisation of the human body. It is used clinically for diagnostic, prognostic, treatment planning and other purposes. Many current uses of medical imaging involve qualitative or semi-quantitive assessment by experts. Radiomics seeks to automate analysis of medical imaging for clinical decision support. At its core, radiomics involves the extraction and machine learning-based analysis of quantitive features from medical images (Lambin et al., 2017). However, very few–if any–radiomics tools have been translated to the clinic (Huang et al., 2022). One of the essential prerequisites for translation is reproducibility and validation in external settings (O’Connor et al., 2017). This can be facilitated through the use of standardised radiomics software. Here we present mirp, a Python package for standardised processing of medical imaging and computation of quantitative features. Researchers can use mirp for their own radiomics analyses or to reproduce and validate radiomics analyses of others.