Paving a Path to Clinical Impact with Radiomics: Enabling Reproducibility and Reach

Abstract Radiomics, the extraction of quantitative data from images, holds promise for noninvasively characterizing tumor phenotypes. Tools like LIFEx have improved the accessibility, transparency, and reproducibility of radiomic feature extraction by offering standardized, user-friendly workflows across imaging modalities. Introduction of such a platform that enables consistent and transparent analytics has helped democratize access to the exploration of radiomics and has highlighted other fundamental challenges in radiomics, addressing upstream heterogeneity in image acquisition, reconstruction, and region-of-interest segmentation that impede reproducibility. Differences in these upstream steps can drastically alter radiomic features, even when downstream processing is standardized. We highlight ongoing efforts and fundamental challenges that the community will need to tackle collectively to enable the clinical translation of radiomics. By addressing variability throughout the radiomic pipeline, we can ensure that radiomic features better reflect tumor biology, as well as fulfill their potential as robust, generalizable biomarkers for precision oncology. This article is part of a special series: Driving Cancer Discoveries with Computational Research, Data Science, and Machine Learning/AI. See related article by Nioche and colleagues, Cancer Res 2018;78:4786-89