Uncertainties associated to the extraction of texture features in single-energy contrast-enhanced mammography

Added value of feature uncertainty in a radiomic analysis of contrast-enhanced digital mammography boosted by deep learning

“…Different approaches have been taken to consider RF variability, one of them being the assessment of uncertainty over the RF and the subsequent CADx model. Mendel et al 5 conducted a study to quantify feature robustness across different mammography unit manufacturers, and Castillo-Lopez et al 6 investigated the potential of uncertainty analysis to improve the evaluation of texture and the appraisal of prediction models from CEDM single-energy images. Yet, possible solutions to consider uncertainty effects have not been fully explored.…”