Extended Data Fig. 4: Confusion matrices for each radiologist and experiment.
From: Image prediction of disease progression for osteoarthritis by style-based manifold extrapolation
Seven radiologists were tasked to judge whether a given knee radiograph is likely to undergo OA progression in the future (prog) or not (non-prog), based on the baseline radiograph alone (a) or the baseline radiograph and the synthesized predicted radiograph (b). For both experiments, we randomly selected 486 knee radiographs from the OAI test set.
