Radstract should mention in its docs early exactly what to use it for, with examples. e.g. "processing DICOMs and NIFTIS for machine learning"
We should also mention the reasons for creating it, and what its goals are (moving medical imaging to be more like other CS Tasks)