I have some concerns about model induced false signals but clearly the professionals who have to do this really do value alternate takes on the featureset.
A side issue here is the natural feedback loop: clinicians will tell radiology services too many false positive referrals are coming through (my partner already experienced this push back on finer grained diagnostic for a venous issue the consultant said was un-imagable 5 years ago and was 50/50 for risk/reward to work on. "Why are you here: on this diagnosis you shouldn't be here" response)
My point is that advances in imaging probably always has leaps and bounds and reconsideration.
A side issue here is the natural feedback loop: clinicians will tell radiology services too many false positive referrals are coming through (my partner already experienced this push back on finer grained diagnostic for a venous issue the consultant said was un-imagable 5 years ago and was 50/50 for risk/reward to work on. "Why are you here: on this diagnosis you shouldn't be here" response)
My point is that advances in imaging probably always has leaps and bounds and reconsideration.