Who should read the studies?

I chatted with a neurologist (a stroke specialist) at a gathering yesterday about the MRI techniques he is using in his research. He never depends on radiologist reads, he says, and interprets everything himself. This is the way things are going, he said. I don't know about the general trend – obviously, there are simple studies I read myself, but many things I leave to radiologists. But I'm not sure where that leaves us in terms of using imaging studies as tests to support or reject clinical hypotheses. When it comes time to modify prior probabilities on the basis of a test, would you rather have the prior probability of an involved clinician or a radiologist? I can think of arguments both ways. 

A systematic review in JAMA looked at the effect of clinical opinion on the accuracy of radiologic interpretation and seemed to find improvement in accuracy, though I don't know how clinically significant this improvement is.