The problem is that even if we can do a 'perfect' brain scan and figure out consistent correlations in how its structure differs for people with various disorde

The problem is that even if we can do a 'perfect' brain scan and figure out consistent correlations in how its structure differs for people with various disorders, it still doesn't really help us all that much. It would be like doing a perfect read of a hard-drive and seeing correlations without understanding (for the most part) what the ones and zeroes actually mean.