80-85% range. More recent studies with superior magnets/coils/software with MSK rads reading put those numbers at 95% plus.

In any case, these numbers are only meaningful when looking at populations and pretty meaningless in individual cases.

It’s silly when someone looks at a study showing that MRI has an accuracy of 95% in diagnosing ACL rupture and then tells every patient that there’s a 5% chance their MRI is wrong. That 5% refers to what we tend to observe in a series of patients in a given population, but has little meaning in any given individual patient. For example, if I see a ruptured acl on an MRI with clearly torn and lax fibers with a big fluid filled gap, a pivot shift contusion pattern, massive effusion, and an anteriorly subluxed tibia, then there is not a 5% chance that I’m wrong when I call that ACL ruptured. There’s a close to zero chance I’m wrong. I don’t really care what the exam shows in that case. I know I’m going to be right if it goes to arthroscopy.

On the other hand, when I see a questionably torn acl that I’m unsure of and call another MSK rad for an opinion and we collectively decide we think it’s ruptured and just to call it torn so the report isn’t wishy washy, there’s probably a 20% chance I could be wrong, and in that case the ortho’s exam is going to be probably much more accurate than the MRI. If I get a call that I’m wrong on a case like that I wouldn’t be shocked.

The studies tend to lump these cases together and spit out a number that indicates the average accuracy that you get from a mix of cases, but the degree of certainty in any individual case is much more fluid and may far exceed or fall short of the reported accuracy in any individual study. This is why I love multidisciplinary conferences when the rads and orthos actually get a chance to talk face to face and discuss the nuances of an individual case and the level of certainty. Good thing for elite athletes is there is usually direct conversations between the clinicians and rads about the findings. Not so much with the general population.