I'm just guessing here, but if even 1% of autism cases were caused by cysts or other treatable brain abnormalities, and hypothetical interventions had a 25% success rate, you'd end up spending $1.2M per case of autism cured (using the other poster's figure of $3K for an MRI scan). That seems well worth it to me. Certainly the cost delta between a functional member of society and someone that requires costly support their whole life is way more than $1.2M over a lifetime.
Of course this is like Drake's equation; taking an unknown and breaking it down into made-up numbers doesn't actually increase certainty, but fortunately, these numbers are a lot more knowable than the coefficients in Drake's equation. It's simply that I don't know them.