If there are brain scan hallmarks for autism maybe a therapy could help address it as well if brain manipulations like that become a possible therapy. — TiredThinker

From this paper:

In this paper, we summarized 47 literatures involved in fMRI data classification between ASD individuals and TCs. Most researchers expected to derive the biomarkers of ASD through classification studies and have made some progress in deed, but the overall assessment of classification of ASD using fMRI data thus far falls short of biomarker standards. Despite this, several work directions may need to be paid more attention by researchers...

There is evidence, that at least some percentage of autism, corresponds to variation in cortical minicolumn structure, which are variations with a feature size too small for current fMRI technology to detect directly.

One treatment that I am curious about, which appears to be beneficial in some cases, is treatment with the hormone oxytocin. I haven't looked into trying it myself though. Perhaps next time I go to the doctor...