For example, http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2859218/ notes that head circumference and cross-sectional MRI measurements of brain size show that children diagnosed with autism have (on average) a larger brain size.

However, looking at the charts (for a head circumference chart, see http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3869044/ ), while the data may be statistically significant, there's no clear separation between those diagnosed with autism and those with no indication of autism. Many of those with autism have a size smaller than the average of those without, and many of those without autism have a size larger than those with.

This means that head size or simple cross-sectional MRIs can't be used as a reliable diagnostic test.

Which structural differences where you meaning?