Nice article. Unfortunately articles are not data. It mentions a 2013 study but yet again with a peer group of 32 that is much too small to tell you anything, and gives no further information. That 2013 I can't seem to find, but the p value if the article is correct on sample size would also be quite low, though slightly higher than the previously mentioned studies.

If the difference is so clear in an MRI why not run a simple blinded study where researchers look at brain scans and tell you who is transgender and who is not? Such a study would not only prove this theory outright, but would likely make a researcher's career.