This study sounds flawed on multiple levels, though I'd probably have to read the original study to be sure on a few issues. The biggest problem is this: is it reasonable to assume that linguistic differences account for the differences in the MRIs of the participants?
My first thought would be that it's a result of different pedagogical methods in China and English-speaking countries. According to the article, "They also note that the use of the abacus in many Asian schools may encourage the brains of students in this region to think spatially and visually about numbers." I'd give that a lot more weight than I'd give to the linguistic differences.
The majority of the study seems to hinge on a rather suspect claim:
Reiman and his colleagues suggest that the Chinese language’s simple way of describing numbers may make native speakers less reliant on language processing when doing maths. For example, “eleven” is “ten one” in Chinese “twenty-one” is “two ten one”.
Morphologically speaking, the Chinese number system is marginally "simpler" than the English number system. And even if the morphological simplicity of a language's number system contributes to its speakers math skills, why would it result in a visual understanding of numbers?
And anyway, this is probably ridiculously easy to disprove. All you have to do is find a language with a more morphologically complex number system whose speakers are better at math than English speakers, or, conversely, a language with a more morphologically simple number system whose speakers are worse at math than English speakers.
This study is just one more in a long list of studies that have tried and failed to prove the popular notion that language serves as some sort of programming language for the brain that affects a speaker's ability to think in certain ways.
Language Log has covered this sort of thing many times, so I'll try to dig up some of the relevant links and let you read opinions from real experts.
