Chapter detection for roman numerals??

Hi, when I convert I would like calibre to detect roman numeral chapters. These chapters don't have the word "chapter" in them, just:
I
II
III
IV
etc..

I figured something like:
[IVXLCDM]+ but I'm not sure exactly how to write it into the
"detect chapter" Xpath expression under the Structural Detection section. Some of the books have the <h2> class tag and some of them have the normal <p> class tag. What would be the exact expression I would use in the xpath line that would detect any combination of "IVXLCDM" as chapters?

Thanks!