Detailed protocols are available in the literature. See the attachment. In my working life we used a somewhat simpler version: reference 12 in the attachment. Please note that this method is rather complex and requires analytical skill and training. It is recommended to exercise first on reference foodstuffs with well documented glycemic index.
At present I have no access to the original (colorimetric) version of the Englyst method (= reference 12 of the attachment to my reply). But I think the leaflet by Megazyme attached to the reply of Kashif Ameer gives a good description of the colorimetric version. Here the glucose oxidase-peroxidase (= GOPOD) assay is applied for the quantitative determination of rapidly and slowly liberated glucose (G20 and G120), i.e. after 20 and 120 minutes digestion with amyloglucosidase, respectively. Resistant starch is finally obtained by amyloglucosidase digestion of the insoluble residue after solubilization in aqueous alkali. Of course, the GOPOD assay has to be calibrated by measuring blanks and glucose standards.