Couldn't ya just take the dmso layer with the elemicin dissolved in it, find a solvent that elemicin is insoluble in and is miscible with dmso (most are correct?) and just add an excess of that to crash out the majority of the elemicin?