there are nuances that im sure someone will correct, but this is how i understand it:
-do 3 sensory studies: the radial/median difference to the thumb, the transcarpal using median/ulnar, and the ring finger using median/ulnar. add up the changes as compared to the median conductions (for ex: if the transcarpal showed a median latency of 2.0 and the ulnar showed 1.5, then you'd have a 0.5 difference). if the sum is > 0.9, then you've got CTS. needle for axonal damage.
apparently its the most sensitive for CTS when there is a doubt or if ther eis a mild case. still need to do needle to deteremine axonal damage. also, there have been a few more recent papers (by robinson i think) that looked at which of the three studies to do first and the best algorythm. i dont remember exactly, but i think you do the ring finger first and then if the difference is greater than a certain number you stop. not 100% sure on that, though