Personally, I'd go with the catalog. Or at worst, dyno-test your motor.

If you like doing it the hard way, I can try to explain.

It is possible to mathematically calculate the torque of an electric motor. For simplicity's sake, let's take the 3-arm DC motor (the kind you find on your everyday R/C cars, cd players, etc)

For a current-carrying conductor in an electric field, let:

F=BILsinA (scalar eq)

where F=Force acting on conductor(N), I=current on conductor(A), L = length of conductor immersed in electric field(m), and A is the included angle between the direction of I and the direction of B.

We can now take a motor's coil and divide the coil into very short current-carrying wires. Each elemental length of the coil's wire carries the same amount of current but with different directions. In turn, these elemental lengths of the coil will make contributions to the overall force generated by the coil.

Now we know that we can split the coil into very small pieces of wire that experience small magnetic forces on their own. Like you mentioned, calculating torque is now only a matter of multiplying the distance of these elemental wires by the distance from the axis. However, you have to remember that every single piece of these very short wires vary in distance from the axis.

You also have to remember that as the rotor turns, the electric field strength varies. (depending on design, magnet quality etc)

So there, if you can mathematically describe the magnetic force, coil shape and timing, you can take these equations and stuff them into eg MAPLE. However, the number you get (assuming you get the mathematical model correct) would be the GROSS torque rating, and IMHO, not as useful as simply dyno-testing your motor.