Hi,
Thanks for that, so the sensor is the major project.
You would need to sit down with a materials expert, and work out what has good repeatable rotational stress figures.
Not sure of the torque, but a metal would probably be better, just you may have very small diameters where the rotational deflection would occur.
If it is research into rotational deflection sensing, then what you are using may be fine, proof of concept, keeping in mind its limitations.
A crude setup would be to have thin disks on either end of your torsion element.
Punch one hole on the outer edge of each, use photo-interrupters to detect when the hole passes thorough, then do some maths to work out rotational deflection.
The larger the disc diameter the higher the resolution.
Is just a thought, not sure how low in RPM you want to go, but you might need more holes in the discs.
Tom... 
