Temperature Measurement Circuits in Power Systems

May I suggest the use of a bridge circuit, with the NTC as one leg, the other three legs of fixed resistance values equal to the nominal NTC value at the nominal temperature you want to measure at. A bridge allows a differential output, fed into an appropriate instrumentation type op-amp will give you the best noise rejection at the amplifier. Use the lowest value of NTC that gives you the needed resolution of signal. As suggested, use twisted, shielded (use an active guard drive to the shield, don't ground it) cable to connect it to the input circuitry as it sounds like the cabling must run some distance. Pay attention to power supply decoupling of your instrumentation amp and install it inside a well grounded case. You may also have to keep any power supplies powering this amplifier at a distance to prevent additional power line/supply coupling into the amplifier, decouple the power supply bus at the amplifier as well and shield the power cable and ground it.

Make sure the voltage or current source feeding the bridge is also well decoupled at the bridge. I would recommend a voltage source as it should have a lower output impedance than with a current source. It is always easier to prevent noise getting into the system rather than removing it from the system later. The instrumentation output can be further filtered if necessary to average out the signal even more if necessary. This can be tailored to the system's response time to temperature changes.

Noisy environments are always a challenge to work in, you may find that additional steps may be needed to prevent noise in your signal. It is difficult to say just what you will run into given the minimal information, it is difficult for you to even tell us precisely what you have to begin with.