I think the first step is to understand where the noise is coming from.
If you look at the transducer signal with an oscilloscope (across a 250 ohm resistor will give 5V = 20 mA), is the noise at any obvious frequency?
Does it go away when you turn off adjacent machinery?
How long is the wiring from transducer to ADC?
If you do this with short wires in the lab/workshop, is the noise still there?
How noisy is the power supply to the transducer?
What are the earthing and screening arrangments?
If there is above Nyquist noise that could enter the system then it is better to filter it out with a simple R-C filter, because there is always a danger that some above-Nyquist frequencies could alias down into the pass-band if using digital filtering. For this reason the best approach is to find where the noise is arising and KILL IT AT SOURCE.
