Biasing the PRT with 100 ohms at 5V will produce about 50mW of heat in the sensor at midpoint. That will cause enough self heating in the PRT to throw your reading off a lot more than 1/100th of a degree. How much varies, depending on the sensors thermal mass and dissipation characteristics.
To reduce self heating, you should either:
1) Bias it at a much lower current, and use an op-amp to gain the voltage up.
2) Use a second uC pin to turn the bias current on just a few uSecs before reading it. Then turn the power off for a while. If the on/off time ratio is small, self heating can be minimized.
Another thing to consider when using PRTs is that, since the sensor resistance is so low, the resistance of interconnecting wire can be another source of error. If the sensor is located vary far from the power source and measurement electronics, you may need to do a 4-wire connection.