Noise is the point-to-point variation in image density that does not contain useful information. The magnitude of noise is indicated by the percentage standard deviation of the CT numbers within a region of interest in the image of a uniform substance (generally water), relative to the difference in CT numbers between water and air.
Statistical variability in the response of the assay from all sources, including: variation in actual quantities of reagents added or their concentration, instrument output variability, geometry of plates, uncertainty or variability in background, etc.
In the context of quality management, noise is essentially variability. For example, if you are making ketchup, noise in the process comes from variations in the quality of incoming tomatoes, in changes in ambient temperature and humidity, in variations in machinery performance, in variations in the quality of human factors, etc.