Noises in MRI images primarily are Rician noise, Gaussian Noise and Rayleigh noise. For details, see https://biomedpharmajournal.org/vol11no3/noise-issues-prevailing-in-various-types-of-medical-images/#:~:text=Noises%20in%20MRI%20images%20primarily,it%20converges%20to%20Rayleigh%20distribution.
In terms of the physical mechanisms contributing to the noise, thermal noise resulting from microscopic fluctuations in magnetization is generally the primary source. However, in functional MRI, where time series are analyzed at each voxel in the image, physiological noise is a significant factor as well (due to cardiac pulsatility of blood, changes in the B0 field due to respiration, motion, etc.). This paper by Triantafyllou et al is a seminal paper on the impact of thermal vs. physiological noise on fMRI Article Comparison of physiological noise at 1.5 T, 3 T and 7 T and ...
. There is also a follow-up from 2011 examining this using multi-channel receive coils
Article Physiological noise and signal-to-noise ratio in fMRI with m...