Fat suppression is the process of utilizing specific MRI parameters to remove the deleterious effects of fat from the resulting images , e.g. with STIR, FAT SAT sequences, water selective (PROSET WATS - water only selection, also FATS - fat only selection possible) excitation techniques, or pulse sequences based on the Dixon method.
Spin magnetization can be modulated by using special RF pulses. CHESS or its variations like SPIR, SPAIR (Spectral Selection Attenuated Inversion Recovery) and FAT SAT use frequency selective excitation pulses, which produce fat saturation.
Fat suppression techniques are nearly used in all body parts and belong to every standard MRI protocol of joints like knee, shoulder, hips, etc. Imaging of, e.g. the foot can induce bad fat suppression with SPIR/FAT SAT due to the asymmetric volume of this body part. The volume of the foot alters the magnetic field to a different degree than the smaller volume of the lower leg affecting the protons there. There is only a small band of tissue where the fat protons are precessing at the frequency expected, resulting in frequency selective fat saturation working only in that area. This can be corrected by volume shimming or creating a more symmetrical volume being imaged with water bags.
Even with their longer scan time and motion sensitivity, STIR (short T1/tau inversion recovery) sequences are often the better choice to suppress fat. STIR images are also preferred because of the decreased sensitivity to field inhomogeneities, permitting larger fields of views when compared to fat suppressed images and the ability to image away from the isocenter.
See also Knee MRI.
Sequences based on Dixon turbo spin echo (fast spin echo) can deliver a significant better fat suppression than conventional TSE/FSE imaging.

In MR, saturation is a nonequilibrium state with no net magnetization. The same amount of nuclear spins is aligned against and with the magnetic field. Saturation methods like FatSat, SPIR etc., work with a frequency selective saturation pulse for a specific chemical shift applied before the actual sequence starts. This saturation pulse adjusts the magnetization from tissue components to zero. The hydrogen nuclei of fat and water resonate at different frequencies, which makes it possible to excite just the fat with repeatedly applying RF pulses at the Larmor frequency with interpulse times compared to T1. The resulting signal is then destroyed with a gradient pulse (Spoiler Gradient Pulse). Fat is the chemical compound to be saturated at a fat saturation sequence. When the actual sequence follows, (e.g., a spin echo sequence) the unwanted suppressed component will not resonate.

See also Saturation Recovery.

The gradient recalled echo MRI sequence generates gradient echoes as a consequence of echo refocusing. The initial slice selective RF pulse applied to the tissue is less than 90° (typically rotation angles are between 10° and 90°). Immediately after this RF pulse, the spins begin to dephase.
Instead of a refocusing 180° RF pulse, reversing the gradient polarity produces a gradient echo. A negative phase encoding gradient and a dephasing frequency encoding gradient are used simultaneous. The switch on of the frequency encoding gradient produces an echo caused by refocusing of the dephasing, which is caused by the dephasing gradient.
TR and flip angle together control the T1, and TE control T2* weighting.