MRI Fat Suppression Techniques: Choosing CHESS, SPAIR, STIR or Dixon

Fat is bright on most of the sequences we run, and that becomes a problem when it hides edema, enhancement or a fluid-filled tear. Choosing among MRI fat suppression techniques is one of the most practical protocol decisions a technologist makes, and the right answer changes with field strength, field of view, anatomy, hardware and whether contrast is on board. Here is what each common method does well, where it fails, and what to check at the console.

How MRI fat suppression techniques work: two physical handles

Every method uses one of two differences between fat and water:

• Frequency. Fat protons resonate slightly slower than water. The shift is about 3.4 ppm, roughly 217 Hz at 1.5T and 435 Hz at 3T, according to a 2021 Radiologia Brasileira review of Dixon in musculoskeletal MRI. CHESS, SPAIR and water excitation target that frequency, and Dixon uses the phase difference it creates.
• T1. Fat recovers quickly after an inversion pulse. STIR times its excitation for the moment fat signal passes through zero.

Frequency-based methods need a homogeneous main field (B0); if it is off, the fat peak moves and the pulse misses. STIR barely cares about B0, but it nulls anything with a short T1.

Method by method

CHESS (spectral fat saturation)

A narrow RF pulse tuned to the fat frequency saturates fat, and a spoiler gradient dephases it before the imaging sequence runs, as described in RadioGraphics' 2013 review of MR lipid detection techniques. The 2021 Radiologia Brasileira review calls it the most commonly used method and reliable for post-contrast T1 imaging.

Where it struggles: large fields of view, off-center anatomy, irregular contours and metal. Del Grande and colleagues' 2014 RadioGraphics review of 3T musculoskeletal fat suppression shows unsuppressed subcutaneous fat at the edge of a large FOV, patchy suppression in the foot, and residual fat beside an interference screw.

According to Siemens' syngo MR E11 Ortho operator documentation, spectral fat saturation comes in a weak and a strong setting, with strong giving darker fat, and is suggested for bone bruise and edema work and for showing cartilage surface lesions. The same manual describes a joint-optimized fat suppression option for TSE, SE and SPACE sequences, opened from a button beside the fat suppression selection, supplied in 3T joint protocols and intended for foot and ankle, knee, hand and wrist, and elbow.

SPAIR

SPAIR inverts fat with an adiabatic, spectrally selective pulse instead of saturating it. The 2013 RadioGraphics review notes that this makes it less vulnerable to an uneven RF field (B1) than SPIR, at the cost of SPIR's time savings. Del Grande's review shows SPAIR giving more homogeneous suppression than CHESS in the thigh and calls it SNR-efficient for neurography. It is still frequency-based, so it shares CHESS's sensitivity to B0 problems and metal. Siemens' E11 Ortho documentation points to SPAIR for the hip at 3T.

STIR

An inversion pulse is followed by excitation at the inversion time (TI) at which fat signal crosses zero. Because it relies on T1 rather than frequency, STIR is the go-to when the field is the problem: large FOVs, off-center anatomy and hardware. In Del Grande's review STIR gave homogeneous suppression at the edge of a large FOV and fewer susceptibility artifacts near hip implants than SPAIR.

The trade-offs:

• Lower SNR than the other methods (Radiologia Brasileira 2021; RadioGraphics 2014).
• Not for post-contrast imaging. Gadolinium shortens T1, so enhancing tissue can be suppressed along with fat (Radiologia Brasileira 2021; RadioGraphics 2013).
• TI matters. Del Grande shows weak marrow fat suppression at a TI of 205 ms improving at 220 ms in a knee example. Siemens' E11 Ortho documentation gives starting TIs of about 160 ms at 1.5T and 200 ms at 3T when STIR replaces spectral fat sat near MR Conditional implants. Treat these as starting points; your approved protocol and radiologist set the final values.

Water excitation

Rather than removing fat, water excitation uses a pair of RF pulses separated by a short delay so that only water ends up tipped into the transverse plane (RadioGraphics 2014, which shows it on a 3D gradient-echo knee sequence). It depends on the fat-water frequency difference, so B0 homogeneity still matters.

Dixon

Dixon acquires in-phase and opposed-phase echoes, then calculates water, fat, in-phase and opposed-phase images, so one acquisition gives several contrasts. The Radiologia Brasileira review gives opposed-phase echo times of about 2.3 ms at 1.5T and 1.15 ms at 3T, and reports that multi-point Dixon gives more uniform suppression near metal and in tricky regions such as the hands, feet and neck. Del Grande shows Dixon outperforming CHESS beside a screw and delivering higher SNR than STIR.

Watch for:

• Water/fat swaps, where the reconstruction labels fat as water in part of the image. Del Grande shows swapping with severe field heterogeneity after a total knee replacement. Two-point Dixon is more vulnerable to B0 problems than three-point or multi-point versions (Radiologia Brasileira 2021).
• Scan time. The same review notes acquisition can take about three times the minimum of a conventional sequence.
• Combinations. Siemens' E11 Ortho documentation notes SEMAC metal artifact reduction cannot be combined with Dixon.

Quick comparison

Checklist when fat suppression fails at the scanner

• Is the anatomy at isocenter? Re-center the patient or table before anything else.
• Does the shim or adjustment volume cover the anatomy of interest rather than air?
• Is the FOV larger than the exam needs?
• Is there metal or hardware near the region? STIR or multi-point Dixon is usually the more robust choice.
• Is contrast on board? Do not use STIR for post-contrast series.
• On Dixon, check the water image for swapped regions before the patient leaves the table.
• Any change from the approved protocol goes through your radiologist.

If the images are also suffering from motion, our motion artifact checklist pairs well with this one.

Key takeaways

• Frequency-based methods (CHESS, SPAIR, water excitation) need a good shim and centered anatomy.
• STIR is the most field-tolerant option but has lower SNR and is not used after contrast.
• SPAIR is a common upgrade over CHESS where B1 is uneven, such as at 3T.
• Dixon gives uniform suppression and multiple contrasts, but check for water/fat swaps.
• Confirm settings in the manual for your own system and software version.

FAQ

What is the difference between fat sat and STIR?

Fat sat (CHESS) targets fat's resonance frequency, so it depends on a homogeneous field. STIR uses fat's short T1 with an inversion pulse, so it tolerates field problems better but has lower SNR and also suppresses other short-T1 tissue.

Can you use STIR after gadolinium?

It is not recommended. Gadolinium shortens T1, so enhancing tissue can be nulled along with fat. Post-contrast series normally use a frequency-based method such as spectral fat sat, per your facility's protocol.

Why is fat suppression uneven at the edge of my FOV?

The field is less homogeneous away from isocenter and over large FOVs, so spectral pulses miss the shifted fat peak. Re-center, tighten the FOV, or use STIR or Dixon if your protocol allows.

What is a water/fat swap in Dixon?

It is a reconstruction error in which fat and water are mislabeled in part of the image, often where the field is very uneven, such as near metal. Multi-point Dixon methods are generally more robust than two-point.

Questions about fat suppression on your own scanner are always welcome, and if your team wants hands-on help building these protocols, you can request applications training.

Sources

• Lins CF, Salmon CEG, Nogueira-Barbosa MH. Applications of the Dixon technique in the evaluation of the musculoskeletal system. Radiologia Brasileira, 2021
• Del Grande F, et al. Fat-Suppression Techniques for 3-T MR Imaging of the Musculoskeletal System. RadioGraphics, 2014
• Pokharel SS, et al. Current MR Imaging Lipid Detection Techniques for Diagnosis of Lesions in the Abdomen and Pelvis. RadioGraphics, 2013
• Manufacturer operator documentation: Siemens MAGNETOM, syngo MR E11, Ortho operator manual