What is the g-factor in SMS, and why does CAIPI matter?

The geometry factor, or g-factor, quantifies how much noise is amplified when the reconstruction unfolds overlapping slices using coil sensitivities; it is always greater than or equal to one, and it is worse when the overlapping slices are poorly distinguished by the coil. Controlled-aliasing methods such as CAIPIRINHA, and its echo-planar form blipped-CAIPI, deliberately shift the simultaneously excited slices relative to one another in the image, which improves how well the coil array can separate them and sharply reduces the g-factor penalty.

Does SMS reduce image signal-to-noise like in-plane acceleration?

Not in the same way. Because pure SMS acquires all k-space lines, it does not incur the square-root-of-acceleration signal-to-noise loss that in-plane undersampling does. The dominant penalty is the g-factor, which controlled aliasing keeps small. When SMS is combined with in-plane parallel imaging, the two penalties multiply, and the square-root term returns from the in-plane component.

What are the clinical benefits of SMS?

SMS can roughly halve the acquisition time of a clinical diffusion-weighted sequence, enable whole-abdomen diffusion imaging within a single breath-hold, and let functional MRI sample the whole brain in well under a second instead of a few seconds. Faster acquisition reduces motion sensitivity, improves patient tolerance, and in functional and diffusion imaging increases the amount of data collected per unit time.

Simultaneous Multi-Slice (Multiband) MRI Explained

Q: What is simultaneous multi-slice (multiband) MRI?

Simultaneous multi-slice (SMS), also called multiband, is an acceleration technique that excites two or more slices at the same time using a composite radiofrequency pulse, then separates the overlapping slices in reconstruction using the spatial information in a multi-channel receiver coil. Because it shortens the time needed to cover a given number of slices, it is most valuable for slice-hungry sequences such as echo-planar diffusion-weighted imaging and functional MRI.

Q: How is SMS different from in-plane parallel imaging?

In-plane parallel imaging, such as SENSE or GRAPPA, speeds acquisition by skipping phase-encode lines within each slice, which reduces the number of acquired samples and therefore lowers signal-to-noise by roughly the square root of the acceleration factor, in addition to a geometry-factor penalty. Pure SMS does not skip any k-space samples; every line of every slice is still acquired. As a result SMS has only a marginal intrinsic signal-to-noise penalty at fixed echo time, and its main cost is the geometry (g-) factor from separating the simultaneously excited slices.

Q: How high can the multiband factor go?

It depends on the anatomy, the field strength, and the number of receiver coil elements. In brain imaging at 3 tesla with a 32-channel coil and no in-plane acceleration, multiband factors as high as eight have been used routinely with acceptable image quality, and nine has been shown to be feasible. Body and abdominal imaging, with fewer usable coil elements and more field inhomogeneity, generally tolerates substantially lower multiband factors.

Q: Does SMS raise safety concerns such as SAR?

Composite multiband radiofrequency pulses excite several slices at once and can deposit more radiofrequency power than a single-slice pulse, so specific absorption rate management is a real design consideration, and the gradient activity used by blipped-CAIPI contributes to acoustic output and peripheral-nerve-stimulation limits. These quantities are governed by the applicable MR equipment safety standard, and scanner software enforces the operating limits during the exam.

Simultaneous multi-slice imaging excites several slices at once and then uses the spatial information in a multi-channel receiver coil to pull the overlapping slices apart — so a sequence that used to acquire slices one at a time can cover them in a fraction of the time. Because the technique does not skip any k-space samples, its intrinsic signal-to-noise penalty is small; the real cost is a geometry factor from separating the simultaneously excited slices, and that is exactly what controlled-aliasing methods such as CAIPIRINHA and blipped-CAIPI were designed to minimize.¹⁶⁹

For the slice-hungry echo-planar sequences at the center of modern neuroimaging and body diffusion — functional MRI and diffusion-weighted imaging — simultaneous multi-slice (SMS), also called multiband, is one of the most consequential acceleration advances of the past two decades.⁴⁸⁹

Introduction

In echo-planar imaging the bottleneck is usually not the in-plane readout but the number of slices needed to cover the anatomy. A whole-brain functional or diffusion protocol may require dozens of slices, each demanding its own excitation and readout, and the total time scales with the slice count. In-plane parallel imaging speeds up the readout within each slice but does nothing about the slice dimension itself.⁹¹¹

Simultaneous multi-slice attacks that bottleneck directly. Instead of exciting one slice, waiting, and exciting the next, an SMS sequence excites several slices with a single composite radiofrequency pulse and reads them out together. The slices overlap in the raw data, and the reconstruction separates them using the distinct sensitivity patterns of a multi-channel receiver coil — the same principle that underlies parallel imaging, but applied across slices rather than within a slice.¹⁹

The payoff is large and the cost is subtle. This article explains how SMS works, why its signal-to-noise behavior differs fundamentally from in-plane acceleration, what the geometry factor is and how controlled aliasing tames it, how high the multiband factor can realistically go, and where the clinical and safety considerations lie. The single most important conceptual point — and the one most often stated incorrectly — is that pure SMS does not undersample k-space, so it does not pay the square-root-of-acceleration signal-to-noise penalty that in-plane parallel imaging does.⁶⁹

Topic Explanation

Exciting several slices at once

A conventional two-dimensional sequence excites a single slice by applying a slice-selective radiofrequency pulse in the presence of a slice-select gradient. SMS replaces that single-slice pulse with a multiband pulse: a superposition of several single-slice pulses at different frequencies, each tuned to a different slice position. One excitation therefore tips the magnetization in several slices simultaneously, and one echo-planar readout records signal from all of them at once.¹⁹

The problem is that the readout cannot, by itself, tell which signal came from which slice — the simultaneously excited slices are superimposed in the acquired data. Separating them is where the receiver coil does its work.¹

Separating the slices with coil sensitivity

A modern MRI system receives signal on many coil elements, each with a spatially varying sensitivity. Because the simultaneously excited slices sit at different positions, each coil element "sees" them with a different weighting. Given a measured calibration of each element's sensitivity, the reconstruction can solve a system of equations to recover the separate slice images from the overlapped data. This is the same linear-algebra idea that parallel imaging uses to unfold in-plane aliasing, implemented here across slices; the widely used k-space form is called slice-GRAPPA.¹⁶

The quality of that separation depends entirely on how different the coil sensitivities are at the slice positions being separated. If two simultaneously excited slices are far apart and the coil array distinguishes them well, the separation is clean. If they are close together or the coil provides little differentiation along the slice direction, noise is amplified — and that amplification is quantified by the geometry factor.²⁶

SMS is not in-plane parallel imaging

It is worth stating the distinction plainly, because it drives everything downstream. In-plane parallel imaging (SENSE, GRAPPA) accelerates by acquiring fewer phase-encode lines within each slice. Fewer acquired samples means less signal collected, and the signal-to-noise ratio falls by roughly the square root of the acceleration factor, on top of a g-factor penalty.²

Pure SMS acquires every k-space line of every slice; nothing is skipped. The time saving comes from reading out several slices per excitation, not from collecting fewer samples. Consequently SMS carries only a marginal intrinsic signal-to-noise penalty at fixed echo time, and its dominant cost is the g-factor alone.⁶⁹ For the in-plane counterpart and its g-factor behavior, see our companion guide to parallel imaging and the g-factor.

Key Technical Principles

The geometry factor and the SNR penalty

The geometry factor, introduced in the original SENSE formulation of parallel imaging, measures how much noise the unfolding reconstruction amplifies at a given pixel. For acceleration achieved by k-space undersampling, the accelerated signal-to-noise ratio is:²

SNR_{acc} = \frac{SNR _{full}}{g R}, g \geq 1

The pixel-wise g-factor follows from the coil-sensitivity matrix $C$ and the receiver-noise covariance $Ψ$ :

g_{ρ} = [(C^{H} Ψ^{- 1} C)^{- 1}]_{ρρ} [(C^{H} Ψ^{- 1} C)]_{ρρ}

The crucial point for SMS is which terms apply. Because pure SMS does not skip k-space samples, the $R$ term — which represents the loss from acquiring fewer samples — is absent. The signal-to-noise behavior of pure SMS is therefore governed by the geometry factor alone:⁹

SNR_{SMS} \approx \frac{SNR _{single-slice}}{g}

When SMS is combined with in-plane undersampling by a factor $R_{in}$ , the penalties multiply and the square-root term returns from the in-plane component:

SNR \approx \frac{SNR _{full}}{g _{total} R _{in}}

This is why the literature describes SMS as having only a "marginal" intrinsic signal-to-noise penalty: at fixed echo time, with no in-plane acceleration, there is no $R$ loss to pay.⁹

Controlled aliasing: CAIPIRINHA and blipped-CAIPI

If the simultaneously excited slices are simply overlapped, the coil array must separate slices that may project onto nearly the same image locations, and the g-factor can be severe. Controlled aliasing solves this by shifting the slices relative to one another in the image. The technique, introduced as CAIPIRINHA (controlled aliasing in parallel imaging results in higher acceleration), applies a linear phase modulation to the multiband pulse so that each slice is displaced in the phase-encode direction.³

A linear phase increment $Δ ϕ$ applied line-to-line produces an image-domain shift:

Δ y = \frac{Δ ϕ}{2 π} FOV

A field-of-view/ $n$ shift between adjacent slices corresponds to $Δ ϕ = 2 π / n$ per line, with common choices of FOV/2, FOV/3, and FOV/4. By spreading the overlapping slices across the field of view, controlled aliasing lets the coil array separate them far more cleanly, cutting the g-factor penalty.³

For echo-planar imaging, the phase-only implementation produced a "tilted-voxel" blurring artifact. The blipped-CAIPI technique fixed this by generating the inter-slice shift with slice-select gradient blips played alongside the echo-planar phase-encode blips. In the original demonstration, blipped-CAIPI reduced the average g-factor penalty for threefold slice acceleration to under one percent at 3 tesla, compared with about thirty-two percent without the shift; for combined threefold slice and twofold in-plane acceleration the penalty fell to about nineteen percent at 3 tesla and ten percent at 7 tesla, versus roughly forty-one and twenty-three percent without it. These figures are specific to the coils, field strengths, and simulation method of that study, not universal constants, but they illustrate how large the controlled-aliasing benefit can be.⁶

Acceleration factors and leakage

The net acceleration of an SMS acquisition combines the multiband (slice) factor with any in-plane factor:

R_{total} = MB \times R_{in-plane}

In one 7-tesla functional demonstration, four simultaneous slices combined with fourfold in-plane undersampling achieved sixteenfold total acceleration.⁴ A separate 3-tesla evaluation found multiband factors up to eight usable routinely, and nine feasible, in brain imaging with a 32-channel coil and no in-plane acceleration.¹²

Separating many simultaneously excited slices is imperfect, and residual signal can "leak" between slices. That residual aliasing is quantified by the leakage (L-) factor, and the split slice-GRAPPA reconstruction was developed specifically to suppress inter-slice leakage and improve time-series stability in accelerated diffusion and functional imaging.⁷¹² The following table summarizes how the two acceleration families differ.

Property	In-plane parallel imaging	Pure SMS / multiband	SMS + in-plane
How speed is gained	Skip phase-encode lines within a slice	Excite and read several slices at once	Both, combined
k-space undersampled?	Yes	No	Partly (in-plane only)
SNR penalty	$g \cdot R$	$g$ only (marginal at fixed TE)	$g_{total} \cdot R_{in}$
Effect on echo time	Can shorten TE	Unchanged at fixed TE	Depends on in-plane factor
Main artifact risk	Residual in-plane aliasing	Inter-slice leakage	Both
Typical factors (brain EPI, 3 T)	2–3	up to 8	product of the two

Clinical Impact

The clinical value of SMS is concentrated where slice coverage is the bottleneck: diffusion-weighted imaging and functional MRI. In these echo-planar sequences the slice dimension dominates the scan time, and SMS converts almost directly into faster exams or richer data.

In neuroimaging, SMS enabled a step change in temporal and angular sampling. Multiplexed echo-planar imaging brought whole-brain functional MRI to sub-second repetition times, and sped up high-angular-resolution diffusion imaging severalfold.⁵ The Human Connectome Project built on SMS to acquire whole-brain 2-millimeter functional data in about 0.7 seconds and to cut total diffusion acquisition time roughly threefold.¹⁰¹¹

The benefits translate directly to clinical body and oncologic imaging. In prostate diffusion-weighted imaging, SMS reduced acquisition time by about half — from roughly six minutes to about three — with equivalent image quality and apparent diffusion coefficient agreement.¹³ In abdominal diffusion imaging, SMS enabled whole-abdomen coverage in a single breath-hold of roughly twenty-three seconds, compared with around three minutes of respiratory-triggered acquisition, with comparable apparent diffusion coefficients in most organs, though with somewhat lower signal-to-noise and lower measured liver values that must be weighed.¹⁴ For how apparent diffusion coefficient reproducibility is assessed, see diffusion-weighted imaging and quantitative ADC QC.

Faster acquisition also means less time for the patient to move, better tolerance, and more opportunities to repeat or extend a protocol within a fixed appointment. These advantages are real, but they are strongest in the brain studies where the technique was developed; body applications rest on smaller feasibility studies and warrant protocol-specific validation.¹³¹⁴

Practical Optimization Tips

SMS rewards deliberate protocol design. A few principles help a program get the speed without paying for it in image quality.

Match the multiband factor to the coil and anatomy. High multiband factors depend on many well-separated coil elements along the slice direction. The factors of eight achievable in a 32-channel brain study do not transfer to body imaging with fewer usable elements and more field inhomogeneity.¹²
Always use controlled aliasing for EPI. Blipped-CAIPI (or an equivalent FOV-shift scheme) is what keeps the g-factor penalty small; without it, the noise amplification from separating overlapping slices can erase the benefit.⁶
Watch for inter-slice leakage. Residual signal between simultaneously excited slices can mimic or obscure findings. Split slice-GRAPPA and attention to the leakage factor help, and a careful reader should know which slices were acquired together.⁷¹²
Combine SMS and in-plane acceleration judiciously. The two penalties multiply, and the in-plane component reintroduces the square-root signal-to-noise loss. Push slice acceleration first, add in-plane acceleration only as needed.⁶⁹
Validate quantitative outputs. Before relying on apparent diffusion coefficients or functional statistics from an accelerated protocol, confirm agreement against the unaccelerated reference for your scanner and anatomy.¹³¹⁴
Calibrate well. The slice-separation reconstruction depends on an accurate coil-sensitivity or calibration acquisition; a poor calibration shows up as leakage and elevated g-factor.

Regulatory Considerations

SMS is a pulse-sequence and reconstruction technique, so it operates inside the existing MRI safety and quality framework rather than creating a new regulatory category — but two of its physical mechanisms deserve explicit attention. First, the composite multiband radiofrequency pulse excites several slices at once and can deposit more radiofrequency power than a single-slice pulse, making specific absorption rate (SAR) a genuine design and operating consideration; some implementations deliberately mitigate this.⁵⁹ Second, the gradient blips that blipped-CAIPI adds contribute to gradient activity, which bears on acoustic noise and peripheral-nerve-stimulation limits.

These quantities are governed by the applicable MR equipment safety standard (IEC 60601-2-33), and the scanner's software enforces the operating limits — normal and controlled operating modes for SAR and dB/dt — during every exam. The physicist's role is to verify that accelerated protocols operate within those limits and that the clinical benefit justifies any added radiofrequency burden. For the underlying radiofrequency-safety context, see SAR and RF safety in MRI.

Quality assurance is the other regulatory touchpoint. MRI quality control and accreditation — under ACR MRI accreditation and ACR–AAPM quality-control guidance — require periodic performance evaluation by a qualified medical physicist or MR scientist, and an accelerated protocol should be validated as part of that program rather than assumed equivalent to its unaccelerated parent. MRI is non-ionizing, so it falls outside radiation-machine registration programs; the binding requirements come from accreditation and equipment-safety standards. For the phantom-based foundation of that QC program, see the ACR MRI phantom and QC.

Frequently Asked Questions (FAQs)

What is simultaneous multi-slice (multiband) MRI?

SMS excites two or more slices at the same time with a composite radiofrequency pulse and separates the overlapping slices in reconstruction using a multi-channel receiver coil. Because it shortens the time to cover a given number of slices, it is most valuable for echo-planar diffusion and functional imaging.¹⁹

How is SMS different from in-plane parallel imaging?

In-plane parallel imaging skips phase-encode lines within each slice, losing signal-to-noise by roughly the square root of the acceleration plus a g-factor penalty. Pure SMS skips no k-space samples, so it has only a marginal intrinsic signal-to-noise penalty at fixed echo time, dominated by the g-factor.²⁹

What is the g-factor, and why does CAIPI matter?

The g-factor measures noise amplification when the reconstruction unfolds overlapping slices; it is always at least one and worse when the coil distinguishes the slices poorly. CAIPIRINHA and blipped-CAIPI shift the slices in the image so the coil separates them better, sharply reducing the g-factor penalty.³⁶

Does SMS reduce signal-to-noise like in-plane acceleration?

Not the same way. Pure SMS acquires all k-space lines, so it avoids the square-root loss of in-plane undersampling; the dominant penalty is the g-factor. Combining SMS with in-plane acceleration multiplies the penalties and reintroduces the square-root term.⁶⁹

How high can the multiband factor go?

It depends on anatomy, field strength, and coil elements. Brain imaging at 3 tesla with a 32-channel coil has used multiband factors up to eight routinely, with nine feasible; body imaging generally tolerates much lower factors.¹²

Does SMS raise safety concerns such as SAR?

Multiband pulses can deposit more radiofrequency power than single-slice pulses, so SAR management matters, and the added gradient blips bear on acoustic noise and nerve-stimulation limits. These are governed by the MR equipment safety standard and enforced by scanner software.⁵⁹

Key Takeaways

SMS excites several slices at once and separates them with coil sensitivity, cutting scan time nearly in proportion to the multiband factor.¹⁹
Because pure SMS does not undersample k-space, it carries only a marginal intrinsic signal-to-noise penalty at fixed echo time — the dominant cost is the g-factor, not a square-root loss.⁶⁹
Controlled aliasing (CAIPIRINHA, blipped-CAIPI) shifts the simultaneously excited slices to improve coil separation and can reduce the g-factor penalty dramatically.³⁶
Net acceleration is the product of the multiband factor and any in-plane factor; combining them multiplies the penalties.⁴⁶
The biggest clinical gains are in echo-planar diffusion and functional imaging — halved diffusion scans, single-breath-hold body DWI, and sub-second whole-brain fMRI.⁵¹⁰¹³¹⁴
SMS lives inside the existing MRI safety and QC framework; SAR from multiband pulses and acoustic/nerve-stimulation effects from gradient blips are the mechanisms to watch.⁵⁹

Conclusion

Simultaneous multi-slice imaging is a rare acceleration technique that buys speed without the usual signal-to-noise bill. By exciting several slices at once and leaning on the receiver coil to separate them, it sidesteps the k-space undersampling that makes in-plane parallel imaging costly, leaving the geometry factor as the principal price — a price that controlled aliasing was engineered to keep small. The result is faster diffusion and functional imaging, more data per unit time, and better patient tolerance.

Realizing those benefits in a clinical program takes judgment: matching the multiband factor to the coil and anatomy, insisting on controlled aliasing for echo-planar sequences, watching for inter-slice leakage, combining SMS and in-plane acceleration deliberately, and validating quantitative outputs against the unaccelerated reference. Treated as a tool to be tuned rather than a switch to be flipped, SMS is one of the most useful additions to a modern MRI protocol library.

How DRPS Can Help

Diagnostic Radiation Physics Services supports MRI facilities across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware with MRI physics testing, accelerated-protocol validation, signal-to-noise and g-factor evaluation, ACR phantom and accreditation support, and radiofrequency-safety review — all performed by qualified medical physicists.

A strong MRI program does not just turn acceleration on; it verifies that an accelerated protocol still delivers the image quality and quantitative accuracy the clinical question demands. DRPS helps facilities fold that verification into their medical physics consulting and accreditation support workflow.

Related Resources

References

Larkman DJ, Hajnal JV, Herlihy AH, Coutts GA, Young IR, Ehnholm G. Use of multicoil arrays for separation of signal from multiple slices simultaneously excited. J Magn Reson Imaging. 2001;13(2):313-317. doi:10.1002/1522-2586(200102)13:2<313::AID-JMRI1045>3.0.CO;2-W. doi.org
Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42(5):952-962. PubMed
Breuer FA, Blaimer M, Heidemann RM, Mueller MF, Griswold MA, Jakob PM. Controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA) for multi-slice imaging. Magn Reson Med. 2005;53(3):684-691. doi:10.1002/mrm.20401. doi.org
Moeller S, Yacoub E, Olman CA, et al. Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn Reson Med. 2010;63(5):1144-1153. doi:10.1002/mrm.22361. doi.org
Feinberg DA, Moeller S, Smith SM, et al. Multiplexed echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging. PLoS One. 2010;5(12):e15710. doi:10.1371/journal.pone.0015710. doi.org
Setsompop K, Gagoski BA, Polimeni JR, Witzel T, Wedeen VJ, Wald LL. Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planar imaging with reduced g-factor penalty. Magn Reson Med. 2012;67(5):1210-1224. doi:10.1002/mrm.23097. doi.org
Cauley SF, Polimeni JR, Bhat H, Wald LL, Setsompop K. Interslice leakage artifact reduction technique for simultaneous multislice acquisitions. Magn Reson Med. 2014;72(1):93-102. doi:10.1002/mrm.24898. doi.org
Feinberg DA, Setsompop K. Ultra-fast MRI of the human brain with simultaneous multi-slice imaging. J Magn Reson. 2013;229:90-100. doi:10.1016/j.jmr.2013.02.002. doi.org
Barth M, Breuer F, Koopmans PJ, Norris DG, Poser BA. Simultaneous multislice (SMS) imaging techniques. Magn Reson Med. 2016;75(1):63-81. doi:10.1002/mrm.25897. doi.org
Setsompop K, Kimmlingen R, Eberlein E, et al. Pushing the limits of in vivo diffusion MRI for the Human Connectome Project. Neuroimage. 2013;80:220-233. doi:10.1016/j.neuroimage.2013.05.078. doi.org
Uğurbil K, Xu J, Auerbach EJ, et al. Pushing spatial and temporal resolution for functional and diffusion MRI in the Human Connectome Project. Neuroimage. 2013;80:80-104. doi:10.1016/j.neuroimage.2013.05.012. doi.org
Xu J, Moeller S, Auerbach EJ, et al. Evaluation of slice accelerations using multiband echo planar imaging at 3 T. Neuroimage. 2013;83:991-1001. doi:10.1016/j.neuroimage.2013.07.055. doi.org
Weiss J, Martirosian P, Taron J, et al. Feasibility of accelerated simultaneous multislice diffusion-weighted MRI of the prostate. J Magn Reson Imaging. 2017;46(5):1507-1515. doi:10.1002/jmri.25665. doi.org
Ohno N, Yoshida K, Ueda Y, et al. Diffusion-weighted imaging of the abdomen during a single breath-hold using simultaneous-multislice echo-planar imaging. Magn Reson Med Sci. 2023;22(2):253-262. doi:10.2463/mrms.mp.2021-0087. doi.org