Steady State Free Precession (SSFP) in MRI

• SSFP is a sequence family that uses rapid, repeated RF excitations with short repetition times to establish a dynamic steady state between excitations.
• It enables high SNR efficiency and offers tradeoffs in contrast and artifact sensitivity, with variants like bSSFP preserving transverse coherence through balanced gradients.
• Advanced SSFP techniques exploit frequency dependence to encode relaxation, diffusion, and microvascular features while managing off-resonance and multi-component interference.

Steady State Free Precession (SSFP) denotes a family of rapid magnetic-resonance sequences in which repeated radiofrequency (RF) excitations are applied with short repetition time so that longitudinal and transverse magnetization reach a dynamic steady state rather than fully relaxing between excitations. In the balanced variant, balanced SSFP (bSSFP), all imaging gradients are fully balanced over each repetition so the net gradient moment is zero and transverse coherence is preserved, yielding high SNR efficiency together with pronounced sensitivity to off-resonance phase accrual (Plähn et al., 2023). Across MRI and NMR, SSFP encompasses true steady-state operation, phase-cycled acquisitions, and controlled transient regimes; its characteristic frequency dependence, long-range coherence-pathway structure, and sensitivity to intra-voxel frequency distributions make it both a source of artifacts and a mechanism for encoding relaxation, diffusion, exchange, and microvascular information (Assländer et al., 2018, Coudert et al., 2024).

1. Definition and sequence family

In standard NMR and MRI usage, “free precession” in SSFP refers to the evolution between RF pulses, during which magnetization undergoes Larmor precession about the static field or off-resonance axis together with relaxation via $T_1$ and $T_2$. For the symmetric balanced-type experiment, one repetition can be written as

$$\mathbf U = \mathbf E\, \mathbf R_z \mathbf R_y \mathbf R_z\, \mathbf E ,$$

where $\mathbf R_y$ is the RF rotation, $\mathbf R_z$ the phase accumulation during free precession, and $\mathbf E$ the relaxation operator (Assländer et al., 2018). In this sense, SSFP is not a separate physical effect from other pulsed NMR sequences; it is a repeated-pulse regime built from RF rotation, phase accrual, and relaxation.

Within the broader SSFP family, bSSFP is the case in which all gradient moments over each repetition are balanced, whereas unbalanced or spoiler-containing experiments such as FISP or PSIF alter the coherence pathways and desensitize the signal to inhomogeneous broadening at an SNR cost (Assländer et al., 2018). A deuterium-metabolic-imaging study at 9.4 T explicitly contrasts FLASH, FISP, and bSSFP: FLASH as RF-spoiled gradient echo, FISP as a spoiled or lowest-order SSFP-FID–like readout, and bSSFP as a fully balanced coherent steady state (Valsala et al., 30 Jan 2025). That comparison is representative of a wider usage: “SSFP” is the umbrella class, while balanced and unbalanced implementations select different tradeoffs among coherence preservation, off-resonance sensitivity, and sequence robustness.

A steady state in this literature means convergence to the fixed point of the repetition operator; a transient state means that the magnetization has not yet converged and evolves from repetition to repetition (Assländer et al., 2018). The distinction is operationally central. Classical steady-state designs are analytically simple and robust, but their signal trajectories occupy a relatively small subspace; fully transient designs can encode more parameters, but are typically more sensitive to field imperfections and model mismatch. Much of modern SSFP methodology is structured around navigating that tradeoff.

2. Steady-state geometry and off-resonance structure

The most recognizable property of bSSFP is its frequency profile. Signal magnitude varies periodically with resonance offset, with passbands and banding nulls; in passband operation, the broad, relatively flat high-signal regions are used, whereas transition-band operation exploits the narrow highly frequency-sensitive regions (Birk et al., 2021, Reynaud et al., 2018). This periodic off-resonance dependence is responsible both for the sequence’s high contrast efficiency and for its vulnerability to field inhomogeneity.

For phase-cycled bSSFP, the complex steady-state signal as a function of RF phase increment admits an elliptical signal model. One form used for parameter mapping is

$$S_{base}(r) = M(r)\,\frac{1-a(r)e^{i\theta(r)}}{1-b(r)\cos\theta(r)},$$

with

$$a(r)=E_2(r), \qquad b(r)=\frac{E_2(r)(1-E_1(r))(1+\cos\alpha(r))}{1-E_1(r)\cos\alpha(r)-E_2(r)^2(E_1(r)-\cos\alpha(r))},$$

and

$$M(r)=\frac{M_0(r)(1-E_1(r))\sin\alpha(r)}{1-E_1(r)\cos\alpha(r)-E_2(r)^2(E_1(r)-\cos\alpha(r))}.$$

As the phase cycle is varied, the complex samples trace an ellipse whose center, axis lengths, and cross-point encode $M$, $T_2$0, and $T_2$1, and thereby $T_2$2, $T_2$3, and off-resonance under a single-compartment model (Keskin et al., 2021).

The ellipse description has to be handled with phase consistency in multicompartment systems. For a single component, different phase conventions may appear equivalent because constant factors can be absorbed into an unknown global phase. For multiple components, that equivalence breaks down. The experimentally validated Bloch-consistent description predicts destructive interference zeros in mixed-component bSSFP profiles, whereas an alternative phase description predicts their general absence; water–acetone phantom data supported the destructive-interference case and thereby selected the Bloch-consistent convention (Plähn et al., 2023). This matters because near-zero signal in a phase-cycled profile is not necessarily a classic single-component banding null; in multicompartment systems it can arise from cancellation between components.

The same off-resonance structure also becomes sensitive to tissue microstructure when a voxel contains a distribution of resonance frequencies rather than a single frequency. In white matter, the measured bSSFP frequency profile can become asymmetric because susceptibility-driven intra-voxel frequency distributions depend on fiber geometry relative to $T_2$4. One asymmetry summary used for diffusion-related analysis is

$T_2$5

where $T_2$6 and $T_2$7 are the negative- and positive-offset peaks relative to the banding position (Birk et al., 2021). This asymmetry increases with inclination to $T_2$8, increases with FA, is almost unaffected by azimuth $T_2$9, and is nearly twice as high at 9.4 T as at 3 T.

3. Transient, hybrid, and multi-pathway regimes

Although SSFP is named for a steady state, several important developments deliberately exploit departures from classical steady-state operation. Hybrid-State Free Precession identifies adiabaticity conditions that trap the spin dynamics in a one-dimensional subspace: the magnetization direction adiabatically follows the instantaneous steady-state direction, while the absolute value of the magnetization remains in a transient state (Assländer et al., 2018). In SSFP terms, this retains steady-state-like directional robustness while allowing transient radial evolution and therefore richer parameter encoding. The paper introducing this regime explicitly positions it between conventional steady-state SSFP and fully transient methods such as MR fingerprinting.

A distinct controlled non-steady regime appears in multi-slice passband bSSFP fMRI. Interleaved 2D slice acquisition interrupts the strict per-slice steady state, but after a finite establishment period the signal becomes reproducible from one volume to the next; this was termed a pseudo-steady-state. Bloch simulations and measurements showed that the characteristic passband and transition-band structure is preserved, although the optimal flip angle shifts upward relative to conventional steady-state operation (Reynaud et al., 2018). This suggests that SSFP contrast can survive structured interruptions provided the transient is itself reproduc