Inversion Recovery (IR) Sequence

Inversion Recovery (IR) Sequence

1. Introduction: Historical Evolution and Clinical Purpose

Inversion recovery is the oldest and most physically elegant T1 contrast manipulation technique in clinical MRI, and it is the parent concept of three of the most diagnostically important sequences in routine practice: STIR, FLAIR, and DIR. It was not invented for clinical imaging — it was described by Erwin Hahn in 1949 [9], one year before the spin echo, as a nuclear magnetic resonance spectroscopy technique for measuring T1 relaxation times. The clinical MRI community subsequently recognised that the IR preparation pulse creates a degree of T1-based contrast control that is impossible to achieve with conventional SE or GRE sequences alone: by choosing the TI, the operator can selectively null the signal of any tissue whose T1 is known — effectively erasing a specific tissue from the image.

The clinical problems that IR was designed to solve are three distinct but related challenges:

Fat signal suppression (STIR): fat has the shortest T1 of any major tissue in the body. In musculoskeletal, body, and orbital MRI, the T1-bright fat signal competes with and often masks pathological tissue signal on T1 sequences, and reduces the conspicuity of oedema on T2 sequences. STIR — Short TI Inversion Recovery — uses a TI precisely chosen to null fat signal, producing fat-suppressed images that are B0-field-independent (unlike spectral fat saturation), making STIR the most robust fat suppression technique in regions of field inhomogeneity.

CSF signal suppression (FLAIR): in brain imaging, the T2-bright CSF signal fills the ventricles and surrounds the cortex, producing a bright background that masks periventricular, cortical, and juxtacortical lesions on T2 sequences. FLAIR — Fluid-Attenuated Inversion Recovery — uses a long TI to null CSF signal, revealing periventricular white matter lesions, cortical signal changes, and subarachnoid pathology that are invisible on standard T2.

White matter signal suppression (DIR): Double Inversion Recovery uses two sequential IR pulses to null both CSF and white matter simultaneously, leaving only grey matter visible — enhancing the conspicuity of cortical lesions in epilepsy and MS that are invisible on all other standard sequences.

In modern clinical MRI practice, IR sequences (primarily as STIR-TSE, FLAIR-TSE, 3D FLAIR-TSE, and DIR-TSE) are standard components of virtually every neurological and musculoskeletal protocol. The IR family represents the highest-impact application of T1 contrast manipulation in clinical MRI.

2. Physical Foundations

2.1 Pulse Sequence Logic and Signal Formation

The IR sequence begins with a 180° inversion radiofrequency pulse (the "inversion pulse"), which tips the equilibrium longitudinal magnetisation (Mz = +M₀) completely into the −z direction (Mz = −M₀). This is not a readout pulse — no signal is produced. After the inversion pulse, longitudinal magnetisation begins recovering from −M₀ back toward +M₀ at the rate determined by each tissue's T1. This recovery follows:

Mz(TI) = M₀ · (1 − 2·e^(−TI/T1))

where TI (Inversion Time) is the interval between the inversion pulse and the subsequent readout excitation pulse. The key consequence of this equation: the longitudinal magnetisation passes through zero at a specific TI for each tissue:

TI_null = T1 · ln(2) ≈ 0.693 · T1

At this TI, the tissue in question has zero longitudinal magnetisation — and therefore produces zero signal when the subsequent excitation pulse tips it into the transverse plane. This is the null-point selection: by choosing TI = 0.693 × T1_target, the operator selectively eliminates the signal of any tissue of known T1.

After the TI interval, the readout sequence is applied — typically a 90° excitation followed by a 180° refocusing pulse (creating a spin echo readout) or a TSE echo train. The readout produces the image, and all tissues except the nulled tissue contribute signal proportional to their Mz at time TI.

Inversion Recovery pulse sequence timing diagram

2.2 The Sign of Mz and Magnitude vs Phase-Sensitive Reconstruction

A critical technical detail: tissues that have recovered past the zero-crossing at the time of the readout excitation have positive Mz (returning toward +M₀), while tissues that have not yet reached zero have negative Mz (still between −M₀ and zero). The signal detected in MRI is proportional to the magnitude of Mz at TI — but magnitude reconstruction discards the sign information, meaning that two tissues on opposite sides of the null point may appear with similar brightness despite being in opposite magnetisation states.

Magnitude reconstruction (standard): all signal appears positive. This produces the classic IR contrast where all tissues with |Mz(TI)| > 0 contribute positively. Tissues near the null point appear dark, and their T1-ordering relative to other tissues is partially obscured.

Phase-sensitive (real-valued) reconstruction: preserves the sign of Mz. This produces a richer T1 contrast where tissues before the null point are dark or negative (inverted) and tissues past the null point are bright (recovering). Phase-sensitive FLAIR reconstruction, used in PSIR (Phase Sensitive Inversion Recovery) for myocardial T1 mapping and some brain applications, provides superior grey-white matter contrast compared with magnitude FLAIR.

2.3 Key Equations

Null point TI: TI_null = T1 · ln(2)

Typical null points:

• Fat (T1 ≈ 250 ms at 1.5T / 380 ms at 3T): TI_null ≈ 175 ms / 264 ms → STIR TI = 150–175 ms (1.5T); 200–230 ms (3T)
• CSF (T1 ≈ 3600 ms at 1.5T / 4000 ms at 3T): TI_null ≈ 2490 ms / 2770 ms → FLAIR TI = 2200–2400 ms (1.5T); 1700–1900 ms (3T)
• White matter (T1 ≈ 650 ms at 1.5T / 830 ms at 3T): TI_null ≈ 450 ms / 575 ms → DIR uses two sequential IR pulses to null both CSF and WM

Effective TR for IR-TSE: the TR must be long enough for adequate Mz recovery before the next inversion pulse. For STIR: TR ≥ 3 × T1_longest_tissue (typically ≥ 3000 ms). For FLAIR: TR ≥ 5000–10000 ms to allow CSF to recover adequately between excitations.

3. Key Parameters and Their Clinical Meaning

3.1 Parameter Table

3.2 Parameter Interdependence

The TI is linked to T1, which is field-strength-dependent. Any change in field strength requires TI recalibration. The TR must always be substantially longer than TI + readout duration; insufficient TR produces T1 contamination at the start of the next IR cycle (the steady-state Mz at the beginning of each cycle is not −M₀ but a partially recovered value, shifting the effective null point).

At 3T, the B1+ field inhomogeneity problem is particularly relevant for FLAIR: the 180° inversion pulse is imperfectly executed in regions of B1 heterogeneity (typically the posterior fossa and temporal lobes in brain imaging), producing residual CSF signal that mimics sulcal pathology. Adiabatic inversion pulses (which are insensitive to B1 variations) substantially improve FLAIR uniformity at 3T. All major vendors implement adiabatic inversion for 3T FLAIR.

4. Tissue Contrast Profiles

4.1 STIR (TI = 150–175 ms at 1.5T; 200–230 ms at 3T)

Critical clinical application: bone marrow pathology appears bright on STIR because it replaces the normally nulled yellow fat marrow — making STIR bone marrow oedema conspicuity higher than almost any other sequence.

STIR post-gadolinium contraindication: gadolinium shortens the T1 of enhancing tissues, shifting their null point to shorter TI values. At the standard STIR TI, tissues that were previously non-nulled may now be partially nulled by the gadolinium effect, and tissues that were previously nulled (fat) may no longer be nulled. This produces unpredictable and diagnostically misleading signal changes. STIR must never be acquired after gadolinium injection — this rule is absolute and applies universally across all MRIninja protocols.

4.2 FLAIR (TI = 2200–2400 ms at 1.5T; 1700–1900 ms at 3T)

Post-contrast FLAIR artefact: gadolinium in the CSF (from blood-brain barrier disruption, meningeal disease, or intrathecal administration) shortens CSF T1, shifting its null point. At the standard FLAIR TI, post-gadolinium CSF is no longer nulled and appears bright — this sulcal FLAIR hyperintensity can simulate SAH or leptomeningeal enhancement. When post-contrast FLAIR is the intended diagnostic sequence, it must be acquired within 5–8 minutes of injection (before significant CSF gadolinium accumulation) or after 30–40 minutes (delayed phase). The intermediate window of 10–25 minutes produces maximum artefact.

5. Vendor Implementations

Key Implementation Differences

Adiabatic inversion pulses at 3T: FLAIR quality at 3T critically depends on the inversion pulse design. Standard rectangular 180° pulses produce B1-dependent inversion flip angle — in regions where B1+ is 20% below target, the effective inversion is only ~140°, leaving residual signal in the "nulled" tissue. Adiabatic inversion pulses (Siemens uses an adiabatic half-passage variant; GE uses AFC pulses; Philips uses their MultiTransmit B1 compensation) are B1-insensitive and maintain near-complete inversion regardless of local B1 variations. At 3T, adiabatic inversion is essential for high-quality FLAIR — this is not optional.

VFA in 3D FLAIR: 3D FLAIR (SPACE/CUBE/VISTA) uses variable flip angle TSE readout to extend the echo train over the long acquisition needed for full-brain 3D coverage. The VFA schedule affects the T2 contrast weight — different vendor implementations produce slightly different grey-white matter contrast at identical nominal TI/TE parameters.

6. Clinical Applications Overview

7. Artefacts

8. Advanced Technical Parameters

Optimising TI for Field Strength

The TI null point is T1-dependent, and T1 values increase systematically at 3T compared with 1.5T. The STIR TI appropriate for 1.5T (approximately 150–175 ms) will not null fat at 3T (where fat T1 ≈ 380 ms, requiring TI ≈ 260 ms). The FLAIR TI appropriate for 1.5T (approximately 2200 ms) will over-invert CSF at 3T (where CSF T1 ≈ 4000 ms, requiring TI ≈ 1730 ms for null). Using the wrong TI is among the most common — and most easily preventable — protocol errors when sequences are migrated from 1.5T to 3T.

Practical verification: the correct STIR TI can be verified by confirming that subcutaneous fat and yellow bone marrow appear uniformly nulled (dark) on coronal STIR images. The correct FLAIR TI is verified by confirming that CSF in the ventricles and major cisterns appears uniformly dark. Any residual bright signal in these regions indicates TI miscalibration.

SAR Considerations

IR sequences add one 180° inversion pulse per TR to the standard TSE sequence. For short-TR applications (STIR) this is not problematic because the single inversion pulse adds minimal SAR per second. For FLAIR — with long TR ≥ 5000 ms and a short readout window — the single inversion pulse is also manageable. DIR, however, applies two high-power inversion pulses per TR, which at 3T can push SAR limits, particularly when combined with a long-ETL TSE readout. Adiabatic inversion pulses used for B1 compensation are also high-energy; automatic SAR monitoring and TR extension will be triggered if the total SAR exceeds limits.

Phase-Sensitive IR (PSIR)

PSIR preserves the sign of the longitudinal magnetisation at TI, producing images where tissues below the null point appear dark (negative Mz) and tissues above the null point appear brighter (positive Mz). Clinical applications:

• PSIR-FLAIR for brain: provides superior grey-white contrast and improved sensitivity for subtle cortical pathology compared with magnitude FLAIR
• PSIR (MOLLI-type) for myocardial T1 mapping: cardiac IR-GRE with phase-sensitive readout enables quantitative T1 mapping (described in the dedicated cardiac MRI sequence pages)
• PSIR for post-gadolinium myocardial delayed enhancement (LGE): provides consistent scar-to-blood pool contrast without precise TI optimisation

2D vs 3D IR

2D IR-TSE (standard STIR, standard FLAIR): slice-by-slice acquisition; motion within any slice's readout window does not affect other slices; independent TI optimisation possible per slice stack. The dominant clinical implementation.

3D IR-TSE (3D FLAIR, SPACE/CUBE/VISTA): isotropic voxels; full MPR capability; better periventricular and cortical lesion coverage with no inter-slice gaps; single TI for entire volume. Requires longer TR (8–12 minutes acquisition); entire volume affected by motion; B1 correction more critical. For brain MS and epilepsy protocols, 3D FLAIR is standard; for spine and body STIR, 2D remains preferred.

9. Comparison with Alternative Fat Suppression Techniques

The choice of fat suppression technique is one of the most clinically significant protocol decisions in MRI, and STIR occupies a specific niche in this landscape:

The key clinical rules derived from this comparison:

1. STIR is the preferred fat suppression technique when B0 homogeneity cannot be guaranteed (extremities, off-isocentre, near metallic hardware, obese patients)
2. STIR is contraindicated post-gadolinium — always use Dixon or SPAIR for post-contrast fat-suppressed sequences
3. Dixon is the preferred fat suppression for post-contrast abdominal and pelvic sequences
4. STIR provides the best bone marrow oedema sensitivity because it simultaneously nulls fat and provides T2-weighted contrast at TE 50–80 ms

10. Evidence Gaps and Ongoing Debate

Optimal TI values across field strengths and patient populations: the TI values for STIR and FLAIR are typically set from population-average T1 values. In paediatric patients (shorter T1 at equivalent field strength), elderly patients (longer T1 due to lower myelin content), and in liver/body applications where T1 can be shortened by pathology, fixed TI values may not provide optimal nulling. Individualised TI calibration remains an area of active research.

Post-contrast FLAIR timing: the safe window for acquiring post-contrast FLAIR without gadolinium-in-CSF artefact is empirically defined as < 5 minutes or > 30 minutes post-injection. The exact timing boundaries and the relationship to gadolinium dose, infusion rate, and blood-brain barrier integrity have not been formally characterised in prospective studies.

DIR standardisation: DIR is implemented on all major platforms but with significant differences in the VFA schedule, the two TI values, and the k-space filling strategy. No multivendor comparison study has established equivalent DIR acquisition parameters for consistent cortical lesion detection across platforms.

AI reconstruction for IR sequences: DLR has been applied to FLAIR and STIR in vendor-specific implementations. Whether DLR introduces false-positive FLAIR lesion appearance (over-sharpening mimicking focal signal changes) has not been systematically evaluated.

3D FLAIR vs 2D FLAIR for cortical lesion detection: 3D FLAIR isotropic is recommended over 2D FLAIR for the MS and epilepsy protocols on MRIninja (see the relevant protocol pages). However, the quantitative performance advantage over optimised 2D FLAIR in routine clinical practice (as opposed to expert centre research) has not been consistently demonstrated across all lesion types.

11. Miscellaneous and Related Topics

MOLLI and ShMOLLI (IR-GRE for Myocardial T1 Mapping)

Modified Look-Locker Inversion Recovery (MOLLI) and its abbreviated variant (ShMOLLI) use a series of IR-GRE acquisitions at different effective TI values, obtained across multiple cardiac cycles, to fit a T1 recovery curve for each pixel of the myocardium. This generates quantitative T1 maps that are clinically used for diffuse fibrosis (elevated T1), oedema (elevated T1), amyloidosis (reduced T1), and for extracellular volume fraction (ECV) calculation with pre- and post-contrast T1 maps. The cardiac IR-GRE family is described in the dedicated cardiac MRI sequence pages.

Driven Equilibrium with IR Preparation

STIR can be combined with a driven equilibrium RF pulse at the end of the readout to return residual transverse magnetisation to the longitudinal axis, shortening effective TR and enabling faster multi-slice acquisitions. Available on selected platforms.

Magnetisation Transfer (MT) Combined with FLAIR

MT pre-pulses applied before the inversion pulse of FLAIR further reduce background myelin signal and enhance the contrast of lesions lacking normal myelin. Used in selected research applications for MS lesion detection but not standard clinical practice.

12. Bibliography

A. Guidelines / Consensus / Society Recommendations

(No guideline is dedicated specifically to IR sequence design. IR sequences appear as components of indication-specific imaging guidelines — see the MS, epilepsy, and MSK protocol pages on MRIninja for the relevant references.)

B. Systematic Reviews / Meta-analyses

(Not applicable as a primary category for this foundational sequence page.)

C. Important Prospective / Original Studies

D. Technical MRI Papers

E. Landmark Historical References

End of text document — Inversion Recovery (IR) — MRIninja Sequence Page v1.0 — May 2026 Pulse diagram (2.1) and magnetisation curve diagrams (3.2) are rendered as separate interactive widgets on the MRIninja page.