Slice Gap

MRIninja Knowledge Base | MRI Parameter Deep Dive Version 1.0 — August 2026

1. Introduction and General Purpose

Slice gap is the empty spatial interval deliberately left between the nominal edges of consecutive 2D slices. It is the smallest and most easily overlooked of the three coverage-defining parameters on this platform — alongside Slice Thickness and Number of Slices — yet it carries a genuinely distinct identity and a genuinely distinct risk profile from both.

Slice thickness answers "how much tissue does each slice represent"; number of slices answers "how many are needed, and can the sequence deliver that many in time." Slice gap answers a third, different question: "how much of the anatomy between slices is deliberately left unsampled, and is that acceptable for this clinical question?" Unlike thickness (which trades SNR for resolution) or slice count (which trades acquisition time and SAR for coverage), gap trades image-quality robustness (freedom from cross-talk artefact) directly against literal, physical coverage — tissue that falls entirely within a gap is not sampled at all, at any resolution, by that acquisition. This is the central tension this page exists to resolve.

Why the parameter exists at all: it would be simplest, in principle, to always acquire slices back-to-back with zero gap. The reason gap exists as a parameter — rather than always being set to zero — is a direct consequence of imperfect RF slice-selective excitation, developed fully in Section 2.

2. Physical Foundations

2.1 The Origin of Cross-Talk

A slice-selective RF pulse, in combination with a slice-selection gradient, never produces a perfectly rectangular excitation profile in practice. Real RF pulses (sinc-shaped or otherwise apodised) excite a central region at close to full flip angle, flanked by transition "skirts" of partial excitation that fall off gradually rather than sharply at the nominal slice boundary. When two slices are positioned contiguously (zero gap), the transition skirt of one slice overlaps the nominal territory of its neighbour, partially exciting tissue that "belongs" to the adjacent slice [1]. This partial excitation, delivered before that tissue has had time for full T1 recovery from its own slice's excitation, produces a measurable signal and contrast disturbance — the cross-talk artefact.

Two features of this mechanism matter for practical protocol design. First, the effect scales with how imperfect the RF pulse's profile is: a simple, unapodised sinc pulse has relatively wide, slowly-decaying skirts and needs a comparatively large gap to suppress cross-talk, whereas modern optimised pulse designs (apodised sinc, SLR-designed pulses) have much sharper, more rectangular profiles and need a much smaller gap for the same degree of suppression — figures reported in the technical/patent literature describe gaps of up to roughly 50% of slice thickness for simple sinc pulses versus as little as approximately 10% for optimised pulse designs to achieve comparable cross-talk suppression [4]. Second, the effect is worse for 180° pulses than for 90° pulses: inversion-recovery preparation (STIR, FLAIR) and the refocusing pulses of turbo/fast spin echo sequences both use 180° pulses, whose imperfect profiles produce more pronounced cross-talk than a simple 90° gradient-echo excitation [1].

2.2 Mathematical and Coverage Foundations

The coverage equation shared with the companion Slice Thickness and Number of Slices pages isolates the gap term (g) explicitly:

C = N × (Δz + g)

Rearranged to express the practical consequence of gap directly, for a fixed number of slices and thickness, the fraction of the nominally-covered extent that is genuinely unsampled tissue is:

Missed-tissue fraction = g / (Δz + g)

Worked example: Δz = 4 mm, g = 1 mm (25% gap) → missed-tissue fraction = 1/5 = 20%. One-fifth of the nominal coverage extent is never excited by any slice — a structure smaller than roughly 1 mm and positioned unluckily within a gap could, in principle, be entirely missed. This framing is the single most important quantitative idea on this page: gap is not a "free" spacing choice — it is a direct, quantifiable sacrifice of coverage, distinct from the time/SAR trade-offs that dominate the companion Number of Slices page.

Why interleaving changes the equation: interleaved acquisition (exciting all odd-numbered slices first, then all even-numbered slices, within the same package) inserts a much longer effective time interval between the excitation of any two physically adjacent slices — long enough for substantial T1 recovery of the partially-excited transition-zone tissue — without requiring any physical gap at all. Interleaving is therefore a way to suppress cross-talk without paying the missed-tissue-fraction cost described above, and is the reason most modern 2D clinical protocols default to zero or near-zero gap [1].

3. Units, Terminology and Vendor Nomenclature

The Siemens Distance Factor conversion trap: because Siemens expresses gap as a percentage of slice thickness rather than an absolute value, a Distance Factor of 20% means a genuinely different absolute gap at 3 mm slices (0.6 mm) than at 6 mm slices (1.2 mm). This is a frequent, entirely avoidable source of error when a protocol is translated from a Siemens system to a vendor that expects an absolute millimetre value (or vice versa) — the numeric value must never be copied directly across the percentage/absolute boundary without conversion.

Vendor terminology confirmed from an actual Philips console (Advanced Parameters, Geometry tab): on Philips systems, gap is set within the same Stack block used for slice count and orientation — specifically the slice gap field, with mode set to "user defined" and the value entered directly as gap (mm) (e.g. 0.4 mm), alongside slice orientation and slice scan order in the same panel. This confirms Philips uses the absolute-millimetre convention, not a percentage.

3.1 Gap Is Not the Same Concept as 3D "Slice Oversampling"

This is a common and consequential point of confusion. In a 3D acquisition, there is no RF-selective slice excitation and therefore no cross-talk mechanism at all (Section 2.1) — the "slice oversampling" percentage available on 3D sequences serves an entirely different purpose: adding extra encoded partitions beyond the nominal slab boundary to prevent wrap-around (aliasing) artefact at the slab edges. It has nothing to do with cross-talk suppression and should never be reasoned about using the logic developed on this page for 2D gap.

4. Typical Value Ranges

5. Parameter Interaction Ecosystem

5.1 Parameter Relationships Matrix

6. Effects on Image Appearance

6.1 Increasing Gap

Cross-talk-related signal and contrast disturbance decreases or disappears, at the direct cost of the missed-tissue fraction quantified in Section 2.2. On the images themselves, there is no visible marker that tissue has been skipped between slices — this is a "silent" trade-off, only detectable by deliberately reasoning about the gap value and coverage, not by inspecting the acquired images.

6.2 Decreasing Gap Toward Zero or Negative

Coverage becomes continuous or redundant (overlapping), eliminating any risk of a structure falling entirely between slices. If interleaving is not also active, cross-talk risk rises correspondingly (Section 2.1) — visible as subtle signal/contrast inconsistency at slice boundaries, most conspicuous on post-contrast T1 sequences where the affected tissue's apparent enhancement can be altered.

7. Effects on Acquisition Time

Gap itself has no direct time cost — it does not appear in the T_acq equations governing 2D or 3D scan time (see the companion Number of Slices page). Its only time consequence is indirect: for a fixed total coverage requirement, increasing gap increases the number of slices needed to still reach that coverage, which can in turn push the protocol toward or past the TR-slice-loop-time ceiling described in full on the Number of Slices page — at which point the familiar step-function scan-time penalty of concatenations/packages applies. Gap should therefore be reasoned about as "free in isolation, but potentially expensive through its effect on required N."

8. Effects on SNR and CNR

Gap has no direct effect on the SNR of any individual acquired slice — SNR is governed by voxel volume, bandwidth, field strength, coil and NSA, none of which the gap value itself alters. The relationship to CNR is indirect but clinically significant: uncorrected cross-talk (too little gap, without interleaving, on a cross-talk-prone sequence) can measurably alter the apparent contrast of tissue near slice boundaries — most importantly, altering the apparent degree of contrast enhancement on post-gadolinium T1 sequences — which is a CNR/contrast-fidelity problem rather than a voxel-level SNR problem.

9. Artefacts Associated with Slice Gap

10. Behaviour Across Sequence Families

Spin Echo (SE) / Turbo Spin Echo (TSE): the repeated 180° refocusing pulses of the echo train compound RF-profile imperfection more than a single 90° excitation, making TSE comparatively more cross-talk-prone at a given gap than simple gradient-echo sequences [1]; interleaving is close to mandatory practice for contiguous 2D TSE.

Gradient Echo (GRE/FLASH, 2D): a single, typically lower-flip-angle 90°-or-less excitation pulse per TR generally produces a sharper effective profile and is comparatively more tolerant of small gaps, though the very short TR of many 2D GRE protocols (see the companion Number of Slices page, Section 10) already constrains achievable slice count independently of the gap question.

Inversion Recovery (STIR, FLAIR): the additional 180° inversion pulse compounds the cross-talk risk described for TSE above, and is the sequence family in which unintended cross-talk is most likely to be clinically consequential (Section 9) — the companion Number of Slices page's Example 3 develops the related TR-ceiling consequence of the same inversion pulse.

EPI (DWI, fMRI, DSC): single-shot EPI's very brief per-slice excitation-and-readout reduces within-acquisition repeated-pulse profile degradation, but standard multi-slice EPI series still follow the same gap/interleave logic as any other 2D multi-slice acquisition for adjacent, sequentially-excited slices.

Simultaneous multi-slice (SMS/multiband): as noted in Section 5.1, the traditional adjacent-slice gap concept does not directly govern SMS slice-group spacing, which follows its own separate design logic (slice-group separation, "PINS"/"MultiPINS"-type RF pulse design) intended to control the distinct inter-slice-leakage artefact rather than classical cross-talk [5].

3D sequences generally: no gap concept applies at all (Section 3.1); the relevant analogous parameter is 3D slice oversampling, governed by an entirely different physical concern (wrap prevention, not cross-talk).

11. Field Strength Behaviour

The gap requirement for cross-talk suppression is primarily a function of RF pulse design (Section 2.1) rather than field strength per se. At 3T and above, B1+ inhomogeneity and increased chemical-shift-related effects can, in some circumstances, marginally degrade the effective sharpness of a given RF pulse's slice profile relative to its performance at 1.5T, which can modestly increase the practical gap or interleaving rigour needed for equivalent cross-talk control — but this is a secondary, pulse-design-mediated effect rather than a direct field-strength relationship in the way SAR (Number of Slices page) or chemical shift (bandwidth-related pages) are directly field-strength-coupled.

12. Vendor-Specific Implementation

Siemens: Distance Factor, expressed as a percentage of slice thickness (Section 3); default values for most brain and spine protocols are typically in the 10–20% range, with interleaving applied automatically by default for standard multi-slice 2D sequences — the operator should still verify interleave status is displayed as active, since it can be manually disabled.

GE: absolute millimetre "Slice gap" field, accepting negative values for overlap; brain protocols commonly default to 0 mm gap relying on interleaved acquisition, while body protocols more often use a small positive gap (1–2 mm).

Philips: absolute millimetre "gap (mm)" field within the Stack block (confirmed directly from console review, Section 3), set to "user defined" mode when a specific non-default value is required; slice scan order (e.g. interleaved, or explicit head–foot/foot–head) is set in the same Geometry tab section.

Canon: absolute millimetre gap value in the protocol card UI, with vendor documentation in some contexts expressing the equivalent value as a percentage — a source of the same conversion-trap risk described for Siemens in Section 3.

Cross-vendor protocol translation: whenever a protocol is transferred between platforms, the gap value must be explicitly recalculated (percentage-to-absolute or vice versa) rather than copied as a raw number — this is one of the single most common, entirely avoidable sources of a protocol behaving differently than intended after a cross-vendor translation.

13. Practical Optimisation Strategies

13.1 Clinical Optimisation Recipes

Prefer interleaving over gap as the primary cross-talk defence: since interleaving suppresses cross-talk without sacrificing any coverage, it should be the default first-line measure; gap should be reserved for situations where interleaving is unavailable, insufficient on its own for a particularly cross-talk-prone sequence (Section 10), or where local practice/vendor default combines a small gap with interleaving as belt-and-braces redundancy.

Set gap according to what the clinical question can tolerate missing, not out of habit: a whole-brain MS surveillance protocol, where a single missed periventricular lesion is a genuine clinical failure, warrants gap close to zero even at some residual cross-talk risk (mitigated by interleaving); a rapid triage screening protocol, where speed matters more than millimetre-perfect small-lesion detection, can accept a larger, deliberately chosen gap — but this should be a conscious clinical decision, not a default left unexamined.

Consider negative gap (overlap) for small, delicate, diagnostically critical structures: internal auditory canal/vestibular schwannoma screening, optic nerve, and pituitary stalk protocols are classic candidates, where the cost of redundant sampling is trivial compared to the cost of a missed small lesion.

Always verify the effective absolute gap after any slice thickness change on a percentage-based (Siemens-style) system: because Distance Factor scales with slice thickness, changing thickness silently changes the absolute gap even if the percentage field is untouched.

14. Parameter Extremes

14.1 Zero or Negative Gap (Contiguous or Overlapping)

The modern default for most diagnostic-standard 2D protocols where interleaving is available and active; negative gap (deliberate overlap) is reserved for small, diagnostically critical structures (Section 13.1) where the redundant acquisition time and data are an acceptable cost for eliminating any possibility of a lesion falling entirely between slices.

14.2 Large Gap (Greater Than ~50%)

Occasionally used in deliberately fast, coverage-sacrificing screening or triage contexts (e.g. some rapid whole-spine or whole-body survey protocols) where the explicit clinical goal is rapid detection of large or obvious abnormality rather than exhaustive small-lesion detection. This use should always be an explicit, documented protocol design decision — not an unexamined default — given the direct, quantifiable coverage sacrifice established in Section 2.2; it is not appropriate for any protocol whose clinical purpose includes reliable small-lesion detection.

15. Common Optimisation Errors

16. MRI Technologist Pearls

Interleave first, gap second: treat interleaved acquisition order as the primary, default cross-talk defence, and gap as a supplementary or fallback measure — not the other way around.

If a referring clinician reports an apparent lesion "discontinuity" or an unexplained enhancement pattern near a slice boundary on a 2D post-contrast series, check gap and interleave settings first before assuming a genuine pathological explanation.

For any protocol explicitly built around small-structure detection (IAC, optic nerve, pituitary stalk, small cartilage defects), actively consider zero or negative gap rather than accepting a default positive gap value inherited from a general-purpose protocol template.

When translating a protocol from a percentage-based system to an absolute-mm system, always show the calculated absolute value to a colleague or supervisor for a sanity check before first clinical use — this single verification step prevents the most common vendor-translation gap error (Section 15).

17. Real Clinical Examples

Example 1: MS Surveillance Brain — Near-Zero Gap Is the Correct Choice

Clinical scenario: longitudinal multiple sclerosis surveillance MRI; the clinical question depends on reliably detecting small new periventricular or juxtacortical lesions, some only a few millimetres across.

Protocol logic: axial FLAIR/T2, 3–4 mm slices, gap ≤ 0.3 mm (≈ 5–10%), fully interleaved. The near-zero gap directly minimises the missed-tissue fraction (Section 2.2); interleaving handles the residual cross-talk risk without requiring a larger gap.

Lesson: for any surveillance protocol whose clinical value depends on detecting small new findings, gap should be actively minimised, with interleaving carrying the cross-talk-suppression burden instead.

Example 2: Internal Auditory Canal Screening — Negative Gap (Overlap)

Clinical scenario: high-resolution 3D or thin 2D T2 screening of the internal auditory canals for a small intracanalicular vestibular schwannoma.

Protocol logic: where a 2D approach is used, a small negative gap (slice overlap) is deliberately set — accepting redundant, overlapping sampling of the same millimetre-scale tissue twice — specifically to eliminate any possibility that a very small tumour is positioned entirely within an unsampled interval.

Lesson: for the smallest, most diagnostically unforgiving target structures on this platform, the missed-tissue-fraction logic of Section 2.2 justifies deliberately paying a small redundancy cost (overlap) rather than accepting any positive gap at all.

Example 3: Rapid Whole-Spine Screening — a Deliberate Large-Gap Trade-off

Clinical scenario: a fast triage whole-spine sagittal STIR sequence used to screen rapidly for metastatic disease or cord compression in an oncology emergency context, where speed to a management decision is paramount and the target abnormalities (cord compression, large metastatic deposits) are not typically millimetre-scale.

Protocol logic: a larger gap than would be acceptable for MS surveillance (Example 1) is deliberately accepted, explicitly traded for either faster acquisition or wider single-pass coverage, on the clinical judgement that the target pathology in this specific context is unlikely to be missed purely by falling within a gap.

Lesson: the "correct" gap is not a universal constant — it is a function of how small the smallest clinically important finding is expected to be, and this judgement should be made explicitly for each protocol's specific clinical purpose, not inherited unexamined from a general template (Section 14.2).

Example 4: Post-Contrast Spine T1 — Cross-Talk Misread as Altered Enhancement

Clinical scenario: post-gadolinium T1 TSE sagittal spine, zero gap, sequential (non-interleaved) acquisition order inadvertently left active after a protocol modification.

Problem: adjacent vertebral body marrow and paraspinal soft tissue show a subtle, alternating signal pattern between slices that does not correspond to a genuine pathological process — the classic cross-talk signature, here affecting interpretation of a borderline marrow enhancement pattern.

Correction: interleaved acquisition order re-activated; the alternating pattern resolves, and the marrow enhancement pattern can be assessed reliably.

Lesson: this is the spine-specific analogue of the classic brain cross-talk case; the underlying mechanism (Section 2.1) and correction (Section 2.2) are identical regardless of anatomical region — gap and interleave settings should be part of the standard troubleshooting checklist whenever an unexplained, boundary-associated signal pattern is seen on any 2D post-contrast series.

Example 5: Multi-Station Whole-Body — Negative "Gap" at the Station Seam

Clinical scenario: whole-body STIR/DWI for suspected multiple myeloma or metastatic screening, acquired across 5–6 table stations.

Protocol logic: a deliberate small overlap (station-level negative gap) is planned between the slice coverage of adjacent stations, ensuring no anatomical seam falls entirely unsampled at the table-position boundary — conceptually the same missed-tissue-fraction logic as within-station gap (Section 2.2), applied at the station-to-station level rather than the slice-to-slice level.

Lesson: the gap concept developed on this page for individual slices generalises directly to station planning in multi-station whole-body protocols, and deserves the same deliberate attention.

18. Visual Educational Material

18.1 Cross-Talk vs. Coverage Decision Flow

IS INTERLEAVED ACQUISITION AVAILABLE AND ACTIVE?
│
├── YES → cross-talk is largely handled → gap can be minimal (0-10%)
│         │
│         └── Does the clinical question require detecting very small
│             (near-slice-thickness-scale) structures?
│             ├── YES → consider zero or slightly negative gap (overlap)
│             └── NO  → small positive gap (5-15%) is a reasonable default

└── NO  → cross-talk suppression depends on gap alone
          │
          └── What RF pulse type does the sequence use?
              ├── 90° GRE-type → smaller gap may suffice (~10-20%)
              └── 180° (TSE refocusing, STIR, FLAIR) → larger gap
                  typically needed (~30-50%) for equivalent suppression [4]

18.2 Missed-Tissue Fraction vs. Gap Percentage (Fixed Slice Thickness)

Gap as % of slice thickness:   0%    10%   20%   30%   50%
Missed-tissue fraction:        0%    9%    17%   23%   33%

At 50% gap, a full THIRD of the nominal coverage extent
is never sampled by any slice — a genuinely large sacrifice,
appropriate only for deliberate screening-speed trade-offs
(Section 14.2), never as an unexamined default.

18.3 Percentage-to-Absolute Conversion — Worked Examples (Siemens Distance Factor)

Distance Factor 20% at:
  3 mm slices  → gap = 0.6 mm
  5 mm slices  → gap = 1.0 mm
  8 mm slices  → gap = 1.6 mm

SAME percentage, DIFFERENT absolute gap — always recalculate
when slice thickness changes or when translating to an
absolute-mm vendor system.

19. Evidence Gaps and Ongoing Debate

Optimal gap percentage by RF pulse design and sequence family has not been comprehensively, prospectively validated: the widely cited practical ranges (approximately 10% for optimised pulse designs, up to approximately 50% for simple sinc pulses) originate substantially from technical/engineering and patent literature rather than from a systematic, prospective, cross-vendor clinical validation study [4]; departmental defaults are therefore reasonable, widely used engineering-derived values rather than formally validated clinical thresholds.

Quantified clinical risk of small-lesion detection failure specifically attributable to gap has not been systematically studied: while the missed-tissue-fraction logic in Section 2.2 is a straightforward geometric certainty, the real-world frequency with which a clinically important small lesion is positioned unfavourably enough to be entirely missed, across different anatomical regions and clinical contexts, has not been the subject of a dedicated prospective outcome study.

Interaction between SMS slice-group spacing and traditional inter-slice gap concepts remains incompletely unified in the literature: as noted in Sections 5.1 and 10, SMS acquisitions follow a distinct slice-group-spacing design logic; a unified framework relating classical gap/cross-talk theory to modern SMS-based multi-slice acquisition has not been comprehensively established.

20. Miscellaneous and Future Directions

Progressive reduction in gap requirements from improved RF pulse design: as SLR-designed, VERSE, and other optimised RF pulse shapes become more universally deployed across all major vendor platforms, the practical gap required for equivalent cross-talk suppression continues to trend downward, gradually reducing the clinical relevance of this parameter for platforms with the most modern pulse designs — though the underlying physical trade-off (Section 2) remains conceptually unchanged.

Continued shift toward 3D acquisition for coverage-critical, small-lesion-detection protocols: because 3D acquisition eliminates the gap concept entirely (Section 3.1), the long-term clinical trend for applications where missed small-lesion risk is paramount (e.g. IAC screening, cartilage imaging) continues to favour 3D isotropic acquisition over gapped 2D multi-slice imaging wherever scan-time and motion constraints allow.

Deep-learning slice-gap interpolation: research-stage approaches to synthetically interpolating plausible tissue signal within a gap, from adjacent acquired slices, are an active area of technical development; such interpolated tissue is a model-generated estimate rather than genuinely acquired data, and its clinical reliability for actual small-lesion detection (as opposed to visual continuity in a reformat) has not been established for diagnostic use.

End of document — Slice Gap — MRIninja v1.0 — August 2026

Parent page: MRI Parameters — Overview and Classification (9501)