FOV — Field of View

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

1. Introduction and General Purpose

The Field of View (FOV) is one of the most fundamental acquisition parameters in MRI. It defines the physical extent of the volume from which MRI signal is spatially encoded and mapped into the image. In practical terms, it is the size of the rectangle (in 2D acquisitions) or cuboid (in 3D acquisitions) that constitutes the displayed image: the number of millimetres covered from edge to edge in each spatial direction.

FOV is not merely an aesthetic framing choice. Its value determines — in direct mathematical proportion — the spatial resolution of each voxel, the degree of aliasing artefact, the signal-to-noise ratio, the acquisition time, and the fat–water chemical shift displacement. Because it interacts with virtually every other acquisition parameter simultaneously, FOV is one of the parameters most frequently misoptimised in routine clinical protocols: set too large, it wastes spatial resolution; set too small, it introduces aliasing; set anisotropically without appropriate phase oversampling, it introduces fold-over artefacts that may obscure pathology.

Understanding FOV at the level required for protocol optimisation means understanding not just what it is, but how it is encoded into k-space, how it interacts with the matrix and slice thickness to determine voxel dimensions, and how its relationship to the anatomy being imaged must be balanced against SNR and time constraints.

Historical evolution: the concept of FOV in MRI emerged directly from the Fourier spatial encoding framework established by Lauterbur [1] and Mansfield [2]. In the earliest clinical scanners (early 1980s), FOV was typically 400–500 mm — the maximum coverage achievable with the gradient linearity of first-generation systems. As gradient technology improved (higher linearity over larger volumes, higher slew rates), smaller FOV became achievable without geometric distortion, enabling high-resolution imaging of small structures. The introduction of surface coils in the mid-1980s allowed small FOV imaging (wrist, knee, temporomandibular joint) without SNR penalty, because the coil's sensitivity naturally limits the spatial extent of signal collection to the region of interest. Today, FOV values range from < 50 mm (targeted finger or inner ear imaging with a small surface coil) to > 500 mm (whole-body DWI coronal acquisitions).

2. Physical Foundations

FOV is defined by the relationship between the encoding gradients and the Fourier sampling strategy. In each spatial dimension, the MRI signal is encoded by applying a gradient field that imposes a linear variation of precession frequency (or phase) across the sample. The spatial resolution and FOV in each direction are determined by how finely and over what range this gradient encodes position.

2.1 Mathematical Foundations

Frequency-encoding direction (readout direction)

The readout gradient G_r creates a linear frequency variation across the FOV in the readout direction. The bandwidth of the receiver (total BW, in Hz) and the gradient strength determine the FOV in the readout direction:

FOV_r = BW_total / (γ × G_r)

where:

• BW_total = total receiver bandwidth in Hz
• γ = gyromagnetic ratio for proton (42.577 MHz/T)
• G_r = readout gradient amplitude in T/m

Clinical meaning: at fixed bandwidth, a stronger readout gradient produces a smaller FOV (higher spatial resolution per pixel). At fixed gradient strength, wider bandwidth gives larger FOV. In practice, BW is set first (for artefact control), and the gradient strength is adjusted automatically by the scanner to achieve the prescribed FOV.

In-plane spatial resolution (readout):

Δx = FOV_r / N_x

where N_x is the number of frequency-encoding points (the readout matrix).

Optimisation implication: increasing N_x at fixed FOV_r decreases voxel size → higher resolution, but requires either wider bandwidth (→ SNR cost) or longer readout time (→ longer TE_min → more T2* weighting).

Phase-encoding direction

In the phase-encoding direction, the FOV is determined by the spacing between consecutive phase-encoding gradient steps (Δk_y). The relationship between k-space sampling and FOV follows directly from the Fourier transform:

FOV_p = 1 / Δk_y

where Δk_y is the k-space step size in the phase-encoding direction.

Physical meaning: Δk_y = γ × G_p × Δt, where G_p is the phase-encoding gradient amplitude and Δt is the gradient duration. Increasing the step between phase-encoding gradient increments (coarser k-space sampling) → smaller FOV_p. Decreasing the step (finer sampling) → larger FOV_p.

In-plane spatial resolution (phase):

Δy = FOV_p / N_y

where N_y is the number of phase-encoding steps.

Critical implication: FOV_p and resolution Δy are independently controllable by adjusting both N_y and Δk_y. If the anatomy fits within a small FOV in the phase direction, N_y can be reduced proportionally (keeping Δy constant) → acquisition time reduction proportional to the FOV reduction. This is the mathematical basis for rectangular FOV (rFOV).

Through-plane (slice) direction (2D)

In 2D acquisitions, the slice direction is not Fourier-encoded (it is selected by the RF pulse bandwidth and the slice-selection gradient). The FOV in the slice direction is therefore simply: the number of slices × (slice thickness + gap). There is no k-space encoding in this dimension.

Through-plane (3D acquisitions)

In 3D acquisitions, the slice direction is phase-encoded (a second set of phase-encoding gradients encodes the z-dimension). The FOV in the slab direction is:

FOV_z = 1 / Δk_z

analogously to the in-plane phase encoding. The 3D voxel size:

Δz = FOV_z / N_z

This is how true isotropic voxels (Δx = Δy = Δz) are achieved in 3D acquisitions — by matching the three FOV dimensions divided by their respective matrix values.

Voxel volume

Voxel_volume = Δx × Δy × Δz = (FOV_x / N_x) × (FOV_y / N_y) × Slice_thickness

This equation encapsulates the fundamental SNR-resolution trade-off:

SNR ∝ Voxel_volume × √(N_acquisitions) / √(BW_per_pixel)

Every reduction in FOV that is not accompanied by a compensating reduction in N_x or N_y reduces the voxel volume → reduces SNR. This is why small FOV imaging requires either higher field strength, surface coils, increased NSA, or reduced BW to maintain diagnostic SNR.

Nyquist criterion and aliasing

The FOV sets the Nyquist limit: any anatomy located outside the FOV in the phase-encoding direction is spatially aliased (folds back into the image). The Nyquist condition requires that the sampling frequency (1/Δk_y) be at least twice the spatial frequency of the highest-frequency component in the object:

FOV_p ≥ Extent_of_anatomy_in_phase_direction

Violation of this condition produces fold-over (wraparound) artefact — the most direct clinical consequence of FOV selection error.

3. Units, Terminology and Vendor Nomenclature

FOV is expressed in millimetres (mm) universally across all vendors and platforms. It is typically specified as two values — one for the frequency-encoding direction and one for the phase-encoding direction — unless the FOV is square (FOV_x = FOV_y).

Common abbreviations:

• FOV: Field of View
• rFOV: Rectangular FOV (asymmetric FOV with smaller phase FOV)
• iFOV: Isotropic FOV (3D acquisitions with equal voxel dimensions)
• FOV_r or FOV_f: FOV in the readout/frequency direction
• FOV_p: FOV in the phase-encoding direction
• FOV_z: FOV in the slice/partition direction (3D)

4. Typical Value Ranges

4.1 FOV by Anatomical Region

4.2 FOV by Field Strength

5. Parameter Interaction Ecosystem

FOV is coupled to virtually every other MRI parameter. The table below summarises the most clinically important interactions.

5.1 Parameter Relationships Matrix

6. Effects on Image Appearance

6.1 Increasing FOV

• Coverage: more anatomy is included in the image. May prevent aliasing. Critical for bilateral anatomical assessments.
• Spatial resolution (at fixed matrix): voxels become larger → reduced spatial resolution → structural detail is lost. The trade-off is most clinically significant for fine structures (cartilage, small nerves, plantar plate).
• Voxel volume (at fixed matrix): increases → SNR increases → image appears less noisy. For large patients where SNR is marginal, increasing FOV slightly while accepting some resolution reduction can improve diagnostic confidence.
• Chemical shift artefact (in mm at fixed matrix): increases proportionally. A large FOV with a coarse matrix produces large chemical shift displacement artefacts at fat-water interfaces.
• Partial volume effects: larger voxels → more partial volume → apparent signal averaging between adjacent structures. Particularly problematic for thin structures (cartilage: 1.5–2 mm; cortical bone; nerve roots).

6.2 Decreasing FOV

• Spatial resolution (at fixed matrix): voxels become smaller → higher spatial resolution. This is the primary reason for selecting a small FOV.
• SNR (at fixed matrix): decreases proportionally to voxel volume reduction. At very small FOV, SNR may become diagnostic-limiting without surface coils or increased NSA.
• Aliasing risk: if the anatomy in the phase-encoding direction extends beyond the reduced FOV_p, fold-over artefact appears. Phase oversampling must be applied.
• Chemical shift artefact (in mm): decreases as voxels shrink, which is beneficial for fine-structure protocols.
• Coil coverage constraint: below approximately 80–100 mm FOV, the receive coil's spatial sensitivity profile may be the limiting factor rather than the gradient encoding.

7. Effects on Acquisition Time

FOV has no direct effect on acquisition time when considered in isolation. Acquisition time in standard 2D TSE is:

T_acq = TR × N_y × NSA / ETL

FOV does not appear in this equation directly. However, FOV affects acquisition time indirectly through the matrix requirements it imposes:

If the target spatial resolution (Δy = FOV_p / N_y) is fixed, then:

• Increasing FOV_p requires increasing N_y proportionally → more phase-encoding steps → longer acquisition time.
• Decreasing FOV_p (rFOV) with proportionally reduced N_y → fewer phase-encoding steps → shorter acquisition time.

This is the operational basis for rectangular FOV (rFOV): if the anatomy in the phase-encoding direction is smaller than the anatomy in the frequency direction (e.g., sagittal spine: AP dimension ≈ 120–150 mm vs SI dimension ≈ 250–300 mm), FOV_p can be reduced to match the anatomy → N_y is reduced proportionally → acquisition time decreases by the rFOV factor (e.g., FOV_p = 60% of FOV_r → 40% time saving).

Example: T2 TSE knee, axial:

• FOV 160 × 160 mm, matrix 320 × 320, TR=3000 ms, ETL=12, NSA=1
• T_acq = 3000 × 320 / 12 = 80,000 ms = 1.33 min (square FOV)
• With rFOV 160 × 100 mm (62.5% phase FOV), matrix 320 × 200:
• T_acq = 3000 × 200 / 12 = 50,000 ms = 0.83 min (37.5% time saving)

FOV and 3D acquisition time: in 3D acquisitions, FOV_z / partition_thickness = N_partitions. Increasing 3D slab coverage (larger FOV_z) requires more partitions → directly proportional increase in acquisition time (for fixed TR, N_y, ETL).

8. Effects on SNR and CNR

8.1 SNR and FOV

The fundamental relationship:

SNR ∝ FOV_x × FOV_y × Slice_thickness / (N_x × N_y) × √(NSA × BW_correction)

Simplifying to voxel volume: SNR ∝ Δx × Δy × Δz

This means:

• Halving the FOV in both phase and frequency directions (at fixed matrix) reduces voxel area by 4× → SNR reduces by 4×.
• Halving FOV in only the phase direction (rFOV, at fixed resolution) reduces N_y by half → scan time halves, SNR is maintained per voxel (voxel size unchanged).

The critical distinction: FOV reduction at fixed matrix → resolution improves → SNR decreases. FOV reduction at reduced matrix (rFOV) → resolution unchanged → scan time decreases → SNR per voxel unchanged. These two scenarios are completely different clinically.

8.2 CNR and FOV

CNR (Contrast-to-Noise Ratio) depends on both the tissue contrast (determined by TR, TE, TI, flip angle) and the SNR. FOV affects CNR only through its effect on SNR. At fixed SNR, FOV does not directly change tissue contrast.

However, partial volume averaging is an indirect CNR effect of FOV: large voxels (from large FOV at fixed matrix) average signal from adjacent structures, reducing the apparent contrast difference between a lesion and its surroundings. For a 2 mm cartilage layer imaged at 1 mm voxel size: good contrast. For the same cartilage at 3 mm voxel: substantial partial volume averaging reduces the apparent cartilage signal and contrast with the adjacent bone.

8.3 Field-Strength Dependency

At higher field strength, the intrinsic SNR advantage allows more flexibility with FOV:

• 3T vs 1.5T: the approximately 1.7–2× SNR gain at 3T can be "spent" on smaller FOV (higher resolution at maintained SNR), or on reduced NSA (faster acquisition), or both.
• 0.55T: lower SNR at 0.55T significantly constrains small FOV use; resolutions achievable at 1.5T with small FOV require substantially increased NSA or reduced matrix at 0.55T.
• 7T: extremely high SNR allows very small FOV (≤ 60 mm) for targeted high-resolution neuroimaging (hippocampus, inner ear, cortical layers) — but B1+ inhomogeneity severely limits the useful FOV beyond approximately 200–250 mm in the head.

9. Artefacts Associated with FOV

10. Behaviour Across Sequence Families

Spin Echo (SE) and Turbo Spin Echo (TSE)

FOV directly determines voxel dimensions and, with the matrix, the acquisition time. rFOV is routinely and safely applied in SE/TSE. The only aliasing risk in SE/TSE is in the phase direction; frequency-direction aliasing is prevented automatically by the readout bandwidth (which extends beyond the FOV in the frequency direction by design, with frequency oversampling). Phase oversampling adds zero-cost protection in the frequency oversampling axis; phase oversampling adds time proportional to the oversampling percentage.

Gradient Echo (GRE/FLASH/SPGR)

Identical principles to SE for FOV-resolution-time relationships. GRE sequences used for DCE (VIBE/LAVA) require a bilateral FOV appropriate for the organ of interest (liver/kidney/breast) within the breath-hold duration constraint — small matrix × large FOV × small rFOV is the typical compromise.

Inversion Recovery (STIR, FLAIR, MPRAGE)

No FOV-specific modification. In MPRAGE/BRAVO (3D), the sagittal acquisition FOV_x × FOV_y × FOV_z must encompass the full brain — typically 256 × 256 × 192 mm for ADNI protocol. Any asymmetry (smaller FOV_z = shorter coverage at the vertex or cerebellum) must be verified on the localiser.

EPI (DWI, DSC)

FOV in the phase direction is the most critical EPI-specific parameter because the EPI readout bandwidth in the phase direction is inversely proportional to the number of phase-encoding steps and the echo spacing. The EPI phase-direction bandwidth per pixel (PE-BW) is approximately:

PE-BW = 1 / (N_phase × ESP)

where ESP is the echo spacing. For a 96-step EPI with 0.8 ms ESP: PE-BW ≈ 13 Hz/pixel. At 3T (fat-water offset 440 Hz), geometric distortion ≈ 440/13 = 34 pixels — catastrophic without correction. Reducing FOV_p (from 96 to 48 steps) doubles PE-BW → halves distortion.

Reduced-FOV EPI techniques (ZOOMit/iZOOM/FOCUS/REVEAL): apply a 2D spatially selective excitation that limits the FOV_p to the target anatomy (e.g., prostate, spinal cord, breast lesion), dramatically reducing phase-encoding steps and geometric distortion while maintaining in-plane resolution.

Dixon (2-point, 3-point, mDixon)

FOV requirements are identical to the underlying GRE sequence. The TE-dependent fat-water phase cycle imposes specific minimum TE constraints (see parent page 9501) but does not independently constrain FOV selection.

bSSFP (TrueFISP/FIESTA/b-FFE)

bSSFP is sensitive to off-resonance banding artefacts that scale with FOV at a given B0 homogeneity level. In cardiac bSSFP, the standard FOV (280–380 mm for the heart) must be centred precisely to keep the banding patterns outside the myocardium. The banding period (in mm) scales inversely with the off-resonance frequency — not directly with FOV — but positioning the FOV to keep the heart far from the banding node is a standard optimisation.

ASL (pCASL)

ASL requires the labelling plane to be placed upstream of the imaging FOV (at the neck, inferior to the imaging slab). The imaging FOV must be large enough to cover the full brain (typically 220–260 mm axial × 50–100 mm slab thickness for 3D pCASL). Insufficient FOV_z in ASL causes incomplete brain coverage → missing perfusion data in superior brain regions.

Spectroscopy (SVS / MRSI)

In single-voxel spectroscopy (SVS), the voxel itself is defined by three perpendicular slice-selection gradients — it is the spectroscopy equivalent of a FOV, but typically 2–8 cm per dimension. In MRSI (multi-voxel chemical shift imaging), the phase-encoded FOV determines the number and size of spectral voxels. The MRSI FOV must be placed to avoid lipid contamination from the skull base — a critical spectroscopy-specific FOV positioning requirement.

11. Field Strength Behaviour

At 3T: the higher SNR can be used to justify smaller FOV (higher resolution) at the cost of increased chemical shift and EPI distortion, both of which require compensatory parameter adjustments (wider BW, reduced FOV_p in EPI, Dixon instead of SPAIR).

At 0.55T: the low SNR severely constrains small FOV use. Most 0.55T applications require FOV ≥ 200 mm to maintain diagnostic SNR, with parallel imaging and DLR as the primary tools for partial SNR recovery.

12. Vendor-Specific Implementation

Siemens

Phase FOV is expressed as a percentage of the frequency FOV (e.g., "Phase FOV 75%"). Phase oversampling is set separately in percentage and adds lines beyond the prescribed FOV_p without aliasing protection cost to acquisition time in the oversampled zone. The Siemens "No Phase Wrap" option automatically applies 100% phase oversampling — eliminates all aliasing but doubles phase-encoding steps → increases scan time unless ETL or parallel imaging compensates. In the body matrix (body coil or multi-element surface coils), the scanner automatically warns if the anatomy exceeds the FOV_p by checking the body landmark.

GE

Phase FOV is expressed as a percentage (same convention as Siemens). "No phase wrap" is the GE equivalent of full phase oversampling. GE uses the term "Zip" for zero-filling (interpolation) in the frequency direction — Zip2 doubles the apparent frequency matrix for display; this is zero-filling, not true resolution increase. Technologists must distinguish the acquired matrix from the Zip-interpolated display matrix when reporting spatial resolution.

Philips

Phase FOV is controlled by "Fold-over suppression" (a Boolean that activates phase oversampling) and the "Phase FOV" percentage parameter. Philips uses the term "overcontiguous" for slice coverage beyond the prescribed slab in 3D acquisitions, which is related to 3D FOV_z management. The Philips system also reports "voxel size" directly on the protocol display, making it easier to verify the true voxel dimensions without manual calculation.

Canon

Canon uses standard FOV terminology consistent with the international conventions. Phase oversampling is termed "Phase oversampling" with a percentage value. Canon's automated protocol management (iDMS) suggests FOV values based on patient body landmark detection — the technologist should verify and override when protocol requirements differ from the automated suggestion.

United Imaging

United Imaging (UIH) follows Siemens-equivalent parameter naming for most parameters including FOV. Phase FOV is expressed as a percentage. UIH scanners include automated body landmark detection for FOV suggestion; their uAI platform provides AI-assisted slice planning that includes FOV optimisation proposals based on anatomical recognition from the localiser.

Hidden parameter coupling across vendors: on most platforms, when the FOV is reduced below a certain threshold for the selected coil configuration, the scanner automatically activates surface coil-based intensity correction (CLEAR/PURE/Prescan Normalize) to compensate for the non-uniform sensitivity profile. This changes the noise texture and apparent SNR distribution in the image in ways that may not be immediately obvious to the technologist.

13. Practical Optimisation Strategies

13.1 Clinical Optimisation Recipes

14. Parameter Extremes

14.1 Extremely Small FOV (< 80 mm)

At very small FOV, three effects converge:

• SNR becomes critically low without a surface coil whose sensitivity matches the FOV. A 60 mm FOV with a body coil produces nearly unusable images.
• Gradient demands are high: small FOV requires either high gradient amplitude or narrow bandwidth (increasing chemical shift). Modern gradient systems (80–100 mT/m) handle 60–80 mm FOV without difficulty; older systems (< 40 mT/m) were limited.
• Aliasing sensitivity is extreme: any anatomy outside the 60 mm zone folds back immediately. Even a small patient movement that shifts the anatomy laterally by 10–20 mm can introduce aliasing.

Clinical applications of extreme small FOV: inner ear (40–60 mm, with dedicated 8-channel inner ear coil); finger joints (50–70 mm); orbital MRI (70–90 mm); temporal mandibular joint (60–80 mm per joint); dental MRI (50–70 mm).

14.2 Extremely Large FOV (> 450 mm)

At very large FOV:

• Gradient non-linearity increases at the periphery → geometric distortion → anatomy at the edge is spatially inaccurate. All modern scanners apply gradient non-linearity correction (gradwarp or equivalent) in post-processing, but residual distortion may remain beyond 450–500 mm.
• B0 inhomogeneity increases over the larger volume → spectral fat saturation (SPAIR) becomes unreliable at the periphery → use STIR or Dixon.
• Phase-encoding aliasing from arm anatomy is the primary practical problem for WB-MRI coronal acquisitions (> 400 mm): the patient's arms, if alongside the body, lie outside the prescribed FOV_p and fold onto the spine/pelvis unless phase oversampling is applied. 100–150% phase oversampling is standard for WB-MRI coronal sequences.
• Coil sensitivity fall-off: beyond the sensitive volume of any coil combination, SNR approaches zero regardless of FOV. Setting FOV to 500 mm when the combined coil sensitivity extends to only 400 mm wastes the outer 100 mm.

15. Common Optimisation Errors

16. MRI Technologist Pearls

Verify anatomy fit before starting: always check the localiser to confirm that the anatomy of clinical interest is completely within the prescribed FOV in all three directions, before starting the first diagnostic sequence. A 30-second check avoids a 20-minute repeated acquisition.

Use the voxel size readout: all modern consoles display the resulting voxel dimensions (in mm) when FOV and matrix are set. Read this number — not the FOV and matrix separately — to confirm you are achieving the target resolution. This is the single most effective habit for avoiding resolution errors.

rFOV is almost always applicable to spine protocols: cervical, thoracic, and lumbar spine axial sequences in the A-P direction rarely exceed 140–160 mm. Applying rFOV (60–70% phase FOV) saves 30–40% of scan time with no penalty. This time saving can be used to add a second sequence or reduce patient time in the scanner.

Phase oversampling is generally free in the frequency direction: on all major platforms, frequency-direction oversampling (preventing fold-over in the readout direction) is applied automatically and adds no scan time. In the phase direction, phase oversampling adds a fraction of the scan time equal to the oversampling percentage / ETL. At ETL=12, 50% phase oversampling adds ≈ 4% scan time — negligible insurance against aliasing.

Protocol rescue for oversized patient: if a patient's anatomy exceeds the prescribed FOV_p and fold-over is visible on the first sequence: (a) increase phase oversampling to 100% as the first step — this eliminates the artefact at minimal time cost; (b) if this is insufficient (very large patient), increase FOV_p by 20% and increase N_y proportionally to maintain resolution.

EPI DWI in the pelvis and thorax: always plan pelvic and thoracic DWI with the smallest FOV_p that covers the organ of interest without including air-filled structures (rectum, colon, lung) in the phase-encoding axis. Including large gas-tissue interfaces in the phase direction increases EPI geometric distortion by 2–5 mm at 3T.

Check 3D slab coverage at both ends: for 3D acquisitions (MRCP, MPRAGE, 3D shoulder), verify that the superior AND inferior boundaries of the 3D slab include the target anatomy completely. The inferior pole of the liver, the superior pole of the kidney, the apex of the prostate, and the calcaneus in foot/ankle MRI are the most commonly clipped anatomy.

Document the FOV for serial comparison: when a protocol will be used for longitudinal monitoring (AD volumetry, tumour follow-up, NAC monitoring), document the exact FOV dimensions and matrix in the report or the protocol record. A subsequent examination with a different FOV produces apparent volume changes that are artefactual.

17. Real Clinical Examples

Example 1: Plantar Plate Assessment (Toes MRI)

Clinical scenario: a patient with metatarsalgia and suspected plantar plate tear at the second metatarsophalangeal joint. The plantar plate is 2 mm thick.

FOV selection: small surface coil (flex coil or dedicated foot coil); FOV = 80 × 70 mm (rFOV); matrix 320 × 280 at 3T.

Impact on resolution: voxel size 0.25 × 0.25 mm in-plane — sufficient to resolve plantar plate tears and grade them. At the standard foot protocol FOV of 160 × 130 mm, matrix 320 × 260 → voxel 0.5 × 0.5 mm — plantar plate tears at the attachment are at the resolution limit.

Impact on acquisition time: small FOV with dedicated coil; NSA increased to 3 for SNR compensation (SNR ∝ voxel volume, so 4× smaller voxel → 4× SNR loss; NSA=3 recovers √3 ≈ 1.7× → net SNR reduction accepted for resolution gain).

Impact on diagnosis: at 0.25 mm in-plane, the intact plantar plate fibres versus partial or complete tear at the metatarsal attachment can be graded (Coughlin/Nery system). This is the only clinical context where this FOV is justified; for general foot survey, 150–180 mm FOV is standard.

Trade-off: limited coverage — only one or two metatarsophalangeal joints visible. A separate larger-FOV acquisition is needed for bilateral overview.

Example 2: Bilateral Brachial Plexus MRI

Clinical scenario: traumatic brachial plexus injury with suspected root avulsion — bilateral assessment required for comparison.

FOV selection: 320 × 280 mm coronal (R-L × A-P); 3 mm slice thickness; matrix 448 × 392 at 3T. Phase direction R-L.

Coverage rationale: the bilateral plexus from C4 to the axillary level spans approximately 280–320 mm laterally and 20–25 cm craniocaudally. A FOV of 320 mm laterally just covers both clavicles + the scalene triangles bilaterally. Reducing to 260 mm risks clipping one plexus.

rFOV: not applied in the phase direction (R-L) because the arms alongside the body contribute anatomy that extends beyond any practical rFOV — 100% phase oversampling is applied instead to prevent arm fold-over.

Impact on acquisition: larger FOV with 100% phase oversampling increases effective phase steps → scan time increase mitigated by ETL 16 (STIR) and parallel imaging R=2.

Diagnostic implication: a technologist who reduces the FOV to 260 mm to "save time" without verifying that both clavicles are included risks missing the retroclavicular portion of the plexus on the larger side — a region critical for assessing middle and lower trunk continuity.

Example 3: Prostate MRI (PI-RADS)

Clinical scenario: elevated PSA, prior negative biopsy; systematic prostate MRI for lesion detection.

FOV selection: axial sequences FOV 200 × 180 mm (rFOV 90%); coronal sequences 200 × 220 mm; 3 mm slice thickness; matrix 320 × 288.

Phase direction: axial sequences: A-P (prevents rectal gas artefact from folding onto prostate in the lateral direction); rectal content may extend posteriorly but is outside the A-P phase direction fold-over risk zone.

DWI FOV: axial DWI FOV 220 × 180 mm; reduced-FOV DWI option (ZOOMit) at FOV 120 × 80 mm for the prostate specifically reduces EPI distortion — particularly important for the posterior peripheral zone adjacent to the rectal air-tissue interface at 3T.

Impact on diagnosis: at standard FOV 200 mm with 320 matrix → voxel 0.625 mm in-plane → adequate for PI-RADS lesion detection ≥ 0.5 cm. DWI at reduced FOV 120 mm with 192 matrix → voxel 0.625 mm in-plane, but dramatically reduced EPI distortion → correct anatomical localisation of posterior peripheral zone lesions.

Example 4: Liver DCE MRI in a Large Patient (BMI > 40)

Clinical scenario: suspected hepatocellular carcinoma surveillance; patient weight 135 kg, BMI 42. Standard protocol FOV 380 × 280 mm produces fold-over from lateral adipose tissue at rFOV.

FOV adjustment: increase FOV_p from 280 mm to 360 mm (square FOV for this patient). Increase N_y from 224 to 288 to maintain voxel size (Δy unchanged at 1.25 mm).

Impact on acquisition time: N_y increase from 224 to 288 → 29% time increase per breath-hold. At TR=3.5 ms, N_y=288, R=3 (parallel imaging): effective scan time = 3.5 × 288 / (3 × ETL) — within the patient's 18-second breath-hold capacity.

Impact on diagnosis: without FOV_p adjustment, the right lateral hepatic lobe (which is enlarged in this patient) would be clipped. Additionally, right-sided adipose tissue would fold over the right hepatic lobe on the rFOV acquisition, potentially mimicking or obscuring a hepatic mass.

Trade-off: slightly longer breath-hold required; compensated by instructing two preparatory deep breaths before the arterial phase acquisition.

18. Visual Educational Material

18.1 FOV, Matrix, and Voxel Relationship Diagram

FREQUENCY DIRECTION (N_x)
├──────────────────────────────────────────┤
│ FOV_x = 200 mm                           │  ↑
│ N_x    = 400 pixels                      │  │
│ Δx     = 0.5 mm/pixel                    │  │ FOV_y
│                                          │  │ = 160 mm
│   [target anatomy — fits within FOV]     │  │ N_y = 320
│                                          │  │ Δy = 0.5 mm
│                                          │  ↓
└──────────────────────────────────────────┘

VOXEL VOLUME = 0.5 × 0.5 × 3.0 mm = 0.75 mm³

If FOV_x → 400 mm, N_x fixed at 400:
  Δx = 1.0 mm/pixel  → resolution HALVED, SNR × 2
If FOV_x → 100 mm, N_x fixed at 400:
  Δx = 0.25 mm/pixel → resolution DOUBLED, SNR × 0.5
If FOV_x → 100 mm, N_x reduced to 200:
  Δx = 0.5 mm/pixel  → resolution UNCHANGED, scan time halved (rFOV)

18.2 Aliasing / Fold-Over Decision Tree

Is anatomy extent in PHASE direction > FOV_p?
│
├── YES → Fold-over WILL occur
│         │
│         ├── Option A: Increase FOV_p to cover anatomy
│         │   Trade-off: larger voxels (if N_y fixed) or longer scan (if N_y increased)
│         │
│         ├── Option B: Apply phase oversampling (≥ 100%)
│         │   Trade-off: small time increase proportional to oversampling / ETL
│         │
│         └── Option C: Change phase direction to the shorter anatomical axis
│             Trade-off: must reassess artefact sources in new direction
│
└── NO → FOV_p is adequate
         Apply rFOV if FOV_p >> anatomy → scan time reduction

18.3 SNR vs Resolution Trade-off at Fixed Field Strength

FOV REDUCTION (fixed matrix) → SMALLER VOXEL → HIGHER RESOLUTION
                                                 ↓
                                         LOWER SNR
                                         
Recovery strategies:
  + Increase NSA (time ∝ NSA)
  + Use surface coil (SNR from coil sensitivity)
  + Increase field strength (SNR ∝ B₀)
  + Enable DLR (SNR recovery without time cost)
  + Increase ETL (time recovery, T2 blurring cost)
  + Increase parallel imaging R (time recovery, g-factor SNR cost)

19. Evidence Gaps and Ongoing Debate

Optimal FOV for EPI-DWI across body regions: while reduced-FOV EPI has demonstrated clear distortion reduction in the prostate and spine, no large multicentre prospective trial has quantified the impact of specific FOV_p reductions on diagnostic accuracy for all body regions. The recommended FOV values in protocol guidelines (e.g., PI-RADS v2.1 for prostate) are based on expert consensus rather than formal FOV optimisation studies.

AI reconstruction and minimum FOV viability: DLR algorithms enable smaller FOV at previously unacceptable SNR. The minimum clinical FOV achievable with DLR (vs without) has not been formally established for any anatomical region. Some manufacturers are recommending smaller default FOV in DLR-enabled protocols without formal validation of diagnostic equivalence.

rFOV threshold for specific protocols: the acceptable rFOV ratio (minimum phase FOV as a percentage of frequency FOV) before clinical quality is compromised varies by anatomy and clinical question. No systematic study has defined these thresholds for routine clinical protocol families. Practice varies widely between centres (60–75% phase FOV for spine is routine in some centres; others use 100%).

3D vs 2D FOV equivalence at equivalent voxel size: 3D acquisitions with a given FOV and voxel size have different SNR properties (√N_z SNR advantage) compared with 2D acquisitions at the same voxel dimensions. The clinical equivalence (or superiority) of 3D for specific diagnostic tasks at matched voxel volumes has been demonstrated for some applications (brain MPRAGE, knee 3D FSE) but not systematically for all.

7T FOV expansion with B1+ correction: at 7T, the effective FOV is limited by B1+ inhomogeneity to approximately 200–250 mm for head imaging. Novel B1+ mitigation strategies (parallel transmit, dielectric pads, RF shimming) are progressively expanding the usable FOV at 7T, but no standardised solution for large-FOV 7T body imaging exists at the time of writing.

20. Miscellaneous and Future Directions

Historical milestone: the first clinical MRI scanners (1980–1983) used a fixed FOV of approximately 400–500 mm, determined by the linear range of their gradient coils. The ability to select variable FOV emerged in 1984–1985 as gradient linearity improved and digital control systems allowed variable gradient amplitudes. The concept of rFOV for spine imaging was introduced in the late 1980s as a time-saving strategy — it was one of the first protocol optimisation innovations adopted universally across vendors.

Reduced-FOV EPI (ZOOMit / iZOOM / FOCUS / REVEAL): the commercial introduction of 2D spatially selective RF pulses that simultaneously excite and spatially confine signal in both phase and frequency directions (using a "butterfly" pulse design) enabled reduced-FOV EPI without aliasing from outside the FOV. These techniques, introduced commercially between 2010 and 2015, represent the most significant FOV-related technical advance for DWI in body MRI.

AI-assisted FOV planning: automated anatomy recognition from localiser images can propose the optimal FOV for a given protocol and patient, reducing operator-dependent variability in FOV selection. This is increasingly available (Siemens AIAlign, GE SmartExam, Philips AutoSurvey) but requires validation of the proposed FOV against expert radiologist/technologist standards for each anatomical protocol.

Synthetic MRI and FOV: synthetic MRI (quantitative T1/T2 mapping followed by synthetic image generation at any TE/TR combination) theoretically decouples FOV from contrast selection — a single acquisition at a given FOV and resolution can generate T1, T2, PD, FLAIR, and other contrast-weighted images. This may reduce the number of separate acquisitions per protocol, each with individually optimised FOV, and instead produce all contrasts from a single high-resolution quantitative acquisition.

Compressed sensing and virtual FOV expansion: CS algorithms combined with highly parallel coil arrays can in principle reconstruct images at a virtual FOV larger than the physically acquired FOV, by exploiting coil sensitivity information to suppress aliasing from outside the acquisition FOV. This "coil-extended FOV" approach is under development for cardiac and body applications.

End of document — FOV Field of View — MRIninja v1.0 — May 2026