MRI Orbits for Globe and Intraocular Pathology

Version 1.0 — August 2026

1. Executive Summary

1.1 Added Value over the Generic Protocol

Ultrasound remains the first-line imaging modality for most intraocular pathology, given its excellent resolution, real-time capability, and wide accessibility; MRI’s specific, genuinely added role in this domain is narrower but clinically essential — extrascleral extension assessment for uveal melanoma (a direct staging and management determinant), differentiating retinoblastoma from its common paediatric mimickers when the diagnosis is genuinely unclear, and characterising the globe when dense vitreous haemorrhage or other media opacity makes ultrasound and direct ophthalmoscopic assessment difficult or impossible. This child page documents the specific quantitative and qualitative imaging signatures that make MRI genuinely diagnostic in these scenarios, built around two central, evidence-dense differentials: uveal melanoma characterisation (exploiting melanin’s distinctive paramagnetic MRI signature) and retinoblastoma-versus-mimicker differentiation in children.

1.2 Limits of the Dedicated Protocol

MRI characterises uveal melanoma and its local staging features with genuinely high accuracy, but retains real, specific limitations: flat, minimally elevated tumours are poorly assessed for basal diameter on MRI relative to ultrasound, and — as directly demonstrated in histopathology-validated original research — small areas of extrascleral extension can occasionally be missed even on dedicated, high-resolution imaging. Similarly, while specific MR imaging features distinguish retinoblastoma from its principal mimickers (Coats’ disease, persistent fetal vasculature) with strong, quantified specificity for several individual signs, no single feature is used in isolation in practice; a structured, multi-feature assessment — as documented in Section 5 — is the correct interpretive approach, and definitive diagnosis in genuinely indeterminate cases still depends on the full multidisciplinary clinical picture.

2. Clinical Context

2.1 Clinical Presentation

Presentation depends entirely on the underlying pathology: uveal melanoma is frequently detected incidentally on routine fundoscopic examination, or presents with visual field defect, floaters, or flashing lights as the tumour or an associated retinal detachment develops; retinoblastoma in children classically presents with leukocoria (an abnormal white pupillary reflex, often first noticed by a parent or in a photograph) or strabismus; and vitreous haemorrhage or dense media opacity of any cause presents with sudden or progressive visual loss and a specifically limited ophthalmoscopic view, which is itself the direct reason MRI (rather than ultrasound, which can also image through haemorrhage, or repeat ophthalmoscopy) is sometimes specifically requested.

2.2 Two Central Diagnostic Tasks — the Organising Framework for This Page

This page is organised around its two dominant, evidence-dense clinical tasks: uveal melanoma characterisation and staging (Section 5.1-5.2), the most common primary intraocular malignancy in adults; and retinoblastoma-versus-mimicker differentiation (Section 5.3-5.4), the most common primary intraocular malignancy in children and a presentation in which false-positive diagnosis risks unnecessary enucleation while false-negative diagnosis risks life-threatening delay. Additional intraocular findings — retinal and choroidal detachment, vitreous haemorrhage — are addressed as supporting semiotic features throughout Section 5 rather than as separate standalone entities, since they most often present as secondary findings accompanying one of these two central diagnoses rather than as independent primary indications.

2.3 Differential Diagnosis (Clinical)

An adult intraocular mass should be distinguished from choroidal naevus (typically smaller, flatter, and stable on serial imaging — addressed further in Section 5.1), choroidal metastasis (usually multiple, flatter, and with a relevant primary malignancy history), and choroidal haemangioma; a child presenting with leukocoria requires the specific retinoblastoma-versus-pseudoretinoblastoma differential addressed in depth in Section 5.3-5.4, given that of children referred with suspected retinoblastoma, a substantial minority ultimately have a benign mimicking condition instead.

3. Indications, Timing, and Patient Selection

3.1 When the Dedicated Protocol Is Indicated

Suspected uveal melanoma requiring extrascleral extension and optic nerve invasion assessment for staging and treatment planning; suspected retinoblastoma, or a child with leukocoria/suspected pseudoretinoblastoma requiring differentiation; and dense vitreous haemorrhage or other media opacity precluding adequate ophthalmoscopic or ultrasound assessment are the principal indications for this dedicated protocol.

3.2 Baseline and Surveillance Imaging

For small, indeterminate choroidal lesions managed by observation rather than immediate treatment, and for retinoblastoma or treated uveal melanoma under surveillance, a well-characterised baseline study using consistent sequence parameters supports reliable interval comparison, exactly analogous to the surveillance principles already established for orbital mass lesions in the companion Orbital Tumour and Mass Lesion child page.

3.3 Red Flags Modifying Urgency

Rapidly progressive visual loss with suspected acute retinal or choroidal detachment, and any presentation raising genuine concern for retinoblastoma given the life-threatening implications of delayed diagnosis, warrant expedited rather than routine imaging and multidisciplinary ophthalmological/oncological review.

4. Dedicated Protocol Design

4.1 Mandatory Core Sequences

The table below lists the complete mandatory protocol for globe and intraocular pathology assessment — the seven generic-protocol core sequences (1-7) plus the one dedicated addition (8) detailed in Section 4.3.

4.2 Protocol Delta vs the Generic Protocol

4.3 Sequence-by-Sequence Utility for Globe and Intraocular Pathology

Sequences 1-2 (fat-suppressed T2) — baseline morphology and detachment characterisation. Beyond general lesion detection, T2 imaging is specifically useful for characterising the shape of any associated retinal detachment: a sharp, V-shaped detachment is a feature reported as almost exclusively associated with retinoblastoma in direct comparative paediatric imaging research, while a Y-shaped detachment favours a pseudoretinoblastoma diagnosis instead (Section 5.4) — a specific, genuinely useful morphological distinction this sequence pair is positioned to demonstrate.

Sequence 3 (non-fat-suppressed T1) — the single most diagnostically important sequence for uveal melanoma. Melanin has genuine, well-characterised paramagnetic properties that shorten both T1 and T2 relaxation times, producing the classic uveal melanoma signature of relative T1 hyperintensity and T2 hypointensity (Sequences 1-2) relative to the vitreous — a dual-sequence signal pattern reported, in a histopathology-validated original study, as significantly associated with tumour pigmentation on pathology. This is the specific reason this sequence is elevated beyond its generic baseline-anatomy role: for a genuinely pigmented uveal melanoma, this single sequence can be substantially diagnostic before contrast is even administered.

Sequences 4-5 (post-contrast T1) — enhancement characterisation and extrascleral extension. Nearly all uveal melanomas enhance, reflecting tumour hypervascularity, and — critically for staging — extrascleral extension appears with a signal intensity similar to the intraocular tumour component on these sequences, directly informing the American Joint Committee on Cancer staging-relevant determination of extrascleral extension, a finding of ≥5 mm specifically associated with increased metastasis and mortality risk and therefore a genuinely staging- and management-changing measurement this sequence pair exists to provide.

Sequence 6 (DWI/ADC) — a specific quantitative discriminator for retinoblastoma. Beyond its generic cellularity-marker role (as in the companion Orbital Tumour and Mass Lesion child page), apparent diffusion coefficient has been specifically investigated as a discriminator between retinoblastoma and Coats’ disease, with retinoblastoma’s higher cellularity producing measurably different ADC values than the exudative, lower-cellularity process of Coats’ disease — adding a genuinely quantitative dimension to the qualitative morphological differential detailed in Section 5.4.

Sequence 7 (whole-brain T2/FLAIR) — trilateral retinoblastoma screening. Unlike the generic protocol’s chiasm-level coverage, this sequence’s role is specifically elevated to a full intracranial screening role whenever retinoblastoma is a genuine diagnostic consideration, since trilateral retinoblastoma — synchronous or metachronous involvement of the pineal or suprasellar region by a histologically related primitive neuroectodermal tumour — is a recognised, life-threatening association that dedicated orbit-only imaging would not detect.

Sequence 8 (dynamic/time-resolved post-contrast T1) — perfusion pattern for uveal melanoma. Analogous in principle to the dynamic imaging technique already established for orbital mass lesions in the companion child page, dynamic contrast-enhanced imaging of uveal melanoma has been specifically characterised as showing either a wash-out or a plateau time-intensity curve pattern in the majority of cases — a specific perfusion signature contributing further diagnostic and, in some centres, prognostically-relevant information beyond static enhancement assessment alone.

4.4 Timing and Field of View Considerations for Dynamic Imaging

As with the analogous technique in the companion Orbital Tumour and Mass Lesion child page, dynamic imaging (Sequence 8) depends on capturing the genuine early-to-mid perfusion phase and should be planned, and injection timed, in advance rather than added retrospectively; given the globe’s small absolute size, this sequence additionally benefits from a deliberately small, globe-centred field of view distinct from the wider orbital FOV used elsewhere in this protocol, to preserve adequate temporal resolution without sacrificing the spatial detail this small structure requires.

4.5 Contrast Strategy

Gadolinium contrast is mandatory for both central diagnostic tasks in this protocol: enhancement pattern (static and, where indicated, dynamic) is central to uveal melanoma characterisation, and — while retinoblastoma-versus-mimicker differentiation depends substantially on non-contrast morphological and diffusion features (Section 5.4) — contrast remains part of the standard protocol for completeness and to support the broader differential. Standard macrocyclic GBCA dosing applies, with post-contrast fat suppression using Dixon or spectral/SPIR technique per the parent master page’s Section 4.3 reasoning — never STIR.

4.6 Sequence Matching to Clinical Question

5. MRI Semiotics of Globe and Intraocular Pathology

5.1 Uveal Melanoma — Signal Characteristics and Morphology

The classic uveal melanoma signature is T1 hyperintensity and T2 hypointensity relative to the vitreous, reflecting melanin’s paramagnetic T1- and T2-shortening properties; this relationship is not absolute, however, since approximately one-quarter of choroidal melanomas are amelanotic and typically appear T1-hypointense instead, though even these lesions may retain some T1 signal shortening from microscopic melanin, protein, or haemorrhage content. Morphologically, uveal melanoma may present as a dome, mushroom (particularly for larger tumours that have breached Bruch’s membrane), or flat/lentiform shape, generally enhances avidly (reported in up to 100% of lesions in one histopathology-correlated series despite variable vascularity on histology itself), and frequently shows restricted diffusion reflecting tumour hypercellularity.

5.2 Uveal Melanoma — Staging-Relevant Findings

Extrascleral extension — tumour signal extending beyond an otherwise normal-appearing sclera — is the single most management-relevant finding in this section, since American Joint Committee on Cancer staging criteria specifically link extrascleral extension of 5 mm or more to increased risk of metastasis and death, directly changing management toward more aggressive treatment (often enucleation) rather than globe-preserving therapy. Optic nerve invasion — tumour signal extending into or along the optic nerve on the sequences reviewed in Section 4.3 — is similarly staging- and management-relevant, and was demonstrated in over a third of cases in one histopathology-validated series where the tumour was noted on imaging to approach the optic nerve head. Associated retinal detachment is a common accompanying finding, more frequent with larger tumour prominence, and should be documented as a secondary finding rather than mistaken for the primary pathology.

5.3 Retinoblastoma — Direct Features

Retinoblastoma characteristically presents as an intraocular mass, frequently associated with calcification (a classically emphasised feature, though its absence does not exclude the diagnosis, per Section 5.4), vitreous seeding (tumour cells disseminated through the vitreous, reported as almost exclusively found in retinoblastoma rather than its mimickers), and, in more advanced cases, retinal detachment with the specific sharp, V-shaped morphology discussed in Section 4.3. Restricted diffusion, reflecting the tumour’s high cellularity, is a further supportive, increasingly quantified feature (Section 4.3).

5.4 Retinoblastoma vs Pseudoretinoblastoma — the Structured Differential

A large, multi-institutional European collaborative study directly comparing retinoblastoma with its two most common mimickers (Coats’ disease and persistent fetal vasculature) identified specific imaging features with strong, quantified discriminating power in each direction: larger eye size, vitreous seeding, and sharp V-shaped retinal detachment were almost exclusively found in retinoblastoma (specificity 93-97%), while smaller eye size, ciliary body/lens deformation, optic nerve atrophy, a central stalk between the optic disc and lens, Y-shaped retinal detachment, and absence of calcification were almost exclusively found in pseudoretinoblastoma (specificity 91-100%). This same study identified three newly-recognised discriminating features, including intraretinal macrocysts, found exclusively in pseudoretinoblastoma cases (specifically Coats’ disease). Critically, this evidence base supports a structured, multi-feature assessment approach rather than reliance on any single sign, since — as the underlying research explicitly emphasises — the combination of features, not an isolated finding, is what achieves genuinely reliable differentiation.

5.5 Relevant Classification Frameworks

Uveal melanoma staging follows American Joint Committee on Cancer (AJCC) TNM criteria, in which tumour dimensions, ciliary body involvement, and extrascleral extension (Section 5.2) are direct T-category determinants; retinoblastoma staging and management follow the International Intraocular Retinoblastoma Classification and International Retinoblastoma Staging System, both documented in dedicated ophthalmic oncology resources rather than reproduced in full here, with this child page’s imaging role specifically feeding the local-extent components of those broader staging frameworks.

5.6 Mimickers and Pitfalls

The single most important interpretive principle in this protocol, established directly by the underlying retinoblastoma-mimicker research, is that no individual imaging feature should be treated as independently diagnostic — eye size, detachment shape, calcification status, and the other features in Section 5.4 each carry strong but imperfect discriminating power individually, and genuinely reliable differentiation depends on their combined assessment. For uveal melanoma, the corresponding pitfall is assuming T1 hypointensity excludes melanoma, when amelanotic melanoma (Section 5.1) presents exactly this way and instead depends on the other features in this section (morphology, enhancement, diffusion restriction) for recognition.

6. Reporting Framework

6.1 Structured Reporting Template

Lesion description: location, size, morphology (dome/mushroom/lentiform for melanoma; mass with or without calcification for retinoblastoma). Signal characteristics: T1/T2 relative to vitreous, explicitly noting pigmented vs amelanotic appearance where relevant. Enhancement pattern: static and, where acquired, dynamic (wash-out/plateau) pattern. Diffusion characteristics: ADC value/qualitative restriction, explicitly stated. Extrascleral extension: explicitly measured and stated present/absent, with size if present. Optic nerve invasion: explicitly assessed and stated. Associated findings: retinal/choroidal detachment (with shape, where relevant to the retinoblastoma differential), vitreous seeding, calcification. For suspected retinoblastoma: explicit brain review for trilateral retinoblastoma.

6.2 Mandatory Reporting Elements

Every report addressing suspected uveal melanoma should explicitly state whether extrascleral extension is present and, if so, its measured extent relative to the 5 mm AJCC-relevant threshold, given how directly this determines management; every report addressing suspected retinoblastoma should explicitly and systematically address each of the structured differential features in Section 5.4 (eye size, detachment shape, calcification, ciliary/lens findings) rather than an isolated, unstructured impression, and should explicitly confirm whether the brain was reviewed for trilateral retinoblastoma.

6.3 Critical/Actionable Findings

Extrascleral extension ≥5 mm in suspected uveal melanoma, and any finding suggesting trilateral retinoblastoma (pineal or suprasellar mass in a child with retinoblastoma), are the most directly actionable, urgent findings in this protocol and should be flagged explicitly and prominently, given their direct, substantial impact on management and prognosis.

6.4 Common Reporting Errors

Reporting a T1-hypointense intraocular mass as “unlikely to be melanoma” without considering the amelanotic subtype specifically; describing retinoblastoma-versus-mimicker differentiation based on a single feature rather than the full structured assessment in Section 5.4; and omitting explicit extrascleral extension measurement in a report where a discrete choroidal mass is identified.

7. Technical Pitfalls

7.1 Resolution Limitations for Flat Lesions

As established in Section 1.2, MRI is specifically limited in assessing the basal diameter of flat, minimally elevated choroidal lesions relative to ultrasound; this is a genuine, acknowledged technical limitation rather than a correctable protocol error, and should be reflected in appropriately calibrated reporting confidence for tumour dimension measurement in this specific scenario.

7.2 Sequence-Specific Technical Considerations

Because extrascleral extension can occasionally be genuinely subtle — as directly demonstrated in histopathology-validated original research, where a small area of extension was initially missed on MRI despite being evident on subsequent histopathology and, retrospectively, on the same imaging — deliberate, careful scleral-margin review, rather than a cursory pass, is warranted specifically at the site of any identified choroidal mass.

7.3 When the Generic Protocol Alone Is Insufficient

A study performed using only the generic orbital protocol’s standard technique, without the globe-centred high-resolution T1 sequence (Section 4.3) and, where indicated, dynamic post-contrast imaging (Section 4.4), risks under-characterising exactly the melanin-based signal pattern and perfusion information this protocol’s uveal melanoma assessment depends on.

8. MRI Technologist Pearls

8.1 Sequence Planning for Globe-Centred Resolution

Plan Sequence 3 (and, where acquired, Sequence 8) with deliberate attention to achieving genuinely high resolution centred specifically on the globe, recognising that the diagnostic features this protocol depends on (melanin signal pattern, millimetre-scale extrascleral extension) operate at a finer spatial scale than the wider orbital structures the rest of the protocol addresses.

8.2 Paediatric Considerations for Suspected Retinoblastoma

Given that retinoblastoma most often presents in young children, sedation planning and coordination — following institutional policy, per the parent master page’s Section 3.1 — deserves particular attention in this specific indication, balanced against the genuine urgency described in Section 3.3.

8.3 Fast Salvage Protocol

If time is genuinely constrained, prioritise the non-fat-suppressed T1 sequence and post-contrast T1 (Sequences 3-5) for suspected uveal melanoma, and the T2/T1 morphological sequences plus DWI (Sequences 1-3, 6) for suspected retinoblastoma, over the dynamic sequence (Sequence 8) or additional anatomical detail, since these carry this protocol’s principal diagnostic weight for each respective indication.

8.4 Disease-Specific Common Avoidable Errors

Failing to specifically plan a globe-centred, high-resolution T1 acquisition distinct from the wider orbital sequences; omitting dynamic imaging for a discrete intraocular mass when it would have been genuinely informative; and, for suspected retinoblastoma, failing to extend brain coverage sufficiently to confidently assess the pineal and suprasellar regions for trilateral disease.

9. Quality Control Checklist

• Globe-centred, high-resolution T1 sequence confirmed acquired, distinct from the wider orbital field of view.
• Dynamic post-contrast imaging, where indicated, confirmed to have captured the genuine early-to-mid perfusion phase.
• Extrascleral extension explicitly assessed and, where present, measured against the 5 mm AJCC-relevant threshold.
• For suspected retinoblastoma, all structured differential features (eye size, detachment shape, calcification, ciliary/lens findings) confirmed explicitly assessed, not an isolated impression alone.
• Whole-brain coverage confirmed adequate to assess the pineal and suprasellar regions whenever retinoblastoma is a genuine clinical consideration.

11. Evidence Gaps and Ongoing Debate

• No single retinoblastoma-versus-mimicker imaging feature achieves perfect discrimination on its own. As explicitly emphasised throughout Section 5.4-5.6, even the strongest individual discriminating features carry specificity in the 91-100% range rather than absolute certainty, and the underlying research explicitly frames this as supporting a structured, multi-feature assessment rather than a single definitive sign — genuinely indeterminate cases remain possible despite this well-characterised evidence base.
• ADC-based quantitative differentiation between retinoblastoma and Coats’ disease is a comparatively recent addition to the evidence base, with fewer, smaller studies than the morphological feature-based differential in Section 5.4, and further validation of specific quantitative ADC thresholds across institutions remains an active area of ongoing research.
• MRI’s specific limitation in assessing basal diameter for flat uveal melanomas (Section 7.1), and its occasional under-detection of subtle extrascleral extension, are genuine, acknowledged technical limitations rather than fully resolved issues, and ultrasound retains a complementary, sometimes superior role for these specific measurements even where MRI is otherwise the primary staging modality.

12. Evidence-Based References

A. Guidelines / Consensus / Society Recommendations

No dedicated society guideline specific to globe/intraocular MRI protocol design, distinct from the general ACR–ASNR–SPR Practice Guideline already referenced on the parent master page, was identified as warranting a separate citation for this child page. Uveal melanoma and retinoblastoma staging follow AJCC and International Intraocular Retinoblastoma Classification/International Retinoblastoma Staging System criteria respectively (Section 5.5), which are broader oncological staging frameworks rather than MRI-protocol-specific guidelines. Category A is therefore not populated for this child page.

C. Important Prospective / Original Studies

D. Technical MRI Papers

Represented by the original studies already listed under Category C, which include substantial technical protocol detail alongside their clinical findings; a separate, non-duplicative Category D entry is not populated to avoid citing the same sources twice.

E. Landmark Historical References

No landmark historical reference specific to globe/intraocular MRI, distinct from the modern comparative literature already cited, was identified as warranting separate citation. Category E is therefore not populated for this child page.

End of document — MRI Orbits for Globe and Intraocular Pathology — Child Protocol under the MRIninja Orbit / Visual Disorders master page — v1.0 — August 2026 Parent page: MRI Orbits — Generic Standard Protocol