Neuroradiology series · Part 1 of 4

Reading the Compressed Cord

Cervical neuroradiology for advanced practice providers — what to look at, in what order, and what fools you.

Companion to the timeline figures.

1Before you open the study

Three habits, ten seconds each. They change what you see.

Know the clinical question first. "Weak hands, dropping objects, wide-based gait, eight months" sends you to the cord. A read without a question is a tour, not an examination — and you will find degeneration, because everyone has it.

Do not read the report first. If you are trying to build recognition, reading the impression destroys the training value of the study. You will see what you were told to see. Read cold, commit in writing, then compare.

Check the study can answer the question. No axial T2 through the level of interest? Then nobody can grade the compression, including the radiologist. Recognising an inadequate study is itself a clinical skill, and catching it early saves a wasted clinic visit.

2Which sequence is which

One heuristic settles it: CSF bright → T2 or STIR. Fat also bright → T2. Fat dark → STIR.

Sagittal STIR cervical spine — STIR
CSF bright, fat dark

Sagittal T2 cervical spine — T2
CSF bright, fat bright

Sagittal T1 cervical spine — T1
CSF dark, fat bright

Look only at the CSF column to tell these apart — it is bright, bright, dark from left to right.
Sagittal STIR / T2 / T1 of one cervical spine, one patient, one sitting. From a paediatric flexion injury, so the vertebral marrow is redder and less bright on T1 than an adult’s would be — the CSF signal is the reliable cue at any age.
GRE is deliberately not in this strip — in a routine cervical protocol it is acquired axially, not sagittally, so it does not belong in a sagittal comparison. See below.
Sjoberg et al., Children 2023;10:1094, Fig 9 CC BY 4.0

The orientation image

One cervical spine, the same sitting, three sequences side by side

This is the single highest-value teaching image in the talk. Everything downstream depends on the audience knowing which sequence they are looking at without being told.

T1: Anatomy, marrow, myelomalacia (dark cord). Fat is bright, CSF is dark.
T2: CSF contrast, cord signal change, disc morphology. Both CSF and fat bright.
STIR: Edema — marrow, ligament. CSF bright, fat suppressed to dark.
GRE gradient echo: Calcification, ossified ligament, haemorrhage. Usually an axial acquisition — see the next card.

The fourth sequence: gradient echo

Gradient echo (GRE) is the one you are least likely to have seen, because it is not part of the sagittal stack. Most cervical protocols acquire it axially, angled through each disc space. Your scanner probably calls it something else — MERGE on a GE machine, MEDIC on a Siemens, or just T2* (say "T-two-star"). Those are vendor brand names for the same family of sequence.

You are not usually asked to identify a gradient echo at a glance. You are asked whether it is worth getting, and there are two answers that matter.

OPLL = ossification of the posterior longitudinal ligament. The posterior longitudinal ligament is the strap running down the back of the vertebral bodies, immediately in front of the cord. In some people it slowly turns to bone. When that happens it stops behaving like a ligament and becomes a rigid bar of bone pressing on the cord from the front — and it keeps growing.

It matters to you for one reason: it changes the operation. Ordinary spondylosis can often be decompressed from behind, letting the cord drift backwards away from the disc and bone in front. If a thick bar of ossified ligament is in the way, the cord may have nowhere to drift to, and the surgeon has to come from the front instead. (The measurement that predicts this is called the K-line; you do not need to draw it, only to know that the question exists.) Commonest in patients of East Asian ancestry, but present everywhere.

Sagittal T2, T1, gradient echo and CT of the same cervical spine showing OPLL — **Reason one — it settles ossified ligament, and that changes the operation.** On T2 (a) and T1 (b) the band behind the vertebral body (arrowhead) is just dark — and dark on both is ambiguous, because ossified ligament, calcified disc and cortical bone all look identical. The gradient echo (c) resolves it; CT (d) confirms.
**One caveat, or it will confuse people:** panel (c) is reconstructed to look like CT, so the ossification appears *bright*. On the standard T2* gradient echo in your own protocol, calcium and blood do the opposite and *bloom dark*. Same family of sequence, opposite appearance.
Yoon et al., *Neuroradiology* 2025, Fig 1 CC BY 4.0

Axial T2 with an eccentric dark spot beside the cord — T2 — dark spot
blood, or just flow?

Axial gradient echo showing the same spot blooming larger and blacker — GRE — it bloomed
bigger, blacker = blood

Rakesh et al., Insights Imaging 2025;16:2153, Fig 5c–d CC BY 4.0

Reason two — it separates blood from flow artifact

Axial T2 first, then axial GRE at the same level

An eccentric dark spot beside the cord on T2 is ambiguous: it could be haemorrhage, or it could just be moving CSF. The gradient echo decides it, and the two answers look opposite:

Blood: Blooms — the dark spot gets bigger and blacker on GRE than it was on T2.
Flow: Vanishes — the dark patch disappears entirely and the CSF goes uniformly bright.

This card shows the blooming version. You will see the vanishing version in section 5, where the same trick is used to throw out a false call of cord compression.

3The eight-step read

Run it in this order every time. The order is front-loaded by urgency, so if you are interrupted after step three you have already found what matters.

4The findings that change management

Five findings. For each: where to look, what you see, how to grade it, and what it means.

The colour of each card is its severity. Orange = mechanical, potentially reversible · Red = the cord itself is injured. They are ordered worst-last: cord atrophy is the most serious finding on this page, not the mildest.

Sagittal T2 with effaced CSF at C5-6 and C6-7

Kwok & Cheung 2020, Fig 1 · CC BY 4.0 — free to use

Loss of the CSF stripe

Sagittal T2 to detect · axial T2 to characterise

The bright CSF outlining the cord thins, then disappears. Trace the anterior and posterior columns separately — they are effaced independently and the pattern matters.

Grade 0: Normal subarachnoid space
Grade 1: Partial obliteration, anterior or posterior
Grade 2: Complete obliteration, no cord deformity
Grade 3: Cord deformity or impingement

What it means: effacement without deformity is the entry point to the "asymptomatic compression" conversation. On its own it is not an operative finding.

Focal ventral impingement, CSF preserved — indented
CSF preserved

Flattened cord, partial CSF loss — flattened
ventral CSF partly gone

Severe circumferential compression, CSF effaced — severe
CSF effaced

Banana-shaped cord with compression ratio 0.37 — the same change, measured
compression ratio 0.37 · cross-sectional area 71 mm²

Grades: Kadanka et al., Front Neurol 2024;15:1341371, Fig 1 CC BY 4.0
Measured example: Bednarik et al., Eur Spine J 2008, Fig 2b · permission pending

Cord deformation

Axial T2 at the narrowest level — never grade shape off a sagittal

The normal cervical cord is a rounded oval, wider side-to-side than front-to-back. Under compression it flattens, then becomes crescentic as the anterior surface is indented.

Compression ratio = smallest AP diameter ÷ largest transverse diameter. Lower is worse. Two readers measuring the same cord agree closely on it (ICC 0.80, where 1.0 is perfect).

Cord shape beats canal diameter — it reflects what is happening to neural tissue rather than to the container.

Arvin et al., Spine 2013, Figure 1b–c · permission pending

T2 hyperintensity in the cord

Sagittal T2 to detect · axial T2 to confirm it is real

Bright signal within the cord substance, usually at or just above the level of maximal compression, often centred on central grey matter. Characterise it as focal (one segment, crisp) or diffuse (spanning segments, hazy), and count the levels.

Multilevel T2	Worse baseline severity and lower recovery ratio (p = 0.001). Present in 27%.
Single-level T2	Outcomes no different from no signal change at all (p = 0.275).

T2 signal alone decides nothing. It raises concern; it does not make the decision. Count levels before you weight it.

Low T1 signal in the cord — low T1 signal

New T1 hypointensity after surgery — new post-op T1 hypo

Arvin, Spine 2013, Fig 1e · Kato, Spine 2018, Fig 2d · permission pending

T1 hypointensity — the money finding

Sagittal T1, deliberately, at every level where T2 was bright

Cord darker than normal cord at the same level. This is cavitation and established myelomalacia, not reversible oedema. It will not announce itself — you have to switch sequences and look.

Reduced recovery ratio	p = 0.03
Lower odds of an optimal outcome	OR 0.45, p = 0.005
Post-op resolution improves outcome?	No (p = 0.36) — unlike T2

The caveat that must travel with it. Inter-rater agreement is only fair (κ = 0.31, versus 0.60 for T2), and it appears in under 20% of patients. When you think you see it, say "possible T1 hypointensity — worth a second look" rather than asserting it.

Say this to the patient: "This doesn't mean don't operate. It means the goal shifts from restoring what you've lost toward protecting what you still have."

Axial and sagittal T2 showing a thinned cord with restored CSF space

The canal is not roomy because the compression resolved — it is roomy because the cord shrank into it.
Two-year follow-up MRI after single-level anterior cervical corpectomy. Cureus 2026;18:e107566, Fig 6 CC BY 4.0

Cord atrophy

Sagittal T2 for impression · axial T2 compared to uninvolved levels

This is the worst finding on the page. Cord tissue has been lost and will not grow back. Where T2 signal change can resolve after decompression, atrophy does not. Surgery here is about protecting what remains, not restoring what has gone — and expectations should be set that way before the operation, not after it.

Reduced cord calibre, often with a deceptively roomy CSF space — the canal looks decompressed because the cord shrank, not because the compression resolved. A classic trap in a severely myelopathic patient.

Mimic: physiologic tapering below the cervical enlargement (roughly below C6) is normal. Compare to expected calibre at that level, not to the enlargement above.

The example shown is a two-year post-operative follow-up, where the contrast is easiest to see — but atrophy is not a post-surgical finding. It occurs in long-standing untreated compression too, and that is where it does the most damage: a scan that reads as "adequate canal" in a patient who is visibly myelopathic. Trust the examination over the roominess.

5Three things that will fool you

Every one of these is a false positive that looks convincing on a single image and evaporates on a second plane. This is the highest-yield section of the talk.

Truncation (Gibbs) artifact — the commonest false-positive cord finding

A thin bright line down the centre of the cord on sagittal T2. It is linear, uniform across many segments, and vanishes on the axial. Real signal change is segmental and persists.

What fools you — sagittal T2

Sagittal T2 with a central bright line in the cord

Reads as a syrinx or extensive cord signal change. Note the yellow localiser line — that is the level of the axial beside it.

The disproof — axial T2, same patient

No cavity — the central signal is lip-shaped, not round. Always confirm cord signal on the axial before you call it.
Nayak & Gaikwad, Korean J Radiol 2019;20:1474, Fig 1, cropped CC BY-NC 4.0

CSF flow artifact — mimics effacement, and mimics blood

Moving CSF loses signal, so dark patches appear in the subarachnoid space and read as loss of the CSF stripe — or, when eccentric, as haemorrhage. Named explicitly as a source of diagnostic error in the cervical stenosis literature. A gradient-echo sequence settles it: the artifact disappears, real blood blooms darker.

What fools you — axial T2

Axial T2 with flow-related signal loss around the cord

Looks like the CSF around the cord has been squeezed out.

The disproof — gradient echo, same level

Axial GRE at the same level, CSF uniformly bright

A clean bright ring of CSF, circumferentially. The T2 was lying.
Rakesh et al., Insights Imaging 2025;16:2153, Fig 5a–b CC BY 4.0

Oblique slice angle — makes a compressed cord look normal

If the axial slice is not perpendicular to the cord — common at the apex of a kyphosis — the cord is cut at a slant. The slant stretches the front-to-back dimension by 1 ÷ cos θ while leaving the side-to-side dimension untouched, so the cord looks rounder and larger than it really is. The compression ratio comes out too high and the cross-sectional area is overestimated. This error runs toward false reassurance — it makes a compressed cord read as an adequate one, which is why it is worth knowing. Check the slice angulation on the localiser before you trust a compression ratio.

What fools you — an axial cut on the slant

The number looks reassuring. It is an artefact of the geometry, not the cord.

The fix — re-angle to the cord

Angle each axial block to the disc space it crosses, rather than one fixed tilt for the whole stack. If you cannot re-angle, grade at a level where the cord runs straight.
Schematic drawn for this handout · geometry, not a case

So how do you fix it?

Reading a study already done: You cannot un-angle it, and nothing you do at the workstation recovers the true number. Do not quote a compression ratio from that level. Grade at a neighbouring level where the cord runs straight, or describe it in words — "severe ventral compression at C5–6" is defensible; "compression ratio 0.64" is not.
If the decision turns on it: Ask for a re-angled axial block through that level, or a reformat perpendicular to the cord if a 3D isotropic sequence was acquired. That is minutes of scanner time, and cheap when the question is operate or watch.
Preventing it next time: Axial blocks should be angled to each disc space rather than one fixed tilt for the whole stack. Worth mentioning to your techs once — it is a protocol habit, not a special request.
The sentence that protects the patient: "Measurement unreliable at this level — slice angulation." That tells the surgeon far more than a falsely reassuring number does, and it is the one thing here that is entirely within your control.

6What you can and cannot call

Two patients with the same degree of cord compression, one myelopathic and one not — **This is the whole reason the section exists.** Top row (A, B) is a patient with symptomatic myelopathy. Bottom row (C, D) is a patient with **no myelopathic signs or symptoms at all.** Same sequences, same planes, and no visible difference in the severity of compression. You cannot read severity of disease off severity of compression — the examination decides that, not the picture.
Kadanka et al., *Front Neurol* 2024;15:1341371, Fig 2 CC BY 4.0

Call it and act on it	Defer it, always
Presence, level and number of levels of compression	Any mass, marrow replacement or enhancing lesion
Loss of CSF signal around the cord	Infection — discitis, epidural abscess, paraspinal collection
Cord deformation and gross atrophy	Intramedullary lesions that are not clearly compression-related
T2 hyperintensity and (cautiously) T1 hypointensity	Anything outside the spine on the visible slices
Gross alignment — lordosis, kyphosis, listhesis	Post-operative scar versus recurrent disc
Study adequacy for the clinical question	Anything you are about to talk yourself into

The calibration that makes all of this safe. 52% of asymptomatic adults have disc bulges; disc degeneration runs from 37% at age 20 to 96% at age 80. Finding degeneration on an MRI is about as informative as finding grey hair. It acquires meaning only when it explains this patient's syndrome at that level — which is why you establish the clinical question before you open the study.

Sources for the figures quoted above

Nouri A, et al. Spine 2017;42:1851–1858 (n = 419) — T1 and T2 signal prognostics, inter-rater reliability. · Kato S, et al. Spine 2018;43:824–831 (n = 167) — post-operative signal resolution. · Karpova A, et al. Spine 2013;38:245–252 — reliability of quantitative measures. · Jensen MC, et al. NEJM 1994;331:69–73 and Brinjikji W, et al. AJNR 2015;36:811–816 — asymptomatic base rates. · Crawford CH III, et al. JAAOS 2026;34(4):e555–e560 — artifact as a source of diagnostic error.

Prepared for the MANS 2026 Summer Meeting APP session. This supports clinical triage and communication; it does not substitute for formal radiologic interpretation.