What this is
Notes on CT-guided spinal injections — what the four main injection types actually do, and why the cervical list looks so different from the lumbar one.
📊 Slide deck: Teaching deck for the team
Why Any of This Works
A spine in pain is almost always one of two stories, an angry nerve or an angry joint, and the job is getting a small dose into the millimetre or two of tissue that’s actually inflamed without hitting anything sitting next to it (cord, artery, dura, lung). I find that when I’m explaining this to new staff the thing that actually lands is that CT lets you see every millimetre of the path before the needle moves, which is the whole reason this works at all.
The whole logic, in one line
Get a small dose into a precise spot, while seeing every structure the needle has to avoid.
On the list day to day it tends to be one of a few stories repeating, a disc bulge pinching a nerve root sending pain down a leg or arm (radicular pain), wear and tear in the small facet joints at the back of the spine (facet pain), inflammation of the joint where the spine meets the pelvis (sacroiliac pain), or a test injection to tell the surgeon which level is the real culprit.
Why CT and Not Fluoroscopy
Fluoroscopy gives you a single flat image with the needle, bone and soft tissue all overlapping. CT gives you cross sections, so the doctor can see exactly how deep the needle is, how close to the dura, where the artery is on this particular patient. I find for anything cervical or thoracic that depth information is the difference between a routine case and a catastrophe.
The other thing that comes up a lot is variant anatomy. About one in three patients has something slightly unusual, a tighter foramen, an artery running a touch more medially, a facet angled off textbook. On fluoroscopy that’s invisible until the contrast goes in. On CT it’s there before the needle is even prepped, which I think is the bigger reason CT has taken over for spinal work.
Radiation isn't a free trade
CT does deliver more dose than fluoroscopy in principle, but intermittent low-mAs technique brings it down to roughly a minute of fluoroscopy. Modern CT-guided spine work is a low-dose procedure.
The Four Flavours
Four injection types account for almost everything on a CT spinal list. They differ in what they’re targeting more than in how the needle gets driven.
Epidural injection — flooding the corridor around the nerves
The epidural space is the thin envelope of fat and veins just outside the dura, which is the tough sleeve around the cord and nerve roots. Think of the cord and its nerves as electrical cables running through a conduit, with the epidural space as the loose packing material around them. The needle goes in from behind, between the laminae (the bony plates) of two adjacent vertebrae, stopping the moment it slips through the ligamentum flavum into the epidural space. Contrast goes in first to confirm position, you’ll see it spread upwards and downwards through the space, then the steroid mix follows. This one is the workhorse for diffuse, multi-level pain where the picture isn’t pointing at one specific nerve.
What you'll hear it called
“Interlaminar epidural” or just “epidural.” The needle came in between the laminae rather than through a foramen.
Transforaminal injection — putting the drug right at the angry nerve
The neural foramen is the doorway where a single nerve root exits the spine. If the pain follows one specific nerve, classic sciatica or pain that runs reliably down one arm, the most efficient thing is to put the drug at that exact doorway. The needle threads in from the side into the foramen, and the steroid lands on the ventral side of the root, which is the side the disc actually contacts.
Why this one works so well
Drug ends up on the right side of the root. Roughly 70% of patients improve meaningfully after a transforaminal at the right level, versus around 45% for a plain interlaminar.
The tradeoff is risk. The foramen is a narrow window with a nerve root and an artery threading through it. That’s what makes transforaminal work a careful procedure in the lumbar spine, and a different sport entirely in the neck.
Selective nerve root block — same picture, different intent
The needle goes roughly where a transforaminal would. The difference is volume and goal.
A nerve root block uses a tiny volume (around half a millilitre) on a single root, with the deliberate goal of not spreading anywhere else. The point is diagnostic, if numbing one root takes the pain away, that’s the generator, and the surgeon knows which level to operate on. A transforaminal uses several millilitres and wants the spread, treating both the root and the epidural inflammation feeding it.
Same approach, different syringe size
The same needle position can be used for either. The giveaway is the volume drawn up and whether the doctor asks the patient afterwards “is your pain gone?” (diagnostic) or “we’ll see you in a few weeks” (therapeutic).
Facet joint injection — into the little joint at the back
The facets are the small paired joints at the back of each spinal level, the bits that let the spine bend and twist. They wear out with age. A worn facet generates a deep central back ache that gets worse with extension (leaning back). It doesn’t run down a leg, because facets aren’t on a nerve path.
The needle goes straight in from behind into the joint capsule. Contrast lights up the joint as a thin crescent, then local and steroid follow.
Two ways to attack a facet
Inject the joint directly, or block the tiny nerve that supplies it — the medial branch of the dorsal ramus. The medial branch block is diagnostic only: if blocking the nerve takes the pain away, the next step is usually radiofrequency ablation (RFA — burning the nerve for months of relief).
Sacroiliac joint injection — the easiest one on the list
The sacroiliac (SI) joint is where the lower spine meets the pelvis. It’s a common source of one-sided low back pain that doesn’t behave like disc pain.
This one is the technical opposite of cervical work, a big posterior joint well clear of the spinal cord, easily reached prone. The literature suggests some departments use it as the teaching case for new operators because of this.
How a Case Unfolds
The setup is the same regardless of flavour. Patient prone, pillow under the abdomen, planning CT through the target level, skin marked, swabbed, draped, local infiltrated.
The needle is usually a 22-gauge spinal needle (Quincke or Whitacre), 90 to 120 mm depending on build. The doctor advances it a few millimetres at a time, taking a short CT slice between each advance. You’ll see the needle as a bright white line on each slice, and every breath nudges it slightly, which is why the slices stay short.
Once the tip is in place, the doctor will confirm position before any drug goes in. Most commonly that’s 0.5 to 1 mL of Omnipaque (iodinated contrast, the standard for spinal work in our department), and the contrast pattern tells you which compartment the tip is sitting in, a feathery spread for an interlaminar, a tubular nerve sleeve for a transforaminal, a thin crescent inside a facet, the outline of the SI capsule. Some doctors prefer to use air instead, especially for facet joints, where a small bubble in the joint shows up clearly on CT and avoids contrast entirely. Wrong pattern means the needle isn’t where it should be. For instance, contrast disappearing rapidly down the screen is intravascular uptake, and the needle pulls back before any drug goes in.
The drug mix is small, typically 1 to 2 mL of local anaesthetic (bupivacaine for longer action, lignocaine for fast onset) and 1 mL of steroid. The steroid choice is the central safety question, and it sharpens as you move up the spine:
- Particulate steroids (triamcinolone — Kenacort; methylprednisolone — Depo-Medrol; betamethasone — Celestone Chronodose) form a depot and release slowly. Longer-acting, but big enough to plug a small artery if they end up in one.
- Non-particulate steroids (dexamethasone — Dexmethsone, Maxidex) are fully dissolved. Particles ~0.5 µm, less than a tenth of a red blood cell, no aggregation. Safer in the wrong place, but the effect doesn’t last as long.
Why Most Lists Are Lumbar, SI, and Thoracic — and Why Cervical Is Different
A typical CT spine list is mostly lumbar transforaminals, plenty of facets, a steady run of SI joints, occasional thoracic interlaminar epidurals, and only a small number of cervicals. The reason comes down to what’s sitting next to the needle.
Lumbar and SI are forgiving
The spinal cord ends at L1–L2 (the tapered tip is the conus medullaris). Below that, the canal contains the cauda equina, a loose bundle of nerve roots floating in CSF. Nudge one with a needle, the patient gets a brief zing, the operator backs up a millimetre. Plenty of space, no cord-supplying arteries at this level.
The SI joint is even easier, no cord, no canal, no foramen, just a big posterior joint.
Thoracic is intermediate
The cord runs all the way through the thoracic spine and sits closer to the needle. The thoracic foraminal approach is mostly not done anymore for that reason, thoracic injections are almost always interlaminar epidurals where the needle stays midline. Done carefully on CT this is routine, but the margin is smaller than in the lumbar.
The cervical spine is a different problem
Three things stack up that don’t apply lower down.
The vertebral artery sits next to every cervical foramen. It supplies the brainstem and the back of the brain, and runs up through a tunnel of bony rings (the foramen transversarium) right next to the doorway the nerve root exits. Variant anatomy is common, accessory vessels, lateral loops, an artery running a touch more anteriorly, and any of it can put the artery directly in the path of a transforaminal needle.
Tiny arteries supply the cord through the foramen. Radiculomedullary arteries are small branches that join the cord’s blood supply through the same foramen the needle is going through. They can be too small to see on planning CT and too small to feel on the needle. Steroid inside one is catastrophic.
Particulate steroid in the wrong artery causes a stroke. Particulate steroids form clumps. If those clumps get into a vertebral or radiculomedullary artery, they embolise into the brainstem, the cerebellum, or the cervical cord. Animal studies showed irreversible infarction. Human cases mostly happened before non-particulate steroid became standard.
The FDA warning that changed practice
In 2014 the FDA issued a drug safety warning after 131 reported neurological adverse events from epidural steroid injections, most tied to cervical transforaminals. The current consensus — international, not US-specific — is that cervical injections use non-particulate dexamethasone only.
On top of the steroid switch, cervical work involves extra safety steps you won’t see on every lumbar case. Pre-procedural MRI or CT review for variant vertebral artery anatomy, digital subtraction angiography (DSA) to detect intravascular uptake that real-time imaging misses, sometimes a small anaesthetic test dose before any steroid, and often a posterolateral or transfacet approach that comes in from behind the artery rather than past it.
Why fewer operators do cervical work
Detailed pre-procedural review, a different steroid, additional imaging steps, a tighter margin for error, and complications that include stroke and tetraplegia. Lumbar and SI are high-volume, low-stakes-per-error, easy to stay sharp at. Cervical is low-volume, high-stakes-per-error, and proficiency takes deliberate maintenance. Most departments concentrate cervical work in the hands of operators who do it often enough to stay good at it.
It’s not that cervical injections are dangerous in absolute terms. Modern series using non-particulate steroid report no major complications across more than a thousand cases, so the risk is well-controlled. The reason fewer people do them is that the consequences of being off by a millimetre are out of proportion to the rest of spinal work.
What Follows the Procedure
Walked to recovery, observed 20 to 30 minutes, obs and a quick neurological check appropriate to the level. Most patients walk out feeling no different from when they came in.
Steroid takes 2 to 5 days to start working. The local in the mix gives an hour or two of immediate relief, which doubles as a diagnostic signal, if leg pain disappears as the local kicks in, the right level was injected.
Worth flagging to the patient: a steroid flare (brief worsening before improvement, around 1 in 10), possible facial flush, a day or two of sleeplessness from systemic steroid effect, and instructions to watch the puncture site.
Things that need a phone call back
Severe headache that’s worse sitting up than lying down (CSF leak from inadvertent dural puncture), fever, expanding swelling at the puncture site, new weakness or numbness, any new neurological symptom. Cervical patients especially — any new dizziness, vision change, or weakness gets seen immediately.
Numbers Worth Knowing
| Stat | Value | Why it matters |
|---|---|---|
| Transforaminal vs interlaminar improvement | ~70% vs ~45% | Why transforaminal is preferred for single-root pain |
| Intravascular uptake at cervical level | Higher than thoracic or lumbar | Why DSA matters more in the neck |
| Cervical steroid choice | Non-particulate dexamethasone only | Post-2014 FDA safety warning |
| Dexamethasone particle size | ~0.5 µm | < 1/10th of a red blood cell — won’t embolise |
| Nerve root block volume | 0.5–0.6 mL | Deliberately tiny — diagnostic only |
| Transforaminal volume | 1–4 mL | Larger because spread is the goal |
| Steroid onset | 2–5 days | Manage patient expectations on the day |
| Steroid flare | ~1 in 10 | The 24–48 hour worse-before-better pattern |
Quick Reference: What Goes Where
| Injection type | Target | Best for | Typical volume |
|---|---|---|---|
| Interlaminar epidural | Epidural space behind the cord | Multi-level or diffuse radicular pain, stenosis | 2–4 mL |
| Transforaminal | Foramen, on the ventral root | Single-level radicular pain (sciatica, arm pain) | 1–4 mL |
| Selective nerve root block | One nerve root, minimal spread | Diagnosing which level is the generator | 0.5–0.6 mL |
| Facet joint | Inside the facet capsule | Axial back pain from facet arthrosis | 1–2 mL |
| Medial branch block | Nerve to the facet | Diagnostic before facet RFA | 0.3–0.5 mL per branch |
| Sacroiliac | Inside the SI joint | One-sided low back / buttock pain | 2–3 mL |
Related
- Liver Tumour Embolisation — different procedure, same idea of letting a planning CT pre-empt the dangerous bits of the anatomy
- Post-Lumbar Puncture Bed Rest — the other CT-guided needle-in-the-spine procedure most departments do
- Genicular Artery Embolisation — another image-guided pain procedure, this time the knee
Sources
- Spine injections: the rationale for CT guidance — Skeletal Radiology (PMC)
- CT-guided transforaminal cervical and lumbar epidural injections — ScienceDirect
- Technique for CT Fluoroscopy–Guided Lumbar Medial Branch Blocks and RFA — AJR
- The Lumbar Neural Foramen and Transforaminal ESI: Anatomic Review — AJR
- Comparison of three CT-guided cervical epidural approaches — Skeletal Radiology
- Major risks and complications of cervical epidural steroid injections (PMC)
- Safety of cervical transforaminal ESI (PMC)
- Factfinders for patient safety: minimising risks with cervical epidural injections (PMC)
- Spine Injectables: What Is the Safest Cocktail? — AJR
- Particulate vs non-particulate steroids for epidural injections — Skeletal Radiology
- DSA vs real-time fluoroscopy for intravascular detection (PMC)
- Facet Joint Injection — StatPearls
- CT-guided sacroiliac joint injections — PubMed
- Spinal Injections for Pain Management — Radiology (RSNA)
- Cervical Epidural Injection — StatPearls
Last updated May 17, 2026.