CT Artifacts (Beam Hardening, Metal, Motion)
- An artifact is the scanner telling you a confident lie: pixels that don't match the patient.
- Beam hardening, metal, and motion are the big three, and each has a signature look you can learn to spot.
- The danger isn't the artifact itself — it's mistaking it for disease, or letting it bury the disease that's actually there.
- Most artifacts have a fix or a workaround: change the window, re-scan, reconstruct differently, or just know it's fake.
A CT scanner is basically a very expensive machine for measuring how much of an X-ray beam survives the trip through you. When that measurement is honest, you get a beautiful map of the body. When something fools the measurement, the scanner doesn't shrug and say "I'm not sure" — it draws the lie in crisp grayscale and hands it to you like it's gospel. Those confident lies are artifacts, and your job is to recognize them before they recognize you.
Let me introduce the three you'll meet constantly.
Beam hardening: the beam gets tougher as it goes
An X-ray beam isn't one pure energy — it's a mix, like a bag of marbles in different sizes. The low-energy ("soft") photons get eaten first as the beam plows through tissue, so what comes out the far side is harder (higher average energy) than what went in. The scanner, bless it, assumes the beam stayed the same the whole way. It didn't. The math comes out wrong.
You see this two ways. Cupping makes the center of a uniform object look darker than its edges, as if someone scooped out the middle — because the beam crossing the thick center hardened more. Streaks show up as dark bands between two dense structures, most famously between the petrous bones at the skull base, where it can smear right across the brainstem.
That dark band across the posterior fossa — the Hounsfield bar — is beam hardening between the dense petrous temporal bones. It loves to hide (or fake) a brainstem stroke. If a low-density "lesion" sits exactly between two chunks of bone and respects no vascular territory, suspect the physics, not the patient.
Beam hardening is woven into how the machine reads density, so it helps to be comfortable with Hounsfield units and windowing before you trust any gray patch near bone.
Metal: the artifact that eats the room
Metal is the diva of CT. A hip prosthesis, dental fillings, surgical clips, a retained bullet — anything dense and metallic attenuates the beam so brutally that, along some angles, essentially zero photons make it through. The scanner is now reconstructing an image from missing data, and missing data plus an unforgiving algorithm equals bright and dark streaks fanning out like a star, often swallowing everything nearby.
This is partly extreme beam hardening, partly photon starvation, partly the reconstruction throwing up its hands. The practical pain is the same: the anatomy next to the metal — exactly where you usually need to look, like the bladder beside a hip replacement — gets erased.
Don't call a periprosthetic abscess, fracture, or tumor based on the dark zone hugging metal — that's streak, not pathology. And don't declare the area "normal" either; the artifact can bury a real finding. The honest read is often "obscured by hardware artifact," plus a plan to image around it.
The fixes are real, though. Metal artifact reduction (MAR) algorithms interpolate over the corrupted data, dual-energy/spectral CT can synthesize higher-energy images that punch through metal better, and simply widening the window often reveals more than you'd expect.
Motion: the patient who wouldn't hold still
CT acquires its data over a slice of time, and it trusts that the patient politely froze. Real patients breathe, swallow, have beating hearts, and occasionally sneeze mid-scan. When the anatomy moves while the scanner is sampling it, the reconstruction smears it into blurring, ghosting (a faint duplicate), or streaks off sharp edges.
The tell is that motion artifact tracks with things that move: a doubled diaphragm, a blurred bowel loop, stairstep misregistration where slices don't line up. On a chest CT, breathing turns crisp vessels into fuzzy comet tails. On cardiac CT, a heartbeat at the wrong instant can blur a coronary into looking stenotic — pure fiction.
Before you diagnose anything subtle, ask whether the finding moves with the motion. If a "ground-glass" haze sits only where the diaphragm blurred, it's probably respiratory motion, not disease. The cleanest fix is prevention: faster scans, breath-hold coaching, and ECG-gating for the heart.
How to keep artifacts from fooling you
Here's the mental checklist I run when something looks weird:
| Artifact | Signature look | Quick tell | Common fix |
|---|---|---|---|
| Beam hardening | Cupping; dark streaks between dense bone | Respects bone geometry, not anatomy or vessels | Correction algorithms; ignore if classic |
| Metal | Starburst of bright/dark streaks | Centered on a dense metallic object | MAR, dual-energy, wider window |
| Motion | Blurring, ghosting, stairstep | Tracks with something that moves | Re-scan, breath-hold, gating |
The single habit that saves you: when a finding doesn't respect anatomy — it ignores organ borders, crosses vascular territories, or sits suspiciously next to bone or metal — treat the scanner as a suspect, not a witness. Artifacts are reconstruction errors, so understanding how the image is built turns most of them from spooky to obvious.
The scanner will keep lying with total confidence. You just have to stop believing it on the days it's wrong.