Radiographic artifact

A radiographic artifact is a feature on a radiograph that comes from how the image was made rather than from the patient — from the operator, the receptor, the scanner or the software. It is neither anatomy nor disease, but it can imitate both, and separating it from a real finding is part of reading any radiograph.

Synonyms: radiographic artefact; dental radiograph artifact; image artifact; radiographic error

What is a radiographic artifact, and how is it different from a technique error?

You may have been taught that an error is the mistake someone made taking the radiograph and an artifact is the mark it left behind. It is a tidy distinction, and the classification studies do not hold it.

The two papers that set out to classify these things put them in one system. Çalışkan and Sumer catalogued 34 image error and artefact types from 2,100 intraoral radiographs and sorted them into four groups according to the causative factors. Deniz and Kaya reviewed 5,000 periapical images, found artifacts in 1,822 of them, and divided the causes into six groups and 45 subheadings. Both sort the whole set by what caused it.

That is the axis worth learning. At the light box the question is never "is this an artifact or an error." It is: what made this mark, and could that thing have been inside the patient?

What are the main classes of artifact, and what causes each?

Deniz and Kaya's six groups are operator errors, superposition of undesirable structures, ambient light errors, plate artifacts (physical deformations and contamination), scanner artifacts, and software artifacts.

Operator errors were the commonest group in that series, at 34.7%, and projection geometry errors were the commonest problem inside it, at 31.9%. The projection errors they list are shape distortion — foreshortening, elongation, image lengthening and distortion — horizontal angulation artifacts producing overlapping, cone-cutting errors where crowns or apices are cut off, and cone vertical level errors where crowns or apices don't show at all.

That group is the one you meet most often. In a separate two-year record of 50,000 intraoral radiographs from one dental school, 3,550 — 7.1% — were retaken because of an image artifact, and cone cut was the commonest operator error at 988 images.

The rest come from the equipment and the software, and each has a look:

Those studies were done on photostimulable phosphor plates, so the receptor-specific rows describe plate systems. The shape of the taxonomy holds wherever the image is made: something in the chain between the beam and the screen can put a mark on the image, and each link fails in its own way. Panoramic radiographs add their own layer — osseous structures, soft tissues, air spaces and ghost shadows all contribute to the final image, and the superimpositions and distortions built into the projection get worse when acquisition goes wrong.

How do you tell an artifact from a real finding?

Work from the cause. An artifact was made by the equipment, so it behaves like the equipment and not like a mouth.

It repeats. A defect in the receptor or in the software produces the same mark in the same place on consecutive images. Deniz and Kaya describe exactly this with a software fault: the system flagged an edge-masking defect for consecutive images that had a cone-cut appearance in the same region. If the same feature lands in the same spot on consecutive radiographs, it is not in the patient.

It answers to the equipment. The same authors advise checking plates daily for scratches, taking a test image when you find one, and replacing the plate if you need to. That test image is a diagnostic step as much as a maintenance one — if the mark survives with no patient in front of the beam, the question is settled.

It doesn't have to agree with the mouth. The 2025 meta-analysis on radiographic optical effects concludes that clinicians should interpret radiographic findings with caution and always correlate them with a thorough clinical examination. A radiolucency you cannot corroborate is a radiolucency you have not yet explained.

And read the whole image the same way every time. Perschbacher recommends a systematic and repeated approach to examining panoramic radiographs precisely so that critical findings are not overlooked — the same search pattern that stops you missing a lesion stops you inventing one.

One honest limit, from the people who did the counting: on retrospective review, Deniz and Kaya could not securely tell a defective edge-masking artifact from a genuine cone cut. The rules narrow the field. They do not always close it, and a second radiograph is not an admission of failure.

What it's confused with

Which artifacts can mimic pathology?

The best-quantified example is the optical one. A 2025 systematic review and meta-analysis of the Mach band effect and triangular-shaped radiolucencies found non-carious triangular-shaped radiolucencies on 26.44% of maxillary molars (270 of 1021), and calculated that these effects led to false positive diagnoses of caries or fractures in approximately 13% of observations (60 of 464). Read those figures with the authors' own caution: only five of 640 identified reports met inclusion, heterogeneity was high at I² above 90%, and they graded the certainty of the evidence low to very low. The direction is clear and the magnitude is not.

Processing can do it too. In an in-vitro study on three digital sensors and a quality-assurance phantom, sharpening with unsharp masking created overshoot artifacts that, in the authors' words, can be misdiagnosed as pathosis or prosthesis misfit — and the artifacts grew as the sharpening increased. Nothing changed in the patient between the raw image and the sharpened one.

And an artifact can imitate another artifact. The defective edge mask above resembles a cone cut or a metal artifact, which is worth knowing before you blame the person who took the radiograph.

Is superimposed normal anatomy an artifact?

No, and the distinction is worth holding onto.

The interpretation literature treats variations of anatomy, artefacts and disease as three separate categories: Perschbacher presents challenging examples across all three, and the paediatric panoramic literature spends its effort on how to differentiate a normal structure from an underlying pathologic condition. A mental foramen sitting over an apex is not an artifact. It is a real structure that the projection has put somewhere unhelpful.

What does belong in the artifact taxonomy is superimposition of objects: Deniz and Kaya's second group is "superposition of undesirable structures", sitting alongside scanner and software faults. The difference is not subtle once you have it. An object can be removed before the retake. Anatomy cannot, and the answer there is a different angle, not a different patient.

Common errors

• Treating a radiolucency as a diagnosis. It is a density. The meta-analysis above exists because a common optical effect gets called caries.
• Assuming the mark is in the patient. The first test is cheap: does it repeat?
• Sharpening, then trusting the edge. More sharpening buys visible crispness and more overshoot at the same time.
• Blaming the operator by reflex. Some of what looks like a positioning error was made by the scanner or the software after the beam had already done its work.