Which normal structures are mistaken for pathology on a radiograph?

Radiology has a name for this problem. A normal variant is an incidental, usually asymptomatic imaging finding that mimics a true pathological condition, and knowing the common ones is what stops a reader ordering imaging or treatment the patient did not need. [1] The reason it keeps happening is structural rather than careless: intraoral and panoramic radiographs are two-dimensional, so overlapping structures and displacement of the tomographic layer can confuse a trained radiologist, never mind someone three weeks into their first clinic. [1]

What the literature does not give you is a ranking. There is no study that orders normal anatomical structures by how often they get called as pathology. That absence is not a detail to work around — it changes what an honest page can claim, and it is why the roll call below is organised the way it is.

What the question actually is

This is a companion question to is this shadow real?, which asks whether a radiolucency is genuine at all — whether the beam or your own eye put it there. That page owns the triage test, and it is a different test from this one. Once you know a radiolucency is real, the next question is whether it is a named anatomical structure rather than a lesion. That is the question here, and it is answered by location, not by density.

So this page does not cover cervical burnout, the Mach band effect, or anything else the beam or the visual system manufactures. It covers real bone, real canals, real air, sitting where a lesion would sit.

It is worth being clear about the base rate before the roll call, because the roll call can leave the wrong impression. Most periapical radiolucencies are what they appear to be. A 2023 review reports that non-endodontic lesions account for somewhere between 0.7% and 5.0% of all periapical biopsies — a range, because it varies with the population studied. [2] That range counts pathological mimics, not normal anatomy. Anatomy is a differential you rule in or out, not a default.

What the evidence shows

Five landmarks below, each with the positional cue that identifies it.

Read the order as a teaching order, not a frequency ranking. It runs by region so you can find the entry matching where you are looking, and it starts with the landmark whose position is best documented in the literature. No source ranks these against each other for how often they are misread, and none of what follows should be read as a claim that one is misread more than another.

A sixth landmark, the incisive fossa of the maxilla, was planned for this page and has been left out. Searching for position data, morphometry or a reported case returned nothing that met the sourcing standard used here, and an unsourced entry in a list like this is worse than a shorter list.

Mental foramen — mandibular premolar region

The mental foramen is the anterior opening of the mandibular canal, and on a periapical radiograph of the lower premolars it can project as a round radiolucency at or near a root apex.

Its position is the best documented of the five, and the documentation is mostly a warning about position. In a study of Moroccan panoramic radiographs — 794 screened, 377 meeting inclusion criteria for at least one side — the foramen sat just below the apex of the second premolar in 62.7% of patients and between the first and second premolars in 30%, and was symmetrically placed in 79%. [3] A separate anthropometric study of dry mandibles found the modal position differed by population: in line with the long axis of the second premolar in Chinese mandibles, between the first and second premolars in European and Indian mandibles. Its authors put it bluntly — the notion that the mental foramen lies between the first and second premolars can no longer be accepted. [4]

The cue is not a fixed tooth. It is continuity: the foramen is the terminus of the mandibular canal, and the canal usually leads your eye to it. A periapical lesion has no canal running into it.

Nasopalatine canal — maxillary midline

The nasopalatine canal runs in the midline of the maxilla, from the incisive foramen on the palate up to the nasal floor, between and above the roots of the central incisors. On a periapical of the upper anteriors it projects as a radiolucency in the midline.

It is small and it is variable. Across 200 adult cone-beam scans, the incisive foramen had a mean diameter of 4.49 ± 1.71 mm and the canal a mean length of 11.15 ± 2.87 mm. [5] A smaller study of 50 cone-beam scans in a Sri Lankan population reported a mean canal diameter of 3.692 mm with a range of 2 to 6 mm, and its authors said the difference from published figures may be due to ethnicity or to their sample size. [6] Take that as a description of normal spread, not a threshold: neither study defines a diameter above which the canal stops being normal, and no such cut-off is quoted here because none was retrievable.

The canal also does not behave like one clean tube. At the nasal floor it most often opens through two apertures, and three or four openings occur. [5]

The cue is the midline. A radiolucency sitting squarely between the central incisor roots, with both teeth's lamina dura intact, is the canal until something says otherwise. The nasopalatine duct cyst is the look-alike here, and it is a genuine lesion rather than an anatomical variant — it is carried in the differential on palatoglossal air space and is not restated here.

Maxillary sinus alveolar recess — maxillary posterior region

The maxillary sinus can extend downwards between the roots of the posterior teeth, and that extension — the alveolar recess — can present on a radiograph as a well-defined radiolucency sitting on a root apex.

This is the one landmark on the list with a retrieved source about the misreading itself, and that source's own finding is how thin the ground is. A 2026 systematic literature search for enlarged maxillary sinuses with alveolar recesses resembling a periapical cyst found four articles, and its authors concluded that knowledge gaps about the alveolar recess remain in the literature and should be addressed to avoid misdiagnosis. [7] Treat the mimic as real and the evidence for it as early.

What is better documented is the underlying geometry, which is what makes the recess reach the roots at all. Using the Jung classification on 110 cone-beam scans, the sinus floor sat above the root tip most often at the first and second premolars, while apical protrusion over the sinus floor was the most common pattern at the first and second molars. [8] A second cone-beam study, of 60 scans and 1,066 roots, found the mesiobuccal root of the second molar closest to the sinus floor at a mean distance of 0.44 ± 3.05 mm, followed by the palatal root of the first molar at 1.58 ± 4.01 mm, with the buccal roots of the premolars furthest away. [9] Note how wide those standard deviations are — the spread is the finding, not the mean.

The cue is the border. The sinus floor is a continuous corticated line you can trace past the tooth you are worried about and on into the rest of the sinus. A periapical lesion's border stops at the lesion.

Submandibular fossa — mandibular molar region

The submandibular fossa is the concavity on the lingual surface of the mandible below the mylohyoid line, where the submandibular gland sits. Less bone in the beam's path means a radiolucent zone under the molar roots.

It is the usual anatomy, not an oddity. In 100 spiral CT examinations of patients being assessed for lower-jaw implants, a lingual concavity of 2 mm or more was present in 80% of jaws: flat depressions under 2 mm in 20%, concavities of 2 to 3 mm in 52%, and concavities deeper than 3 mm in about 28%, up to a maximum depth of 6.6 mm. [10] That population is a surgical-planning one rather than a general dental one, so read the percentages as describing jaws being assessed for implants.

The cue is what the radiolucency lacks. The fossa is a thinning, so it fades into the surrounding bone with no corticated margin of its own, and it sits below the roots rather than around an apex.

A Stafne bone cavity is a different thing, and worth separating deliberately, because the two get spoken about together. It is a discrete depression in the lingual surface of the mandible, and a meta-analysis of 54 studies covering 355,890 individuals put its pooled prevalence at 0.17% (CI 0.14–0.21%), four times more common in males and unilateral in almost all cases; radiographic studies alone returned a lower figure of 0.12% (CI 0.09–0.15%). [11] A concavity is common; a discrete corticated cavity is not.

Nutrient canals — mandibular anterior region

Nutrient canals carry neurovascular bundles through bone, and they are best seen on mandibular anterior periapical radiographs, where they appear as fine radiolucent lines. In one cross-sectional sample, 72.4% of the canals seen were beyond the root apex. [12]

The cue is shape and direction. A canal is linear and runs through bone; a periapical lesion is a shape centred on an apex. This is a landmark where an orderly search pattern does most of the work, because the difference is obvious once you are looking at direction rather than density.

Where it's unsettled

Nobody has ranked these. Searches for a study ordering normal anatomical structures by how often they are misinterpreted as pathology returned nothing across two literature channels. Position data exists for each landmark; misreading data essentially does not, beyond the four articles that one systematic search found for the maxillary sinus recess. [7] So a page can honestly tell you where each structure sits. It cannot honestly tell you which one catches people out most, and any page that does is telling you something the literature has not established.

Nutrient canals are the clearest genuine disagreement. Whether visible canals mean anything systemic has been argued for decades. A cross-sectional study of 200 patients found them more often in diabetes (84%), hypertension (66%) and periodontitis (52%). [12] A controlled study of 32 hypertensive and 32 normotensive subjects found them in 37.5% of hypertensive and 53.1% of normotensive patients, a difference that was not statistically significant (P = 0.209), and its authors noted that the normalcy of nutrient canals is itself controversial among investigators. [13] A study of 139 normotensive and 76 hypertensive edentulous patients found no statistically significant difference between the groups. [14] The largest of the four, 1,040 patients, reported only that occurrence seemed related to age, edentulous mandible, periodontal problems and high blood pressure. [15] The studies differ in sample size, in dentate status and in the direction of the effect — and since edentulism itself seemed related to canal occurrence in the largest of them, an all-edentulous sample cannot settle the hypertension question. For reading a radiograph, the practical answer is that a visible nutrient canal is a canal, and what it implies systemically is unresolved.

Position varies by population. The mental foramen work is the clearest case, and the finding is that a single textbook position does not hold across populations. [4] That is a reason to use continuity with the mandibular canal as the cue rather than a tooth number.

What this means at the chairside

Read the radiograph second. ADA and FDA guidance is explicit that the recommendations are not substitutes for clinical examinations and health histories, and that a thorough clinical examination, patient history, review of prior radiographs and caries risk assessment should precede radiographic examination. [16] The same document says radiographs should only be ordered when the additional diagnostic information is expected to affect patient care. [16] A suspicious shadow is a reason to examine the patient, not a reason to reach straight for another exposure.

Correlate rather than rely on any single test. The 2023 review's instruction for the anterior maxilla generalises: vitality testing should be performed and correlated with radiographic findings, and clinicians must correlate the radiographic and clinical presentation of every case. [2] Correlation is the discriminator, not the pulp test on its own.

And know what a negative pulp test does not prove. A critical review of nonodontogenic periapical radiolucencies found that a lack of pulp response was registered in 68.4% of them, and that swelling was a non-contributory feature for distinguishing periapical lesions. [17] That figure is about pathological mimics rather than about normal anatomy — but it makes the point in the direction that matters. A non-responding pulp does not confirm an endodontic cause, and the reverse case is the useful one for this page: a responsive pulp with an intact lamina dura and an intact alveolar crest alongside a radiolucency in a named anatomical position is a strong argument for anatomy.

When a second view is worth taking. Two-dimensional projection is what created the ambiguity, so the useful second view is one that changes the projection rather than repeating it — which is a different decision from retaking a radiograph that failed on quality, and part of why a periapical and a bitewing answer different questions. Where a suspected anatomical variant still cannot be resolved on plain radiographs, the authors who studied the maxillary sinus mimic recommend three-dimensional imaging such as CT or CBCT for accurate detection. [7] That is those authors' recommendation from a thin evidence base, not a standard of care, and it sits under the guidance above: the additional imaging has to be expected to change what you do. [16]

When to refer. The honest answer this page can give is a negative one. No source retrieved for it defines a referral threshold for a suspicious anatomical landmark. What the evidence supports is the sequence: examine, correlate, and escalate the imaging only when the extra information will change management — and treat a radiolucency you cannot place anatomically as a lesion until it is placed.

Knowing where the mental foramen sits is one thing. Calling it correctly on a real radiograph, at speed, with a patient in the chair, is another. Try it on real radiographs and find out which way you err.