Why does a radiograph show a double image?

A doubled structure on a radiograph has five causes, and they are not versions of one another. One of them is normal anatomy behaving exactly as the machine's geometry says it should. Two are faults at the receptor. The rest sit in between. What separates them is not how the second copy looks but where it sits — and the same two words, double image, name both the question and one of its five answers.

That last point is why the question stays live. In panoramic radiography, "double image" is a term of art with a specific meaning, sitting alongside "real image" and "ghost image" as one of three named categories a reader has to tell apart. Outside that literature it gets used loosely for anything that appears twice. The published vocabulary does not help much either: the same shadow has been printed under several different names, which means two clinicians can describe the same appearance and disagree about what it is called before they disagree about what it is.

What the question actually is

Start with the machine. A panoramic radiograph is made by a tube and a receptor rotating around the patient, forming a focal trough shaped like the dental arch. Structures outside that trough still reach the image. The result is a complex projection carrying multiple superimpositions and distortions, showing many anatomical structures that sit well outside the jaws — a point Susanne Perschbacher makes in her 2012 review of panoramic interpretation in the Australian Dental Journal.

Within that projection, three categories are named. Jong-Won Kim and Yo-Seob Seo put it plainly in Imaging Science in Dentistry in 2024: when interpreting panoramic radiographs, dentists must distinguish among real, double and ghost images. The three-way naming is old — it is the title of McDavid and colleagues' 1983 paper in Dentomaxillofacial Radiology, and Wakoh and Kuroyanagi were still working out the geometry of all three in 1991.

The naming is not tidy in practice. A 2014 review in Imaging Science in Dentistry lists the names in circulation for shadows cast by structures outside the focal trough: reverse shadows, secondary images, attached pictures, double images, triple pictures, contralateral images, ghosting images. So when a colleague says "there's a double image on this," find out which of the five things below they mean before you agree with them.

What are the causes of a doubled structure on a radiograph?

A double real image. This is the one that owns the term, and the one that is not a fault. In the 2015 case report by Valai-Kasim and colleagues in the Journal of Clinical and Experimental Dentistry, a double real image forms in a diamond-shaped zone in the central portion of the oral and maxillofacial region, where an object is intercepted twice by the beam. A structure sitting there is shown once in the midline and twice, symmetrically, on either side — a mirror pair. The structures that may produce them include the hard and soft palate, the epiglottis and a palatal torus. On machines whose centre of rotation travels outside the jaw, Kim and Seo report the cervical spine doing it too.

A ghost image. Different geometry, different position. A ghost image comes from an object lying between the X-ray source and the centre of rotation, and it lands on the opposite side, enlarged both horizontally and vertically, blurred, and higher than the object that cast it — without being left-right reversed. Kaugars and Collett called it what it is in 1987: a panoramic artifact. Lawful, predictable, and still not anatomy.

A double exposure. Here the geometry is innocent and the receptor is not. A double exposure lays one exposure over another, so what is doubled is the whole radiograph rather than one structure in it.

A phosphor plate that was not fully erased. The same entry covers the near neighbour. Elkhateeb and colleagues, auditing 50,000 intraoral radiographs in World Journal of Clinical Cases in 2022, describe double exposure in a plate arising either from partial erasing of the previous image or from failure of the image scanning. What shows through in that case is the previous radiograph, not a second copy of this one.

Anatomy that overlaps rather than duplicates. This is the commonest thing mistaken for duplication, and it is not duplication at all — one structure crosses another. In a 2003 audit of 460 panoramic radiographs taken in a dental school, the hyoid bone was superimposed on the mandible in 26.30% of them and the palatoglossal air space appeared in 46.30%; only 37.61% of the radiographs were free of any error. A forward head tilt is one reason the hyoid ends up over the anterior mandible, which puts this squarely among the positioning errors.

How does where the second copy sits tell you which one it is?

Make the coarse cut first: is the whole radiograph doubled, or one structure in it? If it is the whole thing, the cause is at the receptor and the geometry is irrelevant. If it is one structure, you are asking a question about the beam's path.

For that question there is a single organising idea, and it is the centre of rotation. Valai-Kasim and colleagues describe a real image as forming when the structure lies between the centre of rotation and the receptor. Sewerin's 1990 survey defines a ghost image as one representing structures located between the X-ray source and the centre of rotation. And Kim and Seo state the consequence directly: real and ghost images are divided by the rotation centre, and a double image forms in the region where that centre's path extends. So the object's position relative to that moving centre decides which of the three you are looking at, before anything about sharpness or size comes into it.

One caution on the geometry. Where a unit's centre-of-rotation path runs outside the jaw, the double-image zone extends with it, and Kim and Seo showed a single marked mandibular condyle producing three images at once — real, double and ghost. That was a single case reproduced on one machine, a Planmeca ProMax, so treat the extent of the zone as a property of the unit in front of you rather than a constant.

Which of them are real anatomy and which are faults?

The double real image is real anatomy, imaged twice because that is what the geometry does with a midline structure. Nothing went wrong. The ghost image is lawful in the same sense — its position is predictable — but it is an artifact, not anatomy, and its job is to be identified rather than diagnosed. A double exposure and an unerased plate are faults, and they are receptor-handling faults rather than anything to do with how the patient stood. Superimposed anatomy is real anatomy in the wrong place on the image, which is a positioning question.

If that sorting feels familiar, it is the same move as separating an artifact from a pathological finding anywhere else on a radiograph, and the same move that stops cervical burnout being read as caries. Both times the fix is a systematic search pattern rather than a sharper eye — you find the second copy, then you ask where it sits. The general case is covered in the radiographic artifact hub, and the ground-level habits in how to read dental radiographs.

Where it's unsettled

Three things are genuinely open, and none of them is the mechanism.

The vocabulary is the first. Beyond the list of alternative names already noted, the disagreement reaches the categories themselves: a 2023 study in the Journal of Oral Science on reducing cervical-vertebra shadows in the incisor region calls those shadows ghost images, where Kim and Seo treat cervical-spine duplication as a double image. Read the description, not the label.

The second is which anatomical structures belong on the double-image list. The retrievable primary literature names the hard and soft palate, the epiglottis and a palatal torus, with the cervical spine added on machines whose rotation centre travels wide. Other structures are named in teaching material without a primary behind them, and this entry does not repeat them.

The third is how far the geometry generalises. The zone's extent is machine-specific, and the papers that map it map one unit at a time.

Which need a retake?

There is no published rule of the form "a doubled structure needs another exposure." What there is, is a standard for judging the radiograph you already have. The European Commission's Radiation Protection 136 guidance grades image quality as Excellent (no faults), Acceptable (some faults but not affecting image interpretation), or Unacceptable — faults that leave the radiograph unsuitable for interpretation. It sets a minimum target of no more than 10% unacceptable, halving at each audit cycle, while noting that the 10% figure is a UK, non-evidence-based audit standard for primary dental care rather than a measured one.

Apply that and the five buckets sort themselves. A double real image is the geometry working, so the grading question is only ever about the rest of the radiograph. A ghost image needs identifying, not repeating. A double exposure or an unerased plate is judged against diagnostic acceptability like any other receptor fault, and the double exposure entry sets that standard out in full. Superimposed anatomy is a positioning question, and whether to repeat depends on what the overlap is hiding.

The reason not to reach for the retake button by reflex is in the same guidance: every radiographic examination must be justified on an individual patient basis, by showing the benefit to that patient outweighs the potential detriment. A repeat is a new exposure and needs its own reason. For scale, in that 50,000-radiograph audit 3,550 radiographs — 7.1% — were retaken because of an image artifact.