Ghost image

A ghost image is a radiopacity on a panoramic radiograph cast by an object that sat between the X-ray source and the machine's centre of rotation. It appears on the opposite side of the image from the real object, enlarged, and not mirror-reversed. Earrings, the cervical spine and the mandibular rami all cast them.

Synonyms: ghost shadow; panoramic ghost image; contralateral image

What is a ghost image?

A ghost is defined by where the object was, not by what the object is. When something lies between the X-ray source and the machine's centre of rotation during the scan, the beam sweeps past it and writes it onto the image a second time, somewhere it never was [1].

That makes it a radiographic artifact with an unusual property: it is lawful. Most artifacts tell you a technique went wrong. A ghost tells you where something was standing. The panoramic image is already a complex projection carrying multiple superimpositions and distortions, which is exactly why a systematic reading approach is recommended for it [2] — and a ghost is one of the superimpositions you can name and account for rather than puzzle over.

What it looks like

Three properties travel together, and you can check all three at the screen.

1. It's on the wrong side.

   The ghost lands on the side of the image opposite the object that made it. Three of the sources below reach for the same word — contralateral [3][4][5].

2. It's bigger, and it's not a mirror.

   In a 2024 case that paired a patient's radiograph with a phantom simulation, the ghosts of a pair of earrings were enlarged relative to the real image and were not reversed left-to-right [6]. Enlargement without inversion is what separates a ghost from a genuine second object.

3. It usually has a partner.

   The object that cast the ghost is often still on the radiograph as a real image. In one reported case a pair of earrings produced two real images and three ghosts [7]; in the 2024 case each side carried one real image and two ghosts [6]. So when you find a suspicious opacity, look across the midline and slightly around it for the thing that made it. Finding the partner settles the question faster than anything else on this page.

Why does it appear on the opposite side, and enlarged?

Because of where the beam was when it caught the object. A panoramic machine rotates the source around the patient, and anything sitting between that source and the centre of rotation gets projected across to the far side of the image [1].

There is a boundary to this. The three-dimensional zone an object has to occupy to cast a ghost is called the ghost envelope, and it has been mapped for individual machines with a lead sphere suspended around a dry mandible [8][9]. An object outside the envelope produces no ghost at all, and an object inside it produces a characteristic ghost according to where it sits [8]. The consequence is practical: the presence or absence of a ghost is information about where the object is. The envelope's exact shape differs between machines, so the rule generalises but the geometry doesn't.

Size plays into it too. Larger radiopaque objects appear to ghost more readily [5].

Which objects and anatomical structures cast ghosts?

Both, and the second half is the one that gets forgotten.

Objects. Jewellery leads — earrings, piercings, necklaces — along with hearing aids, dental crowns, wires, clothing, protective equipment, medical devices and machine parts [3][4]. One narrative review catalogued 37 distinct types of unexpected metallic foreign body appearing on panoramic radiographs, grouped as jewellery, clothing, personal protective equipment, medical devices, iatrogenic foreign bodies and rare incidental findings [4].

Anatomy. Ghosts are commonly cast by anatomy that cannot be removed — the vertebrae, the mandibular rami and the hyoid bone [3]. The cervical vertebrae characteristically ghost into the upper incisor region — two phantom studies were built specifically around reducing that [10][11]. And part of the upper hard palate line is itself a ghost: an in-vitro study of three dry skulls found most of that line is real, but a small part is a ghost of the palatine bones and the posterior palatine processes [12].

This is the half the technique literature skips. You can ask a patient to take their earrings out. You cannot ask them to take their spine out.

What it's confused with

How to be sure

Start with the partner. Look across the midline for the real image of whatever you're looking at [6][7].

Then ask whether the opacity sits where a ghost from that machine would land [8]. Midline and palatal structures deserve extra suspicion: in one reported case a paramedian supernumerary tooth threw multiple ghosts and pushed the clinicians toward an inaccurate diagnosis, and CBCT was needed to establish what was actually there [14].

Where the question stays open, cross-sectional imaging answers it. That is what resolved both the sialolith and the tonsillolith [5], and the authors of that report note plainly that telling a real image from a ghost artifact may avoid unnecessary surgical intervention [5]. That is the stake here — not a retake, a scalpel.

Prevention is partly in your hands and partly not. Radiopaque foreign materials on a panoramic radiograph are often relatively easy to recognise as artifacts from their location, and they are avoidable in most cases [4].

Common errors

Treating a ghost as a preparation failure and stopping there. Removing metal is worth doing, but the vertebrae, rami and hyoid ghost regardless [3]. An entry that ends at "take the earrings out" leaves you with no account of the ghosts that are there regardless.

Assuming it's rare knowledge you'll pick up later. In a written examination covering 28 panoramic regions, 119 dental students scored 53% correct overall — and ghosts of the contralateral nasopalatine line and of the cervical vertebrae were among the regions they scored worst on [15]. This is a known blind spot, not an exotic one.

Using "double image" as a synonym. The literature applies several loose names to these distortions, including double image [3], while also drawing a technical distinction between real, double and ghost images [6][16]. If you mean the contralateral enlarged copy, say ghost image.

Reaching for "pseudo-ghost". You will meet the term. The authors of the 2024 case note the argument that a pseudo-ghost image is theoretically a real image and should therefore be called a second or double primary image rather than a ghost [6]. It is an unsettled naming question, not another thing to memorise.

An artifact you can name is an artifact that stops costing you. Ghosts sit alongside the other patterns worth recognising on sight, like the herringbone artifact from a reversed receptor or the elongation and foreshortening that come from vertical angulation — each one lawful, each one learnable.