How to Read an AI-Assisted Dental Report

Reports differ between services, but most AI-assisted dental reports contain some of the following:

• Your image with overlays — coloured outlines showing where the software is pointing.
• A tooth chart — a diagram of your teeth with markers on the teeth that have observations.
• A list of observations — each with a tooth number, a type and sometimes a short description.
• A summary in plain language.
• A review status — whether a dentist has looked at the observations yet.
• A next step — usually to discuss the report with a dentist or book an examination.

Most reports outside North America number teeth with the FDI two-digit notation, standardised internationally as ISO 3950. It is simple once you know the pattern:

• The first digit is the quarter of the mouth (quadrant), seen from your own point of view: 1 upper right, 2 upper left, 3 lower left, 4 lower right.
• The second digit counts from the middle of the mouth backwards: 1 central incisor, 3 canine, 6 first molar, 8 wisdom tooth.

So 36 is the lower left first molar and 18, 28, 38, 48 are the wisdom teeth. Baby teeth use quadrants 5–8 (for example 75). X-rays are usually displayed as if you are facing the patient, so your left side appears on the right of the screen — another reason never to mirror an image before uploading it.

• Existing dental work — fillings, crowns, bridges, root canal fillings or implants the software recognised. Useful for describing your mouth; not a problem in itself.
• Possible finding — an area that resembles something that may deserve professional attention, such as possible decay or a change around a root tip. It is a candidate for review, not a confirmed condition.
• Nothing marked — the software did not flag anything on that tooth. That does not prove the tooth is healthy; software can miss things, and many problems are only visible in the mouth.

Careful reports use cautious words such as “possible”, “may” and “area to discuss”. Be sceptical of any report that names a disease or a treatment as certain from an image alone. Our guide to AI-assisted X-ray analysis explains why.

Some reports show a confidence value next to each observation. It reflects how strongly that part of the image matched patterns the model learned. It is not the chance that you have a problem, not a measure of how serious it is and not a diagnosis. Machine-learning research also shows that such scores can sound more certain than the model's real error rate unless they are carefully calibrated.

A priority or urgency label is a different thing. It is meaningful when a dentist sets it after reviewing your case, because it reflects clinical judgement. Treat any automatic priority with caution, and ask the dentist how urgent something really is.

FDI World Dental Federation's policy is clear that AI should assist dentists and that responsibility for diagnosis and treatment stays with the human. A report therefore moves through stages:

1. AI observations only — preliminary information, nobody has reviewed it yet.
2. Reviewed by a dentist — each observation may be confirmed, rejected or supplemented with something the software missed.
3. Examined — after an examination, the dentist can make a diagnosis and discuss a treatment plan with you.

If the report does not say whether a dentist has reviewed it, assume it has not.

• Which observations do you agree with, and which not?
• Is there anything the image cannot show that we should check?
• Do we need another image of a particular tooth, or is the examination enough?
• How urgent is this, and what happens if we wait?
• What are the options, including no treatment, and what does each involve?

If you have not yet chosen a dentist, a digital pre-assessment can help you prepare these questions before the first visit.