← Radiology

Guide 8 min read

Putting dental imaging dose in perspective

What a bitewing, a pano, and a CBCT actually cost in radiation, set next to a flight and a day of ordinary background. The honest version, with the numbers.

Patients ask me if the scan is safe, and referring dentists ask me how to answer them. Both deserve a real number rather than a reassuring shrug or a scary one. The good news is that the numbers exist, they are well studied, and once you have a few anchors in your head the whole conversation gets easier. The catch is that “safe” is not a single fact. It depends on the exam, the technique, and the patient in the chair.

The one anchor to memorize

Everyone on earth absorbs radiation constantly, from the ground, from cosmic rays, from the potassium in a banana. In the United States that natural background runs about 3,100 microsieverts a year, per the NCRP’s national survey, which works out to roughly eight or nine microsieverts a day. That daily number is the ruler. Once you can convert an exam into “days of just being alive,” the patient understands it, because everyone already accepts the background dose without thinking about it.

A useful travel anchor sits right next to it. A one-way flight across the country, New York to Los Angeles, delivers somewhere around twenty to fifty microsieverts of cosmic radiation, according to the EPA. So a transcontinental flight is a couple of days of background stacked into a few hours. Nobody cancels the trip over it.

Where the dental exams land

Against that ruler, routine dental imaging is small. A single intraoral radiograph taken with rectangular collimation and a modern digital or fast-film receptor is on the order of one microsievert, a fraction of a day of background. Four bitewings done the same way come to about five microsieverts, and a full mouth series lands near thirty-five microsieverts, roughly four days of background. Those figures are from Ludlow’s phantom study, which is the number set the ADA and FDA guidance draws on. A panoramic runs somewhere in the range of ten to twenty-five microsieverts depending on the unit, call it two or three days. A lateral cephalogram is about six.

Technique matters more than people expect. That same full mouth series jumps from about thirty-five microsieverts with rectangular collimation and a fast receptor to about a hundred seventy with round collimation, and near three hundred ninety with old D-speed film. So the difference between a modern setup and an outdated one is more than a tenfold swing in dose for the identical diagnostic exam. Rectangular collimation is the single biggest lever you control, and it is the cheapest.

CBCT is where the honest conversation gets more interesting, because the range is wide. Ludlow’s meta-analysis of dental cone beam units puts the average effective dose around eighty-four microsieverts for a small field of view, a hundred seventy-seven for a medium one, and two hundred twelve for a large craniofacial volume. So a small localized scan is about ten days of background, and a large one is closer to three or four weeks. But the full spread across machines and protocols runs from single digits to over a thousand microsieverts, which is the important part: two CBCT scans labeled the same way can differ by an order of magnitude depending on the field of view, the resolution, and whether the machine pulses its beam. This is exactly why choosing the smallest field of view that answers the question is not a technicality. It is the dose decision.

For a sense of scale at the top end, a medical head CT is on the order of one and a half to two millisieverts, so around fifteen hundred to two thousand microsieverts. Even a large dental CBCT usually sits well below that, which is one of the real advantages of cone beam for the questions it can actually answer.

The principle underneath the numbers

The rule the whole field runs on is justification first, then optimization. Justification means the scan has to do more good than harm before you take it, which is a clinical judgment made from the history and the exam, not a routine you run on a schedule. Optimization is the part people shorten to ALARA, as low as reasonably achievable. In imaging the more precise version is ALADA, as low as diagnostically acceptable, because dropping the dose so far that the image no longer answers the question helps no one. The pediatric CBCT group refined it further to ALADAIP, adding that the protocol should be indication-oriented and patient-specific, which is a mouthful that means the right dose for a small child asking one question is not the right dose for an adult asking another.

The through line is that “just in case” is not a justification. If you find yourself reaching for a bigger volume or an extra series because it might show something, that is usually a sign the clinical question has not been pinned down, not a sign more imaging is needed.

The pregnant patient

This is the one where fear does the most damage, so the numbers matter most. The dose to a fetus from any dental radiograph is negligible, effectively zero, because the beam is aimed at the head and the abdomen is far outside it. To put a ceiling on the worry, the obstetric literature places the threshold for fetal harm from radiation around fifty milligray, which is fifty thousand microsieverts, and ACOG states plainly that no single diagnostic imaging exam comes anywhere near a dose that threatens a pregnancy. A dental exam is thousands of times below that line. The ADA, echoing ACOG, affirms that necessary dental radiography is safe in pregnancy when it is clinically justified.

There is a related shift worth knowing. In 2023 the AAOMR recommended against the routine lead apron and thyroid collar for dental imaging, including for pregnant patients. The reasoning is that the shield does not block the internal scatter that actually delivers what little dose reaches distant organs, and it can get in the way of the beam or the exposure controls and force a retake, which adds dose. Telling a nervous patient you are skipping the apron can feel backwards, so it helps to have the reason ready: the apron was never doing much, and it occasionally hurt.

Children, and not overstating the risk either

Children genuinely are more radiosensitive than adults. They have more dividing cells and more years ahead for a stochastic effect to appear, and the ADA’s dental radiation guidance notes that the estimated cancer risk from a given dose is roughly double for a ten-year-old compared with a thirty-year-old. That is a real reason to be stingy with pediatric imaging and to size every exam to the child.

It is also a reason to be precise rather than alarmed. The risk model behind these estimates, the linear no-threshold model from the BEIR VII report, assumes that even the smallest dose carries some proportional risk, and it is the right conservative basis for setting protection policy. What it is not built for is predicting one person’s risk from one tiny exposure. BEIR VII itself says its low-dose estimates are uncertain enough that they could be off by a factor of two or three, and the Health Physics Society cautions specifically against taking a minuscule dose and multiplying it out as if it were a precise prediction of harm. So the correct posture with a child is neither dismissive nor fearful. Justify the exam, make it as small as the question allows, and do not turn a few days of background-equivalent dose into a reason for a frightened conversation.

What to actually tell people

When a patient asks if the scan is safe, give them the anchor. A checkup set of images is a few days of the background radiation they already live in. A small CBCT is a week or two of it. A large one is a few weeks, still less than a single medical head CT. The exam is worth taking when it will change what you do for them, and it is not worth taking when it will not, which is the same standard you would want applied to yourself. That answer is honest in both directions, and honesty is what makes it reassuring.

Sources

Still have a question?

Ask me directly.

This page answers the version I get asked most. If your case is the exception, the inbox is faster than the website.

Get in touch →