July 15, 2019

Can we trust Cone Beam CT airway measurements?

Cone Beam CT is a great new technology that has the potential to revolutionise orthodontic treatment.  But is it accurate enough for measurement of the airway?

There is a large amount of interest in attempting to evaluate the effect of orthodontic treatment on the pharyngeal airway space. Currently, claims of increasing airway volume as part of treatment for disordered breathing are becoming common.  These treatments are often illustrated with before and after Cone Beam CT images showing impressive increases in airway volume. Unfortunately, we do not really know if this method of measurement is accurate.  This was the questions asked by these Brazilian based investigators. The AJO-DDO published the paper.

Cone-beam computed tomography airway measurements: Can we trust them?

Daniel Patrick Obelenis Ryan et al

(Am J Orthod Dentofacial Orthop 2019;156:53-60. https://doi.org/10.1016/j.ajodo.2018.07.024

What did they ask?

They did this study to answer this simple question:

“What is the variation in airway volume measured by consecutive CBCT images”?

What did they do?

They collected 27 sets of CBCT records from one surgeon. All the patients were having presurgical orthodontics. They took the CBCTs at diagnosis and following an interval of 4-6 months.  The second CBCT was taken to construct a surgical guide.

All the scans were done on the same iCAT machine with a standardised patient positioning protocol.

As a result, they obtained 2 CBCT scans for each patient taken after a period of 4-6 months.  All the patients had stopped growing and the orthodontic treatment was very unlikely to have changed the dimensions of the airway. Therefore, any differences in airway measurement between the sequential scans would be due to positioning or recording errors.

One trained investigator measured the total airway volume using a standardised protocol. They segmented the airway into the nasopharynx, oropharynx and hypopharynx and calculated the volumes.

They carried out an inter-observer reliability analysis for airway volume. Finally, they measured the agreement between the measurements for the sequential scans, using the Intra Class Correlation coefficient and non-parametric bootstrap techniques. They calculated the agreements and confidence intervals using Bland-Altman plots.

What did they find?

They presented a large amount of detailed data.  I thought that the most important findings were:

  • There was good operator agreement for all the variables.
  • There was low reproducibility for the airway volume measurements.

This table includes the average errors for the population

VariableMean95% CI
Nasopharynx (mm3)681.1472.5-896.3
Oropharynx (mm3)2248.11681.3-2851.6
Hyopharynx (mm3)517.3355.2-688.8

These values are rather large. Importantly, the 95% confidence intervals are very wide. This suggests that there is poor reproducibility for the airway measurement between the readings.

Their overall conclusion was:

“Different CBCT scans can result in significantly different pharyngeal space readings”.

What did I think?

I thought that this was a rather difficult paper to interpret. I hope that I have not simplified it too much. However, I would recommend that you read this paper if you can access it.  This is because  this could be a rather important piece of work.

I felt that the methodology was good. However, we need to consider that they obtained the 27 scans from an original sample of 278 scans.  This may have resulted in selection bias. Nevertheless, I found it hard to imagine that this could have influenced the results of the study.

They used an appropriate methodology and showed that the differences in the measurements were not due to measurement error. Importantly, these differences were not caused by growth, as the population were not growing. Furthermore, orthodontic treatment was very unlikely to result in airway changes.

As a result, the only cause of the marked differences must be the positioning of the patient or other factors such as  breathing and/or swallowing during the scan.

What does this mean?

I feel that this shows that there is a large amount of inaccuracy in airway measurement from Cone Beam CT scans. These findings are similar to an earlier study that I posted about. There are now several studies that reinforce this finding.  I think that this is important because of the many case reports suggesting that certain types of orthodontic treatment can influence the airway. I believe that we cannot make this conclusion because the errors within the current techniques are too great.

Related Posts

Have your say!

  1. From a functional point of view, it is more important to consider the oral functions. There are causes for restricted respiratory function spaces. One needs a competent lip closure in order for the respiratory function spaces to develop properly. At best, these measurements provide information on how quickly recurrences can occur if only mechanical treatment is used. It becomes interesting when myofunctional therapy is used to work on the causes. Then these measurements can be very useful because the recurrences will be significantly reduced, so that mechanically generated improvements can remain stable. Spontaneously tonic habits will not be lost even if the orthodontic result would allow useful tonic functions.

    Translated with http://www.DeepL.com/Translator

  2. Sleep Study is still the Gold Standard. In addition to the CBCT study, having the patients have a sleep study would be ideal. Not an easy task

  3. If these measurements were taken whether standing, laying down or sitting while awake, the study is really useless since the airway while sleeping can be totally different. The airway dynamics are completely different when sleeping than when awake. Whoever did this study should have known this if they were knowledgeable about airway dynamics. Also left out of the equation are the dynamics of tongue position, swallowing pattern, nasal vs. mouth breather, lip seal, head position, etc. This is a simplistic study hoping to prove something with only a portion of the information needed.

    • Martin Denbar,
      Your points about what would make a scan more diagnostic are correct. That was not, however, the point of the study. The point of the study was to take scans how most docs take scans, and assess whether the outcome (cross-section) is reliable and reproducible. It is not.

  4. Creating a “standard” positioning regimen would be very helpful, I feel. Something as simple as tongue position can have a huge effect on CBCT airway measurements. Lip seal, bite position- on and on. Can the AAO or perhaps the Oral Radiologists provide some guidance on this important issue?

  5. I agree with both Dr Matza and Dr Denbar. We should never make any claims about upper airway changes without the corresponding sleep studies. A couple of other points; since the nasopharynx, including the maxillary sinuses are largely enclosed by bony architecture, upper airway changes in these locations may be less dependent on positioning and the phase of the respiratory cycle. Our initial studies appear to support this contention, and will be presented at the World Sleep Society and similar meetings this year and next. I reviewed some studies on upper airway changes, and altho’ no meta-analysis was performed, the review indicated a possible mechanism for non-surgical upper airway remodeling.
    Singh GD. Physiologic remodeling of the upper airway: Pneumopedics. J Sleep Disord Treatment Care 7:3, 2018.

  6. Interesting comments so far. Lots of redirection, nit picking about position, sleep studies, measuring different areas etc, etc. etc. You all are missing the point. This study says that CBCT is probably the wrong tool to measure airways! Lying down, standing, tongue forward or back, Monday or Thursday, it does not matter! It is the wrong tool! Like trying to determine temperature with a tape measure.
    Don’t take it personally, just look for a tool that actually measures what you are claiming. CBCT ain’t it.

    • John,you hit the nail on the head. What you consider to be nit picking is the issue. What I mentioned are significant contributors to the issue. Similar to the dentist using oral appliances to treat OSA without realizing a patient’s therapeutic range can be .25 mm. Small increments impact the total and to have these items categorized as nit picking goes to the heart of the matter. I have never taken it personally, yes it is the wrong tool, but then ask yourself this. Why was the wrong tool used in the first place for this study? Other studies have shown the ineffectiveness already. It can only be from an inherent lack of knowledge of the airway which appears to be pervasive in the field of ortho. and dentistry in general.

  7. From Kevin: “I think that this is important because of the many case reports suggesting that certain types of orthodontic treatment can influence the airway.”

    In 1954 anesthesiologists and respiratory researchers James Elam, MD (Johns Hopkins) and Peter Safar, MD (Yale) demonstrated the importance of forward positioning of the mandible to increase the airway. This would later become a significant component of rescue breathing. With this in mind, I think that a fair assumption is that orthodontic treatment that demonstrates some forward movement of the mandible will increase the airway. Granted this will be on a much smaller scale than airway opening for rescue breathing and is very difficult to quantify as some investigations that Kevin has cited have shown. But in the end it is the same model. It seems to me that certain types of orthodontic treatment certainly can influence the airway and this can be demonstrated without relying on CBCT radiography. Yes? Am I missing something?

    Some interesting facts about doctors Elam and Safar who had remarkable careers with many awards including three Nobel nominations in medicine for Dr. Safar.

    • Barry Winnick, I say this with all due respect, displacing the ENTIRE mandible anterior to create a “forward position” to increase the airway is in no way related to either the orthodontic or dental facial orthopedic aspect of mesial movement of the mandibular dentition effected by orthodontic treatment. To suggest otherwise demonstrates a profound lack of understanding of anatomy and orthodontics. The only thing those two concepts have in common is the word “mandible”. It is not the same model, not even close. Unless you finish your case with the condyle well out of the fossa and on to the eminence, you are nowhere near a forward positioned mandible. If this is the type of logic that drives the airway orthodontic phenomenon, I am speechless at the foolishness of this thought pattern.

      • I certainly do not subscribe to “airway orthodontics” whatever that really means.

        Assuming Elam and Safar’s work is valid, if you demonstrate mandibular movement/repositioning, however small, in a forward/downward direction beyond normal growth, I think that one can expect, however small, an increase in airway space.

  8. The nasal passages must be used at sufficient frequencies so that the mucous membranes are healthy and free of inflammation. Systemically, sufficient respiratory function spaces and the respiratory technique (function) must be regarded as a unit. Orthodontics does not solve a problem if it orients the lower jaw forward and ignores functional causes. I think Mr. Berry Winnick actually missed something. Interesting facts can be found at Buteyko, he has exhaustively dealt with the respiratory function.

    Translated with http://www.DeepL.com/Translator

  9. Dear Kevin

    Your August 30 blog about the Brazilian study is disappointing. You seem to be more concerned with criticising the methodology than you are with the results. The only figures you quote are the errors and variables; there is no mention of the actual pre- and post-treatment pharyngeal measurements T0 and T1, figures which would indicate to the reader whether or not the pharyngeal volume actually changed following orthodontic treatment – something which you clearly do not believe can happen as you state: “ . . . . the orthodontic treatment was very unlikely to have changed the dimensions of the airway”.

    There are many orthodontists who would disagree with you.

    Of course there will be variables and ‘errors’; the patients are human beings and will move constantly in spite of head-holders etc. They are after all alive! They are not architectural or engineered structures.

    Another possible source for variables is the type of pre-surgery orthodontic treatment used in each case; was it expansion/orthopaedic or extraction/retraction? No mention is made of this vital aspect in your blog.

    Although you state “the table includes the average errors for the population”, the source of the measurements in the column headed “95% CI” is not clear; if they include or are related to T0 and T1, then they suggest a significant change in each segment following the orthodontic treatment: naso-pharynx +89%, oro-pharynx +69%, hypo-pharynx +93%.

    What a shame the study subjects were not receiving orthodontic treatment for non-surgical reasons, especially those with a Class II malocclusion.

    Noel Stimson

    • I don’t think Kevin is off base here, and it’s not about the efficacy of treatment. A lot of these studies have been published without any real controls. Oh, they put in things that they call “controls”, but if you look closely all they are for the most part are observer consistency verifications. I’ve looked far and wide for studies that showed that a volumetric analysis of an untreated individual can be reproduced at different time points. You kinda need this as a foundation if you’re going to do analyses on treated individuals, otherwise statements that your treatment caused the changes are scientifically invalid.
      This is the FIRST time that I’ve personally seen anyone actually look at this reproducibility, and I’m not at all surprised to see it lacking. I’ve seen this in my own patients over the last dozen years of clinical CBCTs. A decade ago in speaking with an ENT colleague, she made the comment that she didn’t trust radiography because the airway is dynamic and fluctuates in size. You’re not dealing with a balloon here.
      This is NOT to say that I don’t think that orthodontic treatment and surgical correction can’t benefit affected individuals. It absolutely can, and I’ve seen it hundreds of times personally. But if you’re trying to present scientific justification, volumetric measurements are IMHO completely useless and don’t stand up to real scientific scrutiny. Sure, use measures to show improvements in patients’ symptoms or whatever. But I think Kevin’s observation here is that the volumes can’t be the thing you hang your hat on. Find something else that will justify your claims.

  10. A couple of further points. 1) The accuracy of 3D airway reconstruction depends on the goodness of the edge detection algorithm that is deployed. Next, using a large sample, digitizing noise can be accounted for and, after correcting for size, using say Procrustes superimposition, one can visualize what airway changes occurred, if any, in statistical shape-space. The point I’m making is that if the same level of scrutiny was applied to 2D cephalographs, there might be very few papers in the orthodontic literature.
    2) ‘Jumping the bite’ was first introduced by Kingsley (1880) for orthodontic correction of a Class II condition. Some time later it was shown that by suturing the tongue to the lower lip in cases of Pierre Robin syndrome, not only was the micrognathia addressed but the breathing obstruction was resolved. We can use the findings of these empiric studies using modern technologies to address upper airway issues in patients, even tho’ it’s not an orthodontic issue, and therefore, lies outside the remit of the specialist orthodontist.
    Celenk M, Farrell ML, Eren H, Kumar K, Singh GD, Lozanoff S. Upper airway detection in cone beam images. J Xray Sci Technol. 18(2):121-135, 2010.
    Moyers RE, Bookstein FL. The inappropriateness of conventional cephalometrics. Am J Orthod. 1979;75(6):599-617.
    Kingsley NW. A Treatise on oral deformities as a branch of mechanical surgery. New York, 1880.
    Longmire, W.P. Jr., Sanford, M.C. Stimulation of mandibular growth in congenital micrognathia by traction. Am. J. Dis. Child. 1949;78:750.

  11. This is a link for 50 days of free access to the paper.



  12. I’ve been saying this for ages. I’ve had a CBCT since 2008 and have seen wild swings in volumes in the same patient at different time points. Yet our journals keep publishing studies that have had absolutely NO real controls (ie. untreated individuals at different time points) and spitting this stuff out as “science”. Such is the “rush to publish” in the world today.
    Not that I don’t think that mandibular advancements, expansion, etc actually work. They sure can. But poor science doesn’t do anything to corroborate it. So don’t pass this stuff off as science – it just degrades the body of work in general.

Leave a Reply

Your email address will not be published. Required fields are marked *