Superior Canal Dehiscence -- Radiology

The rationale for using reformatted axial cuts.

Timothy C. Hain, MD   Page last modified: October 12, 2018

In the "superior canal dehiscence" syndrome, or SCD, the roof of the superior semicircular canal is missing. The missing bone is identified using high-resolution CT scan.  A good idea whether bone is missing or not can also be obtained from a 3T coronal T2 MRI (Browaeys et al, 2013). See the main SCD page.

This page is an attempt to outline the optimal technique for doing high-resolution temporal bone CT, when one is looking for SCD. There are also some comments about frequent errors in SCD radiology.

Some general principles:

CT Scanner:

The CT should be done of the temporal bone with at 0.6 mm resolution or better (lower is better). It may be impossible to get a CT scan with a resolution < 0.6 mm. This is often sufficient, but don't accept lower resolution (i.e. more than 1 mm is not good enough).

Conventional CT scans of the brain are nearly always useless to diagnose SCD as their cut resolution is 8-10mm -- this is almost as big as the entire inner ear ! There is also a trade-off between radiation and resolution. One might argue that the tiny lesions that can be discovered with 0.1 mm cuts are not worth the radiation load. This issue is presently unclear.

There are two choices for positioning in the scanner -- on the back (axial), or on the face (direct coronal). There are some pro's and cons of each.  Ideally, for diagnosis, there should be both positions -- axial, and "direct" coronal.  Unfortunately, this means doing the scan twice and doubling the amount of radiation.  Lets look at each method in more detail. 

The bottom line is that it is generally best to do high-resolution axials, with coronal and oblique reformatting. If the facility or radiology is unfamiliar with the radiology entailed with diagnosis of SCD, then it is best to do true-coronal high-resolution.

Cone-Beam CT scans may be superior to ordinary CT scans as they provide better resolution with far less radiation. Cone beam CT scans are standard equipment for dentists, but have not yet become very common outside of dentistry.

Direct Coronals:

True or direct coronal sections are intrinsically better than reconstructed or reformatted coronal images. The direct coronal view is taken in supine or prone with the neck hyperextended and with a tilted gantry (one cannot get true coronal because of limitation of neck and gantry angle). In young patient one can get good images but there are problems with older patients, those with neck problem or children.

Axials can be also  "reconstructed" -- reformatted -- to provide a "pseudo" coronal view.  Our general experience has been that when coronals are produced in this way, they are sometimes fuzzy and can be uninterpretable. However, it is possible to do a pretty good job with reconstructions (see below).

Reconstructed images from Axials:

Although true coronals are more accurate, it is generally best to perform direct axial and to perform reconstructed coronal and oblique parasagittal views. This is to reduce radiation exposure. The oblique images are essential for complete evaluation of SSCC. This is the view at times referred to as the Poschel view. To do the oblique view properly, the CT technician has to identify the top axial section for SSCC, and connect the dots of the limbs of the SSCC. This is the reference section. Then one makes oblique images using this line. The radiology technician will use this line as reference and produce reformatted images parallel and perpendicular to the line).  This results in an image that shows the entire SSCC (one of the images that are coplanar), as well as a series of images that show the canal and surrounding bone.  The reformatted oblique perpendicular to the reference line is the best to see the PSSC in its plane. This is also critical for the anatomy of facial canal as well as the round widow in case there is a perilymphatic fistula.

Why you should not do coronals without an axial

When one does the direct coronal, a fixed reference line cannot be drawn as is the case for axial. The reason is that coronal sections depend on the neck extension and gantry angle and therefore is slightly different in each case. It is therefore very difficult to generate a reproducible oblique views from direct coronal.

It is not easy to get the Poschel view from direct coronals. By reformatting from axials ( which is easy for techs) we know that we get always the right images. For SSCC one depends on the oblique more than direct coronals, as the slope of tegmen and arcuate eminence/ fossa may give a false impression of a defect. All the SCCs can be evaluated by having axial and reformatted coronal and oblique images. Axial images are best for evaluation of VAQ and arachnoid granulation along the posterior petromastoid plate.

Does one lose resolution on reformatting ?

For resolution alone, the ideal study of temporal bone is the direct axial and direct coronal. However, since the multidirector spiral technique provides isotropic volumetric data, one should not lose any resolution on reformatted images. If there is no motion on axial images it is possible to get excellent reformatted images in true coronal and any other planes. There are some cases that reformatted coronal may not be as good as direct coronal. The issue of resolution is complicated and it is not only the matter of pixel size but rather the field of view, detector format, manufacturer of the software (Mafee, 2005)

3D Reconstruction

3D reconstructions will show the full cortical bone thinning/dehiscence.

Figure 2a. Coronal thin cut CT scan showing superior canal dehiscence (SCD). This patient was reported in detail in (Ostrowski, Hain and Wiet, 1997) Figure 2b. Temporal bone CT scan with images taken in plane of superior canal. This is an oblique reformatting, sometimes called the Poschel view. There is a wide area of dehiscence seen at the top.

Blunders encountered in CT scans for SCD.

SCD is apparently difficult for radiologists to appreciate errors are far more common in interpreting these CT scans than other types of inner ear imaging. Radiologists almost never miss acoustic neuromas, given that contrast is given, but they miss SCD, roughly 10% of the time.

If your patient gets a CT scan of the temporal bone -- you HAVE TO LOOK AT IT -- it is not enough to trust the radiologist. If you can't read these scans, you shouldn't be ordering them.

With respect to errors -- first, lets acknowledge that everybody makes mistakes from time to time, and radiologists are a subcategory of everybody. Or in other words, radiologists make mistakes. That being said, there seem to be more mistakes made with SCD interpretations than more common studies like chest X-rays, etc.

The usual method of catching errors is to have someone else check your work. Oddly enough, this rarely happens in radiology. We think this is a structural problem, that perhaps could be fixed with radiology departments by having every scan read twice (by two different radiologists of course). Of course, this would double the cost of interpretation, but likely reduce the errors greatly.

The three types of SCD blunders are:

1. The usual type of blunder in missing SCD is not looking for SCD. Presumably radiologists are sometimes in a hurry, like all of us, and don't always check all of the images. That is why someone else should review their results -- see above -- if you are the ordering doctor, this should be you. If you are not competent to check them yourself, and there was a good reason to do the test in the first place, send them to someone who is.

2. The second most common blunder in missing SCD is having suboptimal CT images. Everything is easier when the images are "optimal", which translates into having someone set up the right images BEFORE the patient is seen. This requires not only that there be a competant person in your radiology department who knows how to do these scans, but also that there is someone who can read the incoming radiology requisitions and transfer them into whatever software your radiology department is using. Thus perhaps two people.

If there are no coronal reconstructions, SCD will not be visible. If the resolution is 1 mm or more, SCD is also easily missed. Generally speaking, if the radiology department didn't do Poschel reconstructions (these are the perpendicular ones), they could certainly miss SCD. These are generally errors of ignorance on the part of the radiology department, who may simply not be used to doing these types of Xrays, but sometimes are errors in transfering information into the radiology department from outside physicians.

3. Sometimes the suboptimal images from the 2nd type of blunder results in a third type of blunder -- diagnosing SCD when there is no SCD, or on the wrong side.

If you have a patient in whom the temporal bone CT was not optimal, rather than doing it over, we suggest starting with a screening tests like an oVEMP and/or a threshold cVEMP.


The Bottom Line

In our opinion, this is what your test prescription for a CT scan of the temporal bone should generally say:

CT scan of the temporal bone, with high resolution (1 mm or less). Direct axial and reformatted coronal and reformatted oblique views parallel and perpendicular to the plane of the Superior Semicircular Canals. No contrast.