Radiopaedia Blog: Technique

For this first post in my planned series on how to prepare for radiology oral exams: essential techniques, I wanted to make a golfing analogy (stifle groans please), but you can replace any sport of your choosing if it helps.

If you want to get good at golf you do not simply play round after round after round. Although you will certainly improve, it does not offer the opportunity of working on problem areas; after all in 18 holes you may only get to use a particular club a few times, or putt from a certain distance a few times. Playing a full round exists mainly identify the areas you need to work on.

So lets say you play 18 holes and you notice that you miss three 6-foot putts.  Your next task is to go to the practice green and hit hundreds of 6-foot putts. If that still doesn’t help, then you need to work out what you are doing wrong, and you may need to read some articles on putting, or get a pro to give you a hand etc…

The same applies to practicing for oral exams. Too many of the trainees I see do the vast majority of practice cases in a group, either in tutorials or in study sessions. In both situations the cases are looked at once in an 'exam style' before moving on to the next case. Moreover these sessions usually contain a relatively random assortment of conditions, at most restricted to a system. It is no surprise that I see these trainees make the same mistakes over and over again.

As a rule of thumb you should probably be doing 10 times more cases on your own than you do in a group. Furthermore when you do cases on your own, you don’t do them the same way as you do in a tutorial, but rather you practice them, until they are nice and polished.

This was my routine (for a ~40-60min study block):

1. pick a topic (e.g. bronchiectasis on chest x-ray, including causes such as ABPA / CF / idiopathic bronchiectasis etc… )

2. pick conditions that can look similar (pulmonary fibrosis, interstitial edema, sarcoidosis, asthma etc..)

3. get a pile of films or create a playlist on Radiopaedia.org (e.g. here is one I created earlier)

4. look at the first case and verbally present it as if you were doing it in an exam (you may want to record yourself)

5. check the answer

6. go back and do it again now knowing what the answer is and try and make your presentation as polished as possible, including relevant positives and relevant negatives. Really try and make the language you use tight and unambiguous.

7. move to the next case

As you try and complete step 6 you will probably find that you uncover areas that you are not certain about (e.g. what is the distribution of bronchiectasis in ABPA). This is knowledge you need to know if your presentation is going to be perfect. Because you are on your own, you can then turn to your trusty textbook or favorite online resource (no prizes for guessing which one I would recommend) to find out the answer. Then you incorporate that into your presentation as you practice it once more. 

This might seem tedious, but by the time you have done a dozen cases in this manner, you will have perfected your approach to this whole group of conditions and be able to knock any similar case out of the park.

Next: Practice for your oral technique in the shower

Dr Frank Gaillard is a neuroradiologist at the Royal Melbourne Hospital, Melbourne, Australia, and is the Founder and Editor of Radiopaedia.org.

NB: Opinions expressed are those of the author alone, and are not those of his employer, or of Radiopaedia.org. 

When reporting follow-up studies the usual practice is to compare to the most recent previous study. Although generally this is a safe thing to do, with increased frequency of follow up exams in many disciplines, one runs the risk of not detecting subtle growth. If this occurs repeatedly then a tumor that may have significantly increased over time will have a series of reports all of which state “no change”.

As an example (Fig 1) look at 4 studies in a patient with a partially excised low grade glioma. Each study (A to D) is approximately 6 months apart. Note how hard it is to discern a change between adjacent pairs.

We simply cannot reliably detect subtle change. How much is below our change threshold will depend on many factors, such as the size and shape of the lesion, scan parameters and partial volume effect, slice position and patient position etc… Regardless, there is an amount below which we simply won’t be convinced that any actual change has taken place. This is obvious if you consider what would happen if you were to scan a patient every day. Even the fasted growing mass would look unaltered on sequential scans.

So what is the solution?

Look at older scans and consider what you expect the biological behavior of the process you are looking at to be; the more indolent the process the longer interval you need between scan pairs to detect change.

In the same patient as before, see how easy it is to see that there has been change when comparing scans 2 years apart (Fig 2).

My practice when assessing gliomas for example, is to look at the most recent scan, and then at the oldest valid comparison; one which does not have intervening surgery, and is not in the  immediate postoperative period.

You are then left with three possible outcomes from such a comparison:

1. change is obvious even when just compared to the most recent scan
2. no change when compared to the recent scan but some change when compared to the oldest scan
3. no change when compared to both the recent and oldest scan

In the setting of obvious change, there is no problem, and in fact there is no need to look at older scans.

If there is, however, no change compared to the recent study but change is evident when compared to the older scan, I usually pull up a few of the intervening scans, to try and assess whether growth is gradual, or something has changed in the behavior of this tumor, suggesting dedifferentiation into a higher grade. After all a tumor that had been stable for years but suddenly starts to grow needs, at the very least, closer follow-up and probably also needs to be considered for a change in management. My conclusion then reflects this distinction; e.g. “Although there is little discernable change when compared to the most recent study, when compared to multiple previous studies dating back 4 years, slow steady growth is evident.”

Only if there is no change when compared to the most recent scan, and no change when compared to the oldest scan does my conclusion read “Stable”, and then I append the time period over which I am claiming no change to have occurred; e.g. “Stable, with no appreciable growth over the past 24 months”.

This approach is valid to all comparison studies, regardless of system or underlying pathology. I hope this approach is useful to you, and will stop you from merely concluding with “stable”.