Approaching the Radiology Core Exam

We’ll start with some general thoughts on the exam and preparation, delve into the image-rich, physics, and non-interpretative skills. We’ll conclude with my personal approach (which you probably shouldn’t do), and some thoughts on adapting that to a reasonable regimen that should serve many people well. This post, like all of my Core Exam posts, is really long. This is partly because there is almost nothing written online about this exam (and partly because I am long-winded). You should also read my post about the actual Core Exam itself.

It should go without saying that how much you need to study and what you should focus on will depend on you, your radiology foundation, the holes in that foundation, your test-taking skills, the board reviews your program may or may not offer, and how much time you have.

Last revised March 2021.

Our tour:

• General Approach
• When to Start Studying
• Arranging Your Schedule
• Resources
• The Image-Rich Part
• Physics
• Non-interpretive skills
• My Approach

General Approach

It’s easy to forget the lessons learned from medical school: more usually isn’t better. Limit your resources, don’t let gunners make you question your prep, and don’t spend all day chasing zebras that aren’t classic Aunt Minnies. Esoteric knowledge may help with achieving a high score but it will never be necessary to pass an exam aimed at minimal competence.

One of the most anxiogenic aspects of the Core Exam is its relative newness. It hasn’t become immediately obvious to anyone exactly what is required to safely pass it. When you took Step 1, you had NBME exams you could take or your UWorld percentage to give you a healthy indication of your potential score. For anybody who made it into radiology, the question wasn’t if you were going to pass the exam; it was exactly what score you were going to get within the range you could expect.

So unsurprisingly, “boards fever” is tough to escape. The test (and radiology in general) is challenging, and you can’t predict your passage safely or know when you “know” enough to achieve your desired goal. As a result, most residents probably freak out and study too much. And with the fact that the Core “study guide” is literally a list of everything ever associated with medical imaging, the ABR isn’t helping.

Don’t get me wrong, radiology is a big field and the Core Exam covers a lot of material, but Radiology residents as a group are a high-performing lot who are as a general rule are going to over-prepare. Especially if you read crazy forum posts. This is an exam with an 87-91% passing rate, and presumably many failures are disproportionately coming from programs 1) too stupid to conduct board reviews or give residents time to study or 2) smaller non-competitive (frequently community) programs (not getting into politics here) / those with significantly less than average-for-radiology USMLE scores. Regardless, you don’t need to know everything: even though the exam isn’t technically normalized, the more or less consistent pass rate means you basically just need to be better at taking this test than the worst 10% of residents. Chances are, if anyone in your program has failed, you know who it is and how they perform compared to their peers and other residents. If, looking at them, you honestly think you’re in danger of failing, study super hard. Otherwise, pick a few resources, learn them well, and stop worrying so much.

Now there isn’t lots of reassuring data to back any of that up, except this fascinating study out of Canada which compared the in-service scores of passers and failures. Basically, the people who failed the Core Exam bombed the in-service as well. If you’re getting 10th percentile on your in-service, then you should also push yourself.1I’m aware most programs don’t care about the in-service. But because no one cares, I’d argue it’s still probably not a terrible measure of your starting point and estimate of your danger of failing. That being said, the in-service is a terrible exam by every conceivable metric, and not all programs even do it.

Update May 2019: there is an interesting new article about Core exam predictors in JACR.

When to Start Studying

I think most people would say it’d be nice to get started with some light general review in the fall, but it’s more common and likely to start with the new year. February is a reasonable time to start on the later side, and March is the absolute latest you should go before diving in. The less time your program allows/schedules for board review or the busier your home life is, the more Core review needs to be a marathon and not a sprint and the earlier you should start. If you’re still taking significant call in the spring, then you should probably be studying in the fall.

Many people will fall behind on their schedule (if they have one) and many others will feel that their peers are doing more (and will be concerned). When everyone subsequently passes, none of the lazy cohort regrets not working as hard. The same rules from your earlier board exam experiences apply: schedule yourself time off if you can, find ways to stay sane, don’t just study until you drool and then binge-watch Game of Thrones until you hate yourself. Take care of yourself:

• I was/am sedentary
• I actually regularly studied at a Cinnabon
• I had a baby at home and thus slept terribly throughout the entire PGY-4 year
• I spent my excess energy writing extremely long articles to post on this website when I should have been studying, working out, or sleeping.
• Don’t be me. Or, since I passed too, maybe some of that is okay.2Probably not the Cinnabon.

Arranging Your Schedule

Don’t buffet-dinner your schedule and over-pack it. Your eyes think they can study more than they can. It’s easier to add more resources than it is to deal with the psychological trauma of falling behind. Do a set of RadPrimer questions or do a few cases in a casebook and see how long it takes you. Extrapolate to the size of the resource. Then tack on (at least) 15% extra for procrastination and misery. If you want to do a step-wise tour through all sections, then starting in January, you have the equivalent of two weeks per subject. (10 subjects x 2 weeks = 20 weeks = 5 months)

To organize by subject? By modality? One at a time or mishmash?

By subject, absolutely, and I would argue for the mix & match approach. If you take a completely linear tour (MSK, then neuro, then cardiac, etc.) then it may be months in between your dedicated review and the exam itself. Learning principles of spaced repetition and interleaving suggest that a strict one-time stop at each topic is suboptimal. So don’t follow that approach strictly.

Physics throughout would be ideal in theory but is difficult in practice. I’d argue for reading a physics book or book equivalent in its entirety toward the beginning, reading the same or another toward the end, and doing physics questions in between on regular basis (more on that later). Note that some parts of physics (most of nucs including radioisotope safety, radiation doses and effects, etc.) should be memorized over and over again for improved penetrance. If you’re a flashcard person, Anki is definitely your friend here. Even with that, you’ll still probably need to cram this material last minute. Don’t leave all of physics for the end.

As a personal example: I started studying (poorly) in February (at AIRP). We had internal board reviews which began late March. I had a ~1-year-old at home. Consequently, I never studied a minute at home until the final two weeks before the exam.

The practice exam and the questions at the end of the study guide are 100% high yield and a must-do. You can take one in the beginning and one toward the end to see how you’re doing. I’d split the difference and do the study guide pdf at the beginning to get a feel for what you’re aiming for and the software version toward the end so you’re freshly accustomed to the software. (Update: they removed the bonus questions at the end of the study guide, oh well)

 

Resources

Pick a reasonable amount of material and stick with it. No resource has a monopoly on radiologic pathology. There are more good/reasonable options than you can and should try to get through, and no perfect universal plan. You know yourself: pick what sounds like a good balance for you (textbooks, qbanks vs case books, in-person board review lectures vs. videos, etc.) and do that. If something doesn’t feel right (like you hate a particular book style), then get rid of that and cover it differently. It’s okay to ditch a resource you hate (like the RSNA physics modules), but it’s less good to keep adding things because someone else said it was good.

• Nothing is used by everyone
• People will use things that you won’t
• And ye all shall pass

Your approach will differ depending on how you feel about the Core. Is it an annoying hurdle? A chance to finally sit down and “learn radiology”? Both? Ideally, at least some of the latter will color your studies so you don’t become too cynical (though commiserating with your colleagues is fun). A general text (e.g. Core Radiology and/or Crack the Core) and some complete source of images/cases (qbanks, casebooks, etc; either one resource or a mix to get there) is probably a good start (and all you really need). I’ve included my personal approach at the bottom.

Here is my detailed post and run-down of common resources (and here is my qbank rundown).

Thoughts on Review Courses

These courses may be useful, but in 2016, none are essential. The Core Exam tests the combination of bread and butter, Aunt Minnies, and bullshit. What you need is to take your studies seriously enough to retain the reasonable material and exposure to enough of the rest to hope to get a reasonable fraction right. If going to a course will keep you honest, by all means, do it. But if you don’t have the money or the time, don’t worry for a second. Any of the numerous books, online sources, question banks, or video series are more than enough. Really, way more than possible. No one has a monopoly on this stuff.

Internal Board Reviews

Internal board reviews, if you have them, are probably good and serve a similar function as casebooks and likely qbanks (depending on format). If your board reviews suck, then come together and advocate for yourselves. The best format for board review is a high volume of rapid cases, as many as possible, ideally followed by high yield brief summary slides as indicated. Didactics are a waste of your time. Classic hot seat cases honestly aren’t ideal, but if you take cases they need to not be old school casemanship. Your faculty should allow for 1 sentence findings + diagnosis and if you don’t know it, you don’t know it, move on. Long case-taking from your fellow residents will make you want to murder each other and destroys progress/volume.

 

The image-rich part

I again want to stress that you should not forget the lessons you learned in medical school: sometimes less is more. I understand the inclination to feel that Radiology is somehow different from the rest of medicine (and it is in many ways), but this test and its multiple-choice single-best-answer format is more like Step 1 than you’d think, which means the odds are in your favor. That doesn’t mean you don’t have to study, you definitely do. Among other things, you need to see the breadth of radiology in rapid succession, because you probably haven’t otherwise seen/studied everything recently enough for it to be fresh in your memory, and it would be a rare night on call to diagnose both a rare fetal anomaly and give a differential for an intraventricular mass.

On the real thing, most questions are image-based but not all are image-reliant. For example, imagine you were shown a radiograph of the classic gamekeeper’s thumb. Then imagine you were also shown an MRI. You don’t have to understand wrist/hand MRI to know that they’re showing you a Stenar lesion. That’s the only thing they would bother to show you on MRI. In fact, if you had a coned down coronal MRI of the base of the thumb, I can guarantee that’d be the question whether you had a plain film or not. So it behooves you to understand the types of high-yield associations and classic presentations, as you can sometimes extrapolate what they must be showing even if you don’t really get it.

Conversely, I do feel that the proportion of second and third-order questions was overstated. 3Colleagues I’ve spoken with from the 2016 exam year agreed, though I’m sure others do not. I felt that the primary diagnosis was by far a more common request, and in most cases where second-order associations were asked, they were classic ones that you are bound to learn with normal review or already know:

As in, knowing cystic fibrosis is autosomal recessive would be reasonable to learn.

As in, knowing the exact percentage (10? 15?) of people with some disease that might also experience X or Y, generally not. General/common associations, like RCC and hemangioblastomas in VHL? Absolutely. Things you’ve never ever heard of not even once during your first three years of training? Not so much. No one is going to fail the Core for not knowing Parry-Romberg Syndrome.4It’s a Qevlar question.

As a result, I would place learning primary image-based diagnosis as by far the most important task for your studying, not focusing on learning tons of random factoids for each diagnosis (just one or two).

I see several possibilities for the primary vs secondary question discrepancy from prior years:

1. Our predecessors understudied associations, struggled with those questions, and thus felt that they made up a disproportionate amount of the test.
2. Because of #1 (whether real or imaged), we consequently overstudied associations and thus found that the primary diagnosis was generally more difficult.
3. The ABR is still working out what kind of test they want this to be, and the exam is changing from year-to-year.

Of the three, I actually favor the final option. I support this belief with the fact that there are a lot of really shitty questions.

As I discuss in The ABR Core Exam Experience, the test questions are short, to the point, and a significant portion are easy giveaways. From this, I believe that getting the easy questions right is by far the most important component of passing.

Please also be aware that despite the fact that are some CT and MRI image stacks and some cardiac and US cine clips (which stutter horribly and are difficult to manipulate), the vast majority of questions are still based on 1-4 still images like you’re used to seeing in books, PowerPoints, etc. Stacks, when provided, are often comically short (like 3-5 scrollable images) and apparently always in one imaging plane. Is this reasonable in 2016 given the realities of daily practice? Perhaps not, but you must still be prepared to “see common things in unusual ways” as per the old board adage. The projection or modality may be unusual, but this is how board question writers think they can get at whether you “really understand” radiology.

The image-based questions are generally fair. IR is particularly poorly written and thus hard to study for. Breast probably comes in second for many people.

Miscellaneous thoughts:

• There are very few negatively framed questions (which of the following are NOT…).
• There are some questions that do not actually need the image, but this is rare.
• Sometimes a question about a disease facet can be answered using a feature of the image. Use what they give you.
• There are questions about things that are given away simply by the anatomy being imaged, the modality, and/or history. Again, use what they give you.

Take-home point: Immerse yourself in the art of image-based diagnosis. Don’t overthink the questions. Don’t try to outsmart the test. There are plenty of easy questions. Be prepared to grab the points and run.

 

Physics

While physics is the most commonly failed section and the only one that’s conditioned (1% of the time), I think the physics was overall surprisingly practical and fair. There are no meaningful calculations. Any formulas to be used in your mind are only to understand the relationships for how different imaging parameters are related. You need to know if you change one thing (e.g. kVp), what else goes up, what goes down, is that relationship linear or exponential. The relationships of kVp, MAS, dose, image contrast, spatial resolution. Or pixel size, matrix size, image contrast, spatial resolution. Common artifacts and how to troubleshoot. Etc etc.

So physics is overall more reasonable, practical, and much more straightforward than most of the sources out there have led us to believe based on the old physics examinations, which apparently was a piece of shit. The exception to these more general principles is the need to know absolute specifics when it comes to radiation safety: doses of different common exams, dose thresholds for deterministic effects, types and lag times of stochastic effects, the relative weighting of body parts – these things are actually important. Also, nucs. Nucs still requires memorizing details (physics and safety more so than clinical nucs). It totally does.

When studying, particularly if you read the RSNA modules or Huda, you may ask yourself: why do I need to know about all the film screen physics and other bullshit including the crystals inside different kinds of radiography systems? Well, I really don’t think you do. If any of this junk appears on the exam in future years, it will likely be so low in volume that you can safely skip it and still pass.

Common resources for physics:

• RSNA physics modules
• Radiologic Physics War Machine
• Huda
• Radiology Physics 300 (an iOS, no longer available)
• Raphex
• Duke Review of MRI Principles
• Qbanks (discussed at length separately)

I review the physics resources at length in my resource rundown post.

The short version is that the RNSA modules are good on the whole but super boring, the site is horrendously ugly, they take forever to get through, and include a bunch of BS you don’t need to know. Ditto Huda, which may have been crucial pre-Core but isn’t necessary now.

I passed physics by a wide margin basically using only War Machine and Physics 300. As of 2021, physics is no longer a make-or-break section, which really eases up on the pressure.

 

Non-interpretive Skills

Noninterpretive skills include a combination of contrast and radiation safety, quality improvement, ethics & professionalism & blah blah blah, and very light statistics.  This results in a section based on legitimate clinical information, common sense, and a nontrivial amount of pseudo-MBA quality science and organizational psychology BS. The bottom line is, if you read the Quality and Safety Syllabus a couple of times, especially the day before the test, you should not have a problem with this section. You probably actually know most of the safety stuff and have common sense, so you may be two/thirds of the way there anyway.

If you think there are limitations to applying the principles used to study manufacturing defects with those involved with patient care, I award you a gold star. With that caveat, the ABR’s study manual for the NIS section is a 62-page 50-page 48-page document and a surprisingly readable and reasonably concise resource. You don’t actually need anything else. Question wording is frequently derived from the actual packet, so definitely read the actual packet. I recommend at least twice, including the day before the exam (on the flight, for example).

The material is covered comically in CTC Vol 2, which may help it stick, but this doesn’t replace the packet.

If you want to ruin a workout or need something more passive, Stanford has a nice 11-part video breakdown that you can listen to/watch (based on the slightly longer version from prior years):

My approach

I took a relatively steady and low-time approach. During prime board review time my son turned one and my wife was finishing up her residency with a couple months on a busy consult service. I didn’t study at home.

• I read Core Radiology around February and March
• I read Radiologic Physics War Machine in April. I re-read the high yield section at the end right before the test both days.
• I read Crack the Core (minus the physics and non-interpretive skills sections, which were covered in War Machine) over two weeks in May
• I did the Physics 300 questions 2-3 times throughout
• I went to internal board reviews from late March-June
• I did a haphazard smattering of different qbank questions intermittently, a few hundred from each of the five products (they’ll be reviewed separately)

Entirely absent: Any video lectures. Any external board review course. Any casebooks. Any completely utilized question bank.5This is what happens when you try to review 5 question banks for your website

Thoughts on this approach:

I took what amounts to a relatively lean approach, and this worked fine and I passed by a healthy margin. I felt anxious mostly as a result of the fact that my peers were studying more and were clearly using more resources. My qbank experience was a result of reviewing these products for this site, and I wouldn’t recommend anybody do that per se (one +/- two case/question sources is fine).

Looking back, in an ideal world: I would have read Core Radiology and started RadPrimer in the fall. Done CTC and physics Feb/March and then filled in the rest of the time with questions, probably primarily via the A Core Review Series. Qevlar is nice for the phone app with offline capabilities and probably would’ve made it in too. Most of the latter would have been important mostly for anxiolysis or possibly long-term retention, as passage wasn’t an issue.

I liked the combination of Core Radiology and Crack the Core and wouldn’t change that. I discussed CTC/War Machine at length here, and I’m convinced that it is sufficient alone as a physics textbook (with Physics 300 as a good question source). After that, you need some source of cases (RadPrimer, etc). It doesn’t matter so much as that you pick a way to cover the exam sections and stick with it. You will use a tiny fraction of the available resources and that is not just fine but desirable.

All Roads Lead to Rome. I’ve included my personal approach at the end not because it’s particularly good, but because otherwise the only thing you’ll hear on the internet is anonymous miscellaneous gunner junk you may otherwise find online and from some of your more…passionate colleagues. Like Step 1 with First Aid and UWorld, you can easily pass this exam alone with Crack the Core and RadPrimer. It’s a time and energy-based process more than a resource-based one.