Ultralearning Radiology

The science of learning has become a lot more popular over the past few years than when I was a student. Contemporary medical students utilize spaced repetition algorithms for their Anki flashcard decks, enjoy high-quality question banks, watch videos at 2x speed, and drill with picture-based mnemonic tools like Sketchy Medical. These techniques have minimal overlap with the medical school I started fifteen years ago (in-person lectures, books, repeat).

A lot of the most compelling educational literature forming the basis for our current conception of optimal learning is well-summarized in the book Make it Stick (the big three being [1] spaced repetition, [2] retrieval practice, and [3] interleaving). A less academic and more casual book repackaging of this evidence-based approach is Ultralearning by Scott Young. I wrote about it back in 2020 in this post about the transfer problem.

We can summarize Young’s key components of “ultra” learning as Directness, Drill, Retrieval, Feedback, Retention, Intuition, and Experimentation.

This post is meant to prompt you to do the work of meta-learning: learning about learning, trying to figure out the best ways to learn the art and science of practicing radiology.

Directness

Directness: the practice of learning by directly doing the thing you want to learn

As in: we learn to do things by actually doing them.

Obviously, we do need to actually read in order to learn. Some passivity is unavoidable, but the point is that passive learning alone (reading and re-reading, watching YouTube) is an inefficient way to retain information in the long term and develop real-world skills.

One of the beauties of residency training is that for the real goal of practicing medicine, much of your daily learning is, in fact, direct. You can learn how to read scans by reading scans. Elegant in its simplicity.

However, it’s also true that part of the purpose of residency training is not just to prepare you to do the job but also to prepare you for jumping through the hoop that is board certification. That means the best way to “directly” prepare for a multiple-choice board exam is to do lots of multiple-choice practice questions. That also means that a key way to prepare for taking cases for the upcoming “new” oral boards is to—you guessed it—practice “taking cases” out loud during conferences or in coaching sessions.

Again–of course–the imperative to avoid passive methods does not mean that you literally shouldn’t read and can never watch a lecture. It’s that whenever possible, you should engage with the material in an active way (pausing, quizzing yourself, etc) and not just let time pass as the material washes over you. This is also why lectures have more staying power when they are engaging and involve cases & testing.

Drill

Drill: Practice skills repeatedly until they become automatic and natural.

Drilling makes a lot of intuitive sense when it comes to procedural skills, but its usefulness is much broader.

When it comes to drilling in diagnostic radiology, I also interpret this to mean honing your search pattern/approach. When you practice doing things the same way every time, you are etching pathways in your brain and making it easier to fall into good habits. When good practices are automatic, you free up cognitive load to handle the harder work.

Having a strong model of normal allows you to pick up on deviations, and that allows you to focus the spotlight of your attention on picking up subtle abnormalities. And having models for different patterns of disease allows you to pick up on subtle constellations of findings. This all largely relies on experience (of course!) but by focusing on developing these internal checklists and really attending to even the normal cases, you can build these models faster and deploy them with greater ease. In time, cases that tighten your sphincter on your first month of nights barely quicken your pulse with experience.

Even if you choose to take a somewhat less concrete approach to a search pattern and not look at every single body part in exactly the same order every single time for uncomplicated studies, you can still develop internal checklists in a problem-focused way. Developing associations like when I see finding X I will also look for findings YNZ to determine its significance. For example: When you see mastoid opacification on a trauma case, you will look extra carefully for a subtle temporal bone fracture. Depending on your imaging protocols, perhaps that means you will open up the accompanying CT of the cervical spine, because the thinner slices on that exam may make a fracture more apparent even though it’s the brain that you’re reading right now.

Everyone employs a countless number of little tricks every day to get the job done, many of them subconscious and forged from the unavoidable litany of past mistakes. One of the ways you can force yourself through these mini checklists is to create task-focused macros in PowerScribe. An example would be creating a tinnitus macro that walks you through important factors predisposing to tinnitus like aberrant carotid anatomy, a persistent stapedial artery, a high-riding and/or dehiscent jugular bulb, and tumors.

Active learning in all of its forms is well worth doing. Be on the lookout for good sources of learning questions, flashcards, and scrollable cases. Use a variety of practice exercises to target specific skills. Yes, that may even mean using an Anki deck to drill anatomy or rapidly review Aunt Minnies so you don’t backslide between rotations.

* * *

A related key point of directness and drilling (as the adage goes): “You play like you practice.”

So one key to making the most of daily practice is deliberate practice. Your goal as a trainee should not just be to get through the cases as fast as possible but rather to build strong habits and grow your expertise. That means not cutting corners. That means reviewing the chart for things like operative notes, making sure to review the previous exams, and looking up new diseases and their features. Speed is partially a natural consequence of a well-honed approach and less of a goal in itself. Once you’re able to do a good job slowly, then you can work on doing a similarly good job quickly. Focusing on speed early often results in shoddy work.

If you learn by doing, then you learn to do well by doing well.

Retrieval

Retrieval: Try to recall information from memory rather than simply re-reading or reviewing it.

Familiarity does not equal knowledge. When you read something and it makes sense, you think you know it–but you often don’t.

Spaced repetition has officially been popular long enough for medical students that flashcards and active recall are now spilling out into residency programs. I can’t vouch for deck quality, but there are a number of decks geared toward the Core Exam and other radiology topics. Active recall is a great way to retain key facts and drill the important things that you otherwise learn just to forget and relearn over and over and over again (like the anatomy of the facial nerve.)

(As an aside, one interesting tool I learned of recently is Jungle. It’s a free Chrome extension that uses some machine learning algorithm to generate text-based flashcards from YouTube videos, websites, and PDFs. You can then individually save generated flashcards into an Anki deck. I haven’t used it, but I suspect this would be a great way to pick out the things you need to work on from articles in Radiopaedia and journal articles after looking up information to handle a case.)

Test yourself regularly on the material using flashcards (or quizzes or trying to match a written search pattern/template).

Use active recall to summarize concepts in your own words or teach them to other trainees or medical students.

But mostly: Be aware of the reality that re-reading and re-exposure are deeply inefficient ways to learn information. When you read something that you think you know, you scan it. But that’s often not enough to sufficiently strengthen those synaptic connections and refresh it into long-term memory. Familiarity is not the same thing as true knowledge. The fact that you’ve learned something before and it makes sense to you when you read it again inevitably affects the energy you marshall in trying to actually relearn it.

That’s the reason why it takes so long to really learn things like the branches of the external carotid artery and why it’s so frustratingly easy to forget. You should be able to narrate these findings and summarize them out loud. If you can’t say it out loud–yes, like you would during an oral board examination–then you probably don’t really know it well enough.

Feedback

Feedback: Actively seek feedback from mentors, colleagues, and patients to identify areas for improvement and adjust learning strategies accordingly.

In an ideal world, you would get meaningful, specific, and actionable feedback from attendings on a regular basis that would help guide your learning. Most of us do not live in an ideal world. The reality is that you will have to create your own feedback whenever possible. That means dictating a report before looking at the wording of the prior report to see how you describe something versus another radiologist or if you made all of the findings. That means keeping a list of cases with unknown diagnoses and following up to see what the follow-up studies, operative report, or biopsy show.

I cannot stress how important it is to do the work of following up your own cases if you really want to get the best feedback in radiology. It’s true you won’t get all the feedback you want, and the feedback you do get may take months to bear fruit. But nonetheless, just like prior exams are a source of educational gold for reading new cases, your own follow-up cases remain an important teaching tool. Don’t shortchange yourself.

So you can, at least to some extent, assess your own performance. Critically review your own work and identify areas where you can improve, such as image interpretation accuracy or communication skills. You don’t need to wait for a third party to evaluate your performance to think of ways to improve.

Retention

Retention: Use memory techniques such as spaced repetition (or even mnemonic devices) to improve long-term retention.

Consider using spaced repetition and incorporating a flashcard approach of some kind into your schedule. When using Anki or a similar service/program, keep up the habit and embrace the algorithm.

Keep drilling so that you don’t fall prey to the forgetting curve.

Do high-quality practice questions every rotation, and then make sure to do some in between rotations as well. This is how you can reinforce the retention of key concepts over the long haul.

While they can be exceedingly lame, mnemonic devices can often help you remember complex information. Just make sure you don’t skip the step of understanding that information first.

Intuition

Intuition: A deeper understanding continuously leads to new insights and connections.

Look for patterns and connections between different concepts to develop a deeper understanding of the material. The human body is a complex machine, but patterns of disease, biomechanical concepts, and imaging features show up again and again across different subspecialties and modalities.

Your goal is to be a real, three-dimensional person with an intuitive grasp of imaging science and not just a cog-in-training. Use your critical thinking and problem-solving skills to approach new or challenging cases.

Never forget that a request for imaging is a question that we attempt to answer with our report.

Experimentation

Experimentation: Try new approaches and techniques to overcome obstacles and optimize learning.

It’s easiest to keep doing what you’ve been doing. It’s obviously worked so far. The more set we get in our ways, the less we are inclined to opt for disruption.

But chances, you haven’t hit on the optimal routine. I know I haven’t.

So, seek out new learning opportunities (like actually paying attention during conferences). Experiment with different approaches to image interpretation, procedural techniques, or patient communication to identify what works best for you. Then, iterate to further hone your practice. Every once in a while, take the plunge and try someone else’s approach.

Try different techniques and see how they affect your learning. For example, while interpreting a case without clinical history is a terrible idea, doing the initial review without the clinical history as an anchor may help you approach cases with an unbiased perspective and develop your eye. Then, following that initial survey with targeted reviews of high-yield areas to exclude specific relevant pathologies is another great practice to try out. You can develop search patterns within search patterns as a tailored way to make sure you address the clinical scenario and provide high-quality care.

* * *

On one hand, we have a time-based training process because, through the sheer number of repetitions, the vast majority of people eventually reach competence.

But that doesn’t mean there aren’t better and worse ways to get there.