Radiology Cases: 52 real films, and you write the answer
A new part of the library. You are shown one real radiograph and three empty boxes — findings, differential, diagnosis — and nothing is revealed until you have committed. Six cases are free on every account.
Radiology Cases is live. You are shown one real radiograph and three empty boxes — what you see, what could cause it, what you would call it — and nothing is revealed until you have written all three.
There are 52 films to work through. Six of them are free on every account: two beginner, two intermediate, two advanced.
Why free text and not multiple choice
Multiple choice teaches a skill you will never use. No patient arrives with four options and exactly one correct answer, and a well-made distractor list quietly does half the diagnostic work for you by telling you what the candidates are.
So these cases give you nothing to eliminate. You get a film, a short clinical note, and a blank box. Naming what you can see, in your own words, before you know what it turns out to be, is the thing that is actually hard — and it is the thing being practised here.
You are never marked wrong for a defensible reading
This is the rule the whole feature is built around, and it is worth being precise about.
Many of these diagnoses were not settled by the radiograph. They were settled by histopathology, or at surgery, or on a CBCT you are not shown. On the film alone, a sound differential is as far as any honest reader can go.
So the marking follows the evidence. Where the answer came from tissue or from theatre, a differential containing the true entity is a complete answer, and the diagnosis is revealed to you rather than scored against you. Where the imaging genuinely does settle it, the diagnosis is marked properly. Where nothing confirmed it at all, it is not scored against a truth that does not exist.
What each case teaches afterwards
The feedback is not a score with a right answer stapled underneath. Each case sets out:
- What is actually visible on the film you were given — not on the CBCT the authors also had.
- What looks decisive and is not. The corticated margin that students read as "cyst" when it is the cortex of the sinus floor. The size. The association with an unerupted tooth. Each one is named, with the reason it fails to separate the options in front of you.
- What finally settled it, and whether a radiograph could ever have done so.
Where the films come from
Every radiograph is a real patient's, taken from openly licensed published sources — CC BY, CC0 or public domain — de-identified by its original publisher, and reproduced with its credit and licence shown in the app. We use nothing licensed NonCommercial or ShareAlike.
They cover caries, four presentations of periodontal disease, cysts and odontogenic tumours, fractures, developmental anomalies, MRONJ and osteoradionecrosis, jaw metastasis, TMJ degeneration and more.
Three of the 52 have no lesion at all. Two are normal films and one is a positioning artefact. That is deliberate: a library where every image hides something quietly teaches you that finding something is always the right answer, which is a habit that harms real patients.
Two honest limits
The films are teaching cases, not radiology reports, and they are not a substitute for reporting by a qualified radiologist. And the marking of your written answer is done by an AI model — it is a study aid, it can be wrong in both directions, and it is not an assessment of your competence.
What the AI does not do is decide the answer. The diagnosis comes from the published case, not from the model.
Posted by Dentalverse Team · Denverse Ltd