Image occlusion gives you the closest thing to a real lab practical inside Anki: a labelled figure, a red mask hiding one structure, and you in the loop trying to remember whether that is the brachial artery or the radial. Used well, it’s the highest-yield card type for anatomy. Used badly, it’s a 200-card deck you delete after one frustrating week.

This post covers what “used well” actually looks like.

Why image occlusion beats plain flashcards for anatomy

Anatomy is spatial knowledge. You don’t just need to know “the median nerve runs through the carpal tunnel” — you need to recognise it on a figure where ten things look similar. Plain Q/A cards train the word. Image occlusion trains the recognition.

Compare the two approaches:

  • Plain card: “What nerve runs through the carpal tunnel?” → “Median nerve.” You learn the verbal pair.
  • Image occlusion: A wrist cross-section with one structure masked → you try to identify it from its position relative to other structures. You learn the spatial map.

Plain cards score you on the question’s wording. Image occlusion scores you on the same skill your exam scores you on.

How Cardivate makes image occlusion cheap

Historically the painful part was making image occlusion cards. You had to:

  1. Find a clean figure in your textbook
  2. Crop it
  3. Manually draw rectangles over each label
  4. Type the answer for each rectangle
  5. Generate the cards

Cardivate compresses all five into one step: feed it a PDF page, click Image Occlusion, and the AI identifies the labelled structures and proposes masks. You review them in the preview pane, adjust any that landed on the wrong label, and accept the batch.

Typical output for a labelled cross-section: 8–15 occlusion cards in under a minute.

The four mistakes that ruin occlusion decks

After watching dozens of students set these up, the same four patterns wreck the experience:

1. Masking too much

The whole point is to test recognition of one structure. If your mask covers half the figure, you’re testing comprehension, not recognition. Aim for one structure per card, with the rest of the figure visible as context.

2. Using ugly source images

Image occlusion only works if you can clearly see what’s around the masked structure. A photocopy of a photocopy or a low-resolution scan kills the technique. Use figures from PDF textbooks or clean atlas pages, not phone photos of whiteboards.

3. Forgetting that the front matters

Many students put no question on the front of the card. The student sees a masked figure and has to guess context-free. Add a clear prompt on the Label field: “Identify the structure passing through the carpal tunnel” — narrower than just “what’s this?”, but not so narrow it gives the answer.

When 12 cards all come from the same diagram, group them. Cardivate auto-tags cards with the source page, so you can filter to “all wrist cross-section cards” during a session. Studying related structures together cements the spatial relationships between them, not just each label in isolation.

A simple drill that works

After your image-occlusion deck has matured for a couple of weeks (each card reviewed at least 3 times), try this:

  1. Open the Cardivate Mock Exam dialog.
  2. Filter to the image-occlusion deck.
  3. Enable confidence-weighted scoring.
  4. Run 15 questions, timed, 15 minutes.

What you’ll discover: you’re often confidently wrong on structures that share a region — radial vs ulnar nerve at the wrist, dorsal vs ventral roots of a spinal segment. The confidence weighting punishes confident-wrong specifically, which is the calibration signal you actually need for an OSCE or a viva.

Iterate on the structures where you keep losing confident points. That’s exactly what the test will throw at you.

When not to use image occlusion

A few cases where plain cards beat it:

  • Pure terminology without a visual referent. “Define osmolarity.” → just use a basic card.
  • Numerical values that need active recall, not visual recognition. Lab cutoffs, drug doses, half-lives.
  • Pathways and cascades where the relationship between steps matters more than identifying any single one. A cloze chain works better here.

Image occlusion is a specialist tool. It pays back 10× the cost when applied to visual recognition, and adds noise when applied to anything else. Pick your sources accordingly.

One last thing

The students who get the most out of image occlusion are the ones who treat the masks themselves as a study activity. While Cardivate’s AI proposes masks, you adjusting them is itself review — you’re forced to look at the figure carefully, decide what’s worth masking, and reject AI guesses that mistook a vessel for a nerve. That five-minute review pass is often more educational than the next hour of Anki sessions.

Don’t skip it.