§ 01
The problem
When a doctor explains a diagnosis, patients nod along to a plastic model or a flat diagram and leave without really understanding it. Most 3D anatomy atlases are 30 MB+ downloads under restrictive share-alike licences, which rules them out for a clinic website or a phone in a waiting room.
§ 02
The decisions that mattered
- Generate the geometry, do not ship it. The eye is nearly spherical and radially symmetric, so all 29 structures are computed at runtime from measured anatomical dimensions. The derived axial length (23.88 mm) and lens thickness (3.53 mm) land on the clinical averages, and the whole app is ~248 KB gzipped.
- Map vessels in clinical coordinates. Retinal arcades are drawn in fundus coordinates (mm from the fovea) and wrapped onto the sphere, across all five circulations, each selectable to explain a specific disease.
- Teach the visual pathway by its failures. Eight lesion sites, from retina to occipital cortex, each paired with the visual-field defect it produces, colour-coded by hemifield.
- Make the screening tests honest. Acuity, Amsler grid, contrast, colour and astigmatism tests are calibrated against a physical reference object, so the measurements mean something on any screen.
- Tie anatomy to symptoms: a disease database and a red-flag system that tells patients when a symptom needs urgent care.
§ 03
What it shows
First-principles engineering instead of reaching for assets: solving the geometry turned a 30 MB problem into a 248 KB one. It is also a product built for a real audience, patients, with clinical accuracy as the requirement rather than a nice-to-have.
Want this kind of thinking on your product?
Book a 20-min call