WebGPU
Modern GPU access from the browser — compute shaders and fast graphics that make client-side medical imaging and on-device ML practical.
In one line
WebGPU is the successor to WebGL: a browser API exposing modern GPU pipelines — including general-purpose compute shaders, not just triangles — with safety and portability across vendors.
How it works
JavaScript (or WASM) creates a device, uploads buffers/textures, and dispatches render or compute pipelines written in WGSL. Unlike WebGL's graphics-only lineage, compute is first-class, so matrix multiplication — the heart of neural-network inference — runs on the GPU without abusing rendering tricks. Browser ML runtimes increasingly use it as their fast backend.
Where it shows up in digital health
Client-side volume rendering of CT/MRI in zero-install viewers; real-time image processing for endoscopy/ultrasound streams; and privacy-preserving on-device inference — running a model on patient data in the browser so the data never leaves the machine. Still maturing across devices; feature-detect and fall back.