Back to browse
Streaming multi-GB CT scans in the browser with WebGPU

Streaming multi-GB CT scans in the browser with WebGPU

by m_panknin·Mar 26, 2026·2 points·1 comment

AI Analysis

●●●BangerWizardryBig Brain

Multi-GB CT scans render in browser when VTK.js requires server-side preprocessing.

Strengths
  • Out-of-core streaming handles datasets larger than available RAM
  • HTTP range requests enable progressive loading without full download
  • WebGPU compute shaders accelerate volume ray-casting on consumer GPUs
Weaknesses
  • WebGPU support limited to Chrome/Edge/Firefox nightly
  • No DICOM native support requires preprocessing pipeline
Category
Target Audience

Medical imaging developers, scientific visualization engineers, radiologists

Similar To

VTK.js · ITK-VTK Viewer · OHIF Viewer

Post Description

Hey folks, I built Kiln, a WebGPU-native out-of-core volume renderer for virtualized volumetric datasets.

Kiln is an end-to-end system for rendering large volumetric datasets in the browser using WebGPU.

It covers the full pipeline: data preprocessing, HTTP streaming, decompression, memory virtualization, and rendering — making multi-gigabyte CT scans and scientific volumes interactive in a standard browser tab using a contemporary tech stack.

The source code and docs are available here: https://github.com/MPanknin/kiln-render

Similar Projects