The demo renders 3D dose distribution and delivery paths directly in a modern web browser.
*A UT Southwestern team replaces two‑decade‑old TPS tech with WebAssembly and WebGPU. The move could break the Varian‑Raysearch monopoly and put treatment planning in patients' hands.*
UT Southwestern Medical Center has unveiled a browser‑based radiotherapy planning demo that challenges a market long held by Varian and Raysearch. The existing TPS software still relies on a two‑decade‑old technology stack, limiting flexibility and inflating costs for hospitals. By moving the entire planning workflow into a modern web environment, the team aims to democratize access and simplify patient education. The demo runs on WebAssembly and WebGPU, delivering 2D/3D visualizations and a GPU‑accelerated dose engine directly in the browser. If the approach scales, it could force entrenched vendors to modernize or lose market share.
The treatment planning station (TPS) market is dominated by Varian's Eclipse and Raysearch's Raystation. Both products still run on a technology stack that dates back roughly twenty years. Legacy code limits integration with modern hardware and hampers user experience. The monopoly means hospitals must accept dated interfaces or pay premium licensing fees. This concentration creates a clear opening for any solution that can deliver comparable functionality on a contemporary stack.
Researchers at UT Southwestern Medical Center assembled a stack that runs entirely in a web browser. They leveraged WebAssembly and WebGPU to render 2D and 3D treatment delivery visuals and to power a GPU‑accelerated dose calculation engine. The engine simulates dose deposition and plan delivery inside the browser, eliminating the need for heavyweight desktop clients. By moving computation to the client GPU, the system achieves real‑time performance while remaining platform‑agnostic.
The browser‑based approach promises easier explanation of the planning process to patients. Because the interface runs in any modern browser, clinicians can show live visualizations during consultations. This accessibility could demystify radiotherapy, improve informed consent, and reduce reliance on opaque, vendor‑locked software. The team emphasizes that the technology is intended to make the planning workflow transparent, not to replace clinical decision‑making.
The original demo was authored by Chris Kabat at UT Southwestern using three.js, a JavaScript 3D library. Neuralrad later rewrote the codebase in Rust, compiling it to a WebAssembly module. The Rust rewrite improves performance, safety, and maintainability while preserving the browser‑first delivery model. The final product runs as a single wasm module, demonstrating that high‑performance medical calculations can be delivered without native installers.
Varian's Eclipse and Raysearch's Raystation currently monopolize the TPS software field.
The team uses modern browser technology, specifically WebAssembly and WebGPU, to run 2D/3D displays and a GPU‑accelerated dose calculation engine inside the browser.
Chris Kabat at UT Southwestern originally developed the demo with three.js; Neuralrad later rewrote it using a Rust stack and compiled it into a wasm module.
The UT Southwestern prototype proves that high‑precision radiotherapy planning can live in a browser, sidestepping legacy code and costly vendor lock‑in. By coupling Rust‑compiled WebAssembly with WebGPU, the team delivers real‑time dose calculations and interactive visualizations on commodity hardware. The demo signals a potential shift toward open, accessible treatment planning tools, pressuring the monopoly to adopt modern stacks or risk obsolescence. The next step will be clinical validation, but the technical proof is already in the browser.
Sources: [Hacker News – Show HN: Radiotherapy Delivery Demo]