a 3D DICOM volume engine built to learn GPGPU, Rust, and Bend2.

The goal is simple: achieve the best fidelity possible with as few issues as possible when rendering a 3D volume based on a DICOM collection (like from a PET/CT scan). I worked on this when I was in college, but my work was mostly theoretical. I know there are standard solutions online, but my goal is to build it myself and learn in the process. This is a study side-project because I want to try something different from my daily boring webdev/REST API routine and experiment with new technologies.

For this project, I intend to use Rust and Bend 2 for the backend. Bend is a new parallel GPU-focused language, and I chose it simply because it sounds interesting.

Planned structure:

Backend: Rust (parser) + Bend 2 (preprocessing & HU matrix normalization).
Rendering: Three.js + custom Ray Marching shaders (WebGPU).
Manipulation: Wasm (spatial operations).

It might take a while, but I'm excited about it, so let's see how it goes! ( :

Go to Homepage	`g` `h`
Go to Done Todos	`g` `d`
Compose a New Todo	`n`
Go to Search	`/`
Show this dialog	`?`

a 3D DICOM volume engine built to learn GPGPU, Rust, and Bend2.

Keyboard Shortcuts