The first 10clinicsnever pay.
A native DICOM viewer on its way to a complete PACS. Those who arrive at first light use it free — forever.
10 of 10 seats · free for life
Where the project stands
No makeup:
the real state.
Akhet started as a DICOM viewer and is halfway through phase 1: the clinical core — window/level, measurements, multi-viewport, MPR — already works on real CT, MR, US, mammography and angiography studies.
What's missing is what shows up least in screenshots: the study browser, so a session doesn't start from zero, and the Windows and macOS installers, so a clinic can simply download and open it. That is exactly where this project needs people.
Phase 1 · working today
This is not a mockup.
It already runs.
Every item below exists in the code, covered by tests that run against real DICOM studies on every commit. Everything that comes next — mini PACS, server, integrations — is built on this base.
Clinical Window/Level
Presets for soft tissue, bone, lung, brain, liver and mediastinum. Mono pixels stored in 32 bits — high-range 16-bit US never overflows.
Full measurement suite
Ruler, angle, Cobb angle, deviation, ellipse, arrow, polygon and pencil — with millimeter values derived from the study's own geometry.
Multi-viewport up to 5×4
Series grouped by geometry, drag-and-drop of whole folders, per-viewport multiframe cine and cross-series reference lines.
Triplanar MPR
Multiplanar reconstruction in patient space with a synchronized crosshair: clicking one plane takes the other two to the same point.
14 color maps
Per-viewport LUTs rendered on the GPU, with CPU–GPU parity tests running in CI on every commit.
~48 keyboard shortcuts
Shortcut model inspired by RadiAnt. If you already report on another viewer, you don't start from scratch.
Native + GPU end to end
Decoding via dicom-rs and wgpu rendering with shader and LUT on the GPU. Large studies open without freezing the workstation.
PHI protected by construction
No patient name, UID or file path ever reaches a log or recorded session. The types themselves redact sensitive data in Debug.
Tested against real DICOM
204 tests over real samples of CT, MR, US, XA, mammography, scintigraphy and endoscopy — including volumetric series.
Straight from the viewer · no mockup
This is how it looks today.
Real captures of Akhet opening public DICOM samples. Multiplanar reconstruction (MPR), 3D volume rendering and the measurement tools — all in the same dark interface, with clinical information in the corners.
Roadmap · where this is going
From viewer to complete PACS
Five phases, in order, documented with a study of six competitors. Each phase only starts when the previous one is in real use.
- 01In development
DICOM Viewer
The clinical core already works. The study browser and installers are missing.
- Zoom, pan and window/level — done
- Measurements and annotations — done
- Multi-series and multi-viewport — done
- Triplanar MPR — in progress
- Study browser — next
- Windows and macOS installers — next
- 02Planned
Mini PACS
The study stops being a file and becomes a searchable archive.
- Study import
- Organization by patient
- Search and study indexing
- Local storage
- Internal sharing
- 03Planned
Complete PACS
Akhet becomes a server: modalities send straight to it.
- DICOM server
- Query/Retrieve and C-STORE
- Worklist and modality integration
- User management and auditing
- High availability
- 04Planned
Clinical platform
Integration with the rest of the hospital, from RIS to the browser.
- RIS and HIS integration
- HL7 and FHIR
- Web viewer
- Desktop and mobile apps
- 05Planned
Artificial intelligence
Support for the radiologist on top of the base the previous phases built.
- Diagnostic support
- Organ segmentation
- Change detection
- Clinical research tools
Founders · 10 seats
The first 10 clinics never pay.
One subscription covers the whole clinic — every workstation, the whole team. The first ten never pay: they reserve their seat now and lock in free-for-life for when the installers are ready. After them, each tier of ten customers rises R$ 0.50 a month.
- Free1—10
- R$ 5,0011—20
- R$ 5,5021—30
- R$ 6,0031—40
- R$ 6,5041—50
FreeForever — the whole clinic, every workstation included.
- Free for life — no card, no deadline
- Every machine and the whole team, no per-seat charge
- A direct line to me for requests and bug reports
- Your workflow and your studies prioritised on the roadmap
In return: use it in real practice and tell me, unfiltered, where it gets in the way.
Claim a founder seatOnce the ten are taken, this tier closes and does not reopen. After that, each customer pays R$ 5,00 per month.
One subscription per clinic or professional, per month, covers unlimited machines and users, in Brazilian reais. The price of the tier you join never changes while your subscription runs — increases apply only to new customers.
Engineering
Medical software
taken seriously.
Medical imaging software is used for years and carries patient data. These guarantees are not promises in a document: they are verified by the compiler and by CI on every commit.
- Architecture enforced by the compiler
- Hexagonal layers where the dependency rule is checked at build time: the domain knows no database, the presentation knows no GPU.
- Discipline you can audit
- At most 200 lines per file, tests separated from code, zero unwrap in production — all verified by custom CI gates.
- Patient data is sacred
- PHI never touches a log, telemetry or session recording. It's not a policy written in a document: the type system prevents it.
- Reproducible by default
- Every session can be recorded and replayed event by event against the current build. A reported bug is a reproducible bug.
Build with us
A PACS is not built alone.
I'm looking for people who want to build real medical software — in Rust, with clean architecture and tests over real studies. In return, you help shape an application you'll be proud of, from the foundation up.
Distribution and installers
CI already compiles on Windows and macOS. What's missing is turning that into installers a clinic can download and use — today it's the project's biggest blocker.
Study browser and local index
The SQLite index exists and compiles. It needs to be wired to the interface so each session doesn't start from zero.
DICOM domain
Compressed transfer syntaxes, new modalities, SUV/PET, anonymization. If you know the standard, there's fertile ground here.
Clinical eye
Radiologists and technologists who test the viewer with real studies and say where the tool gets in the way instead of helping.
Interested? Write about what you do well — Rust, DICOM or clinic routine — and where you'd like to start. Access to the project is arranged in that conversation.
For clinics
Do you work with medical imaging?
If you're from a clinic or diagnostic center, write telling us which modalities you use. We'll get in touch as soon as there's a build for your system — and your feedback becomes a roadmap priority.