ZYVOR IMAGING
Zyvor AI Labs · Medical imaging
Archive, review and report on your infrastructure. A research viewer built for evaluation, now with MPR, 3D and segments. You do not need to be technical to try it: open a link, or drag one app into Applications, and look at made-up patients.
A real recording of the running app on a made-up phantom: no patient data, no mock-up. Research use only.
Try it
Pick the way that suits you. Every option uses made-up (synthetic) patients, never real ones.
1
Easiest · about 2 minutes · nothing to install
2
On a Mac · about 5 minutes
3
Windows or no internet · about 15 minutes the first time
Tick them off as you go. Everything below is in version 0.4.0 (the viewer part is in the live demo section above); if you were sent a demo link, ask which version it runs.
Right-click (or Control-click) the app and choose Open, then confirm. You only need to do this the first time.
Open Docker Desktop and wait until the status says it is running (this can take a minute). Then double-click the installer again.
Open your browser and go to http://127.0.0.1:8080. That address is only your own computer.
If you used the zip, open DEMO-LOGIN.txt in the unzipped folder. For a demo link, reply to our email and we will resend it.
Reload the page, or try another browser (Chrome, Safari or Edge). If it is still stuck, tell us what you clicked last.
Of course. Email sales@zyvor.dev and we will book ten minutes.
Live demo
Watch a thirty-second tour of the real application, then try the same thing on your own machine in about two minutes, or ask us for a hosted link. Every study is a made-up phantom.
Open the shared demo and press Fill in the demo login on its sign-in page. It signs you in as viewer. It is a read-only copy: you can open the phantom, use MPR and 3D and take measurements, but nothing you do is saved, and everything on it is made up. Please do not type anything private into it.
The login is also shown on that page. By hand: username viewer, password zyvor-live-demo-viewer-read-only-2026.
Needs Docker Desktop (Mac or Windows) running once. This starts the real application with made-up studies, including the phantom volume the viewer uses. Nothing leaves your computer.
Loading the command…
A new password is made in your browser each time you load this page; it is never sent anywhere. Prefer no terminal? Use the Mac app or the zip, or ask for a hosted link.
SYN-VOL → open it → press MPR / 3D.30 into the Tilt boxes, then press Reset.The advanced research viewer
New in 0.4.0. Written for this product, from the DICOM standard, in plain browser code: no third-party viewer inside. Research use only.
Axial, coronal and sagittal planes with a shared crosshair. Click or drag to move it; scroll to step through a plane.
Three linked planes, thick slabs, and all three planes tilted together about the patient axes. Checked against a phantom with known shapes and sizes.
Volume rendering, MIP, MinIP and average on WebGPU or WebGL2, with a clip box and your segments in colour. It drops to MIP by itself on a slow computer.
Several series side by side (1×1 to 2×2), placed by a hanging protocol, scrolling together by position in the patient.
Length, angle, ellipse, rectangle and probe with Hounsfield statistics; segments by region grow or threshold, then brush, eraser, fill holes, smoothing and undo. Saved with the series.
A Structured Report of measurements and segments, a DICOM Segmentation download, and Segmentation objects from elsewhere drawn over the images.
JPEG, JPEG-LS, JPEG 2000 and High-Throughput JPEG 2000 decoded in the browser by code written for this product.
On a made-up phantom with known answers, against images from independent encoders, and with the command-line tools of DCMTK and an independent DICOM validator.
No clinical validation. No comparison with other viewers. No third-party viewer has opened our Segmentation or Report files yet. Only synthetic data was used. Read the details.
Built for imaging workflows
Ingest from scanners over DIMSE or DICOMweb STOW. Calling-AE allowlists and TLS where you need them.
Worklist, research viewer and reporting with drafts, finalisation and amendments — not a diagnostic workstation claim.
Compose for a laptop, Helm for a cluster, or a k3s edge with HTTPS and organisation sign-in.
See it
Captured from the running application. Every name and number shown is synthetic.
A short tour: the picture grid, a name-differs badge that jumps to prior studies, the de-identified export preview, search and image checks.
What's new
New in version 0.4.0, released 25 October 2026.
Linked planes you can tilt, volume rendering with a clip box, and several series side by side with scrolling linked by position.
Measurements and segments saved with the series and editable: region grow or threshold, brush, eraser, fill holes, smoothing.
Download or store a Structured Report and a Segmentation object. Binary and fractional Segmentations from elsewhere are drawn over the images.
JPEG, JPEG-LS, JPEG 2000 and High-Throughput JPEG 2000 in the browser, checked against independent encoders.
The scanner gateway is exercised with DCMTK's echo, store, find, get and move tools, and Segmentation intake is fuzz-tested.
The Mac and Windows installers and the Mac app now say 0.4.0 and pull the 0.4.0 image.
Wrong-patient radar (a patient ID under two names gets a badge and a warning; one click shows the priors). Image checks for CT, MR and X-ray, and a warning when text looks burned into a picture: checks, not diagnoses. De-identified export as a pseudonymised research copy, single or grouped, not a certified de-identification. Filters and thumbnails by date and tag. Optional AI helpers (off by default; they need a model your IT team runs). Mac app with priors, the banner and export.
Capabilities
Byte-for-byte storage, UID hierarchy checks, duplicate handling and integrity scans on your volume or bucket.
STOW, QIDO, WADO, metadata, bulk data, frames and rendered PNG — for tools that speak the web APIs.
C-ECHO, C-STORE, Study Root C-FIND and C-MOVE with calling-AE allowlist; TLS for non-loopback listeners.
Named roles, optional TOTP, and OIDC authorization code with PKCE. Organisation sign-in on the k3s edge.
Security posture
Sessions hashed. Mutations origin-checked. Audit recorded. OIDC uses PKCE. See the private product docs for depth.
HttpOnly cookies, Secure when TLS terminates at the edge, SameSite, and hashed session identifiers on the server.
Authorization code with PKCE S256, state, nonce and verified RS256 claims. MFA claim requirement is configurable.
Immutable write path with checksum verification on read. Portable backup tooling for local and shared backends.
Deploy
Doctors start with the Mac DMG or a hosted demo. IT can run the binary zip, pull the image, or stand up a cluster edge.
Drag the notarized Mac app into Applications, or use Docker Desktop once with the air-gap zip — seeded synthetic studies. No GitHub token.
IT can pull the published GHCR image when the package is public, or use FluxVM on Apple silicon.
One command on your cluster: NodePort token login, or full Envoy + Keycloak HTTPS edge — both auto-seed synthetic studies.
Honest limits
Not validated for diagnostic decisions. No MPR/3D diagnostic workstation claim, and no comparison with other viewers.
Built and tested with synthetic data only. Do not load real patient studies into a public evaluation.
No clinical validation or regulatory certification claim (HIPAA, MDR, FDA, etc.).
FAQ
No. The browser viewer is a research viewer for evaluation. It is not a certified diagnostic workstation.
The product repository is private. This site is the public marketing surface and hosts evaluation releases — Mac DMG, binary zip, screenshots, and contact. No application source.
Ask sales for a hosted demo (browser only), download the Mac DMG (drag to Applications), or use the binary zip (Docker Desktop + double-click installer). Ignore automatic “Source code” archives. Passwords are never published here — sales sends logins privately.
Fastest: email sales for a hosted link. On a Mac: download the DMG → drag Zyvor Imaging into Applications → open → paste the Base URL and password from the sales email. Then test: Worklist → open a study → research viewer. Synthetic data only; not for diagnosis.
Not for the demo link: it opens in your browser. The Mac app is a normal drag-to-Applications install. Only the "run it on my own PC" option needs Docker Desktop, once.
Please do not. This is an engineering release tested with synthetic data. Use the sample studies, or ask us for more synthetic ones.
They are off unless your IT team connects a model they run. They help with search and wording, never with diagnosis, and nothing is saved until you press Insert. See the product documentation we send with your evaluation.
Use the "Send feedback" button in the test checklist, or email sales@zyvor.dev. Tell us what confused you; that is the most useful thing.
sales@zyvor.dev for evaluations · security@zyvor.dev for security.
Next step
Evaluations use synthetic data on infrastructure you approve.