
DICOM3D v0.1.0 — Initial Public Release
First open-source release of DICOM3D, a research-oriented 3D medical imaging viewer for patients, clinicians, developers, and researchers.
Features:
Interactive Babylon Lite/WebGPU 3D visualization.
Local GLB and semantic manifest loading.
Structure selection, camera controls, and visibility toggling.
Native InterSystems IRIS for Health integration.
IPM installation and Docker development deployment.
Verified installation, removal, reinstallation, and restart persistence.
Apache License 2.0.
Note: This is an early research presentation viewer, not a clinically validated diagnostic application. Imaging reconstruction and AI capabilities remain under development.
Source and installation instructions: https://github.com/KennethJSmithDev/DICOM3D
DICOM3D is an open development project for preserving the meaning and spatial context of anatomical structures reconstructed from medical imaging. It separates source DICOM reality, semantic and spatial metadata, portable geometry, and renderer-owned presentation.
v0.1.0 is the first installable IRIS for Health development release. The exact Apache-2.0 package passed install, resource enforcement, authenticated browser/API, restart persistence, uninstall/absence and same-artifact reinstall in two isolated Docker targets. See Docker quick start and notice audit. Native Windows 0.0.2 viewer/API qualification remains separate; native Windows IPM support is unverified. Research presentation only; not intended for clinical use.
The current viewer is a small Babylon Lite/WebGPU development shell. It can load a manifest and matching GLB selected locally; the repository does not bundle a patient-derived example. The H000 example was derived from a qualified public research study. Redistribution of derived example assets remains a separate decision.
Presentation transforms, camera, material, lighting, and transient renderer-generated normals do not become source or clinical truth. Current DICOM3D Model Space v1 qualification is limited to one BIPED study and one semantic structure. It does not establish multi-structure composition, anatomical outward-normal correctness, clinical validity, or source-slice correlation.
Requires Node.js and npm. From viewer/:
npm ci
npm run dev
Open the local Vite URL printed by the command. Choose a manifest JSON file first, then its corresponding GLB. The viewer verifies the GLB SHA-256 against the manifest before loading it. The browser must support WebGPU. To produce the static development build, run npm run build.
For the native development host, see NATIVE_DEV_INSTALL.md. The authenticated entry is /dicom3d/DICOM3D.Entry.cls; the read-only identity endpoint is /dicom3d-api/health. Imaging files remain local browser inputs. No OpsDeck installation is needed.
0.0.1: public product baseline.0.0.x: bounded public development milestones before the first installable release.0.1.0: first directly installable IRIS for Health development release, including IPM install/uninstall/reinstall lifecycle and a Docker clean-room quick start.1.0.0: stable, independently installable, documented, reproducible product contract.OpsDeck is an implementation donor for packaging and representation-measurement methods. It is not a DICOM3D runtime dependency. See https://github.com/KennethJSmithDev/DICOM3D/blob/main/ROADMAP.md and https://github.com/KennethJSmithDev/DICOM3D/blob/main/docs/ARCHITECTURE.md.
DICOM3D is licensed under Apache-2.0. Copyright 2026 KennethJSmithDev, personally. See LICENSE, NOTICE and https://github.com/KennethJSmithDev/DICOM3D/blob/main/docs/NOTICE_AUDIT.md. Native64 World and its unreleased engine are excluded. Use GitHub Issues for questions, feedback, and proposed changes.