AIHealth
On-premise clinical platform with local LLMs, RAG on FHIR/DICOM data, diagnostic support, remote follow-up. Architecture designed for the MDR pathway.
Discover AIHealth →
Digital Health
Medical software development compliant with CE and MDR regulatory standards. Clinical decision support systems, AI integration in clinical workflows.
Discover →Displaying DICOM in the browser
A central piece of clinical imaging is the viewer — the software with which radiologists, clinicians, and researchers view diagnostic images. Technical requirements are sharply different from the PACS: fluid interaction with large image series, contrast manipulation (window/level), measurement tools, annotations, cross-series navigation, multimodal fusion, 3D rendering.
The historical viewer landscape has been dominated by installable desktop clients (OsiriX/Horos on macOS, RadiAnt on Windows, ClearCanvas, commercial PACS viewers by Philips/GE/Siemens) and by Java/ActiveX clients embedded in web-based PACS.
With modern browsers maturing — Canvas 2D, WebGL, WebAssembly — it has become technically feasible to build a fully browser-based DICOM viewer, zero-footprint: the user opens a URL, the viewer loads, images scroll. No installation, automatic updates, consultable from any workstation and any operating system. For research, collaborator sharing, hub-and-spoke second-opinion patterns, the operational difference is substantial.
The project that has brought this paradigm to maturity is OHIF Viewer.
OHIF — Open Health Imaging Foundation
OHIF — Open Health Imaging Foundation — was born in 2013-2014 at Massachusetts General Hospital (Harvard Medical School), in Gordon Harris’s group (Director of the Tumor Imaging Metrics Core) with collaborators from Brigham and Women’s Hospital, the 3D Slicer community, and other institutions. Stated mission: build an open source infrastructure for the entire medical imaging lifecycle: visualisation, annotation, analysis, sharing.
OHIF Viewer is the Foundation’s main product: an open source JavaScript DICOM viewer, browser-based, distributed under the MIT licence.
Cornerstone
OHIF Viewer is built atop Cornerstone.js, an open source JavaScript library started in 2014 by Chris Hafey (MGH/Harvard) as a DICOM rendering engine for browsers. Cornerstone is organised in components:
- Cornerstone Core — DICOM image rendering on HTML5 Canvas, window/level management, pan/zoom, multi-layer
- Cornerstone Tools — interactive tools: length/angle/area measurements, annotations, ROI overlays, slice-by-slice scrolling
- dicomParser — low-level DICOM file parser in JavaScript
- cornerstone-wado-image-loader — DICOM loading via WADO-URI (HTTP GET)
Cornerstone is independent of OHIF and used by numerous other viewers and commercial products. MIT licence.
OHIF Viewer architecture
OHIF Viewer is a React SPA (Single Page Application) web app:
- Layout — configurable panels (image viewports, study panel, toolbar, measurements panel)
- Viewports — Cornerstone instances, synchronised or independent for multi-series
- Extension system — plug-in module for adding functionality (custom viewers, external service integration, tools)
- Mode system (introduced in the v3 in development) — preset configurations for specific use cases (mammography, oncology, cardiology)
Version 2 (stable)
OHIF v2 is the current production series as of 2020. Based on Meteor.js in early versions, migrated to standalone React. Provides:
- MPR (multi-planar reconstruction) navigation for volumes
- Linear, angular, elliptical, area measurements
- Textual annotations and free-form ROI
- Overlay of segmentations in RTSS (DICOM RT Structure Set) and DICOM SEG formats
- Integration with WADO-URI, WADO-RS (DICOMweb), local storage
- Primitive hanging protocols
Version 3 (alpha/beta)
OHIF v3 is in active development; alpha and beta releases are available during 2020 with stable release planned for 2021-2022. Structural novelties:
- Full rewrite in TypeScript
- Modular architecture of services (studyService, toolbarService, viewportService, etc.)
- Upgrade to Cornerstone3D — next-generation Cornerstone with GPU support via WebGL/WebGPU, native volume rendering, oblique slice navigation
- Structured Mode system
- Broader support for advanced imaging (fusion, digital mammography with per-quadrant navigation, digital pathology)
DICOMweb
OHIF Viewer is DICOMweb-native (DICOM PS3.18):
- QIDO-RS (Query based on ID for DICOM Objects) — study/series/instance search via HTTP
- WADO-RS (Web Access to DICOM Objects — Restful Services) — metadata and pixel data retrieval via HTTP
- WADO-URI — direct access to individual instances
- STOW-RS — upload of new instances (less used in viewer scenarios)
This lets OHIF connect to any DICOMweb-compatible archive: Orthanc (open source), dcm4chee (open source), commercial PACS exposing DICOMweb, cloud services (Google Cloud Healthcare API, AWS HealthImaging, Azure Health Data Services).
The typical pattern is: Orthanc as archive + OHIF Viewer as interface = a fully open source PACS at zero licence cost, easily deployable for research, clinical trials, small centres.
Extensions and MONAI
OHIF supports extensions via npm packages. As of 2020, the most relevant:
- cornerstone-tools-extension — extended set of measurement and annotation tools
- vtk-extension — integration with VTK.js for 3D/MPR rendering
- dicom-segmentation-extension — loading of DICOM SEG and RTSS segmentations
- dicom-microscopy-extension — viewer for whole slide images in digital pathology
- MONAI Label integration (in development) — AI-assisted annotation in clinical loop, bridging OHIF and deep learning pipelines via REST API
MONAI integration is one of the most promising developments: a radiologist opens a study in OHIF, activates an “AI-segment” tool, the browser sends the volume to a MONAI service that returns an initial segmentation, the radiologist refines interactively. The pattern combines browser-based visualisation with server-based AI, without requiring local installations.
Use scenarios
OHIF Viewer is in operational use in several areas:
Clinical research
Research groups deploy OHIF as consultation interface for imaging cohorts of observational studies. The viewer accesses data via DICOMweb without local copies, preserving tracking and audit.
Clinical trials
The Imaging Core Lab — specialist unit in central imaging evaluation for pharmaceutical trials — is a key use case. OHIF Viewer lets external reviewers assess study imaging under predefined protocols, with exportable structured annotations.
Second opinion and telereporting
Remote-review scenarios: imaging produced at one site is viewed at another via web link, without reviewer-side installation.
Teaching
Medical schools and specialist training use OHIF to distribute teaching cases.
Enterprise PACS integration
Some hospitals have adopted OHIF as primary or secondary PACS viewer, alongside commercial viewers; especially for smartphone/tablet consultation.
TCIA and public dataset integration
The Cancer Imaging Archive (TCIA), the largest public oncology imaging archive, uses OHIF as integrated viewer for public dataset consultation.
Certification and clinical use
Base OHIF Viewer is not a certified medical device for primary diagnostic use. Anyone wanting to use OHIF for primary reporting must submit their deployment to medical device assessment — a full technical and regulatory path (IEC 62304, ISO 14971, CE mark or FDA clearance).
Some commercial OHIF-based implementations have obtained certifications: Radiology Partners, Blackford Analysis, vendors offering OHIF-derived products with complete regulatory documentation.
For research, secondary review, non-diagnostic consultation, base OHIF is widely adopted without regulatory friction.
Comparable ecosystem
OHIF sits in an ecosystem of open source DICOM viewers:
- Weasis (weasis.org) — Java-based desktop viewer, developed at CHU Liège (Nicolas Roduit), mature, used at several European hospitals
- Horos — open source fork of OsiriX, macOS desktop
- Slim — OHIF-like viewer from MGH, focus on digital pathology
- Cornerstone-demo — Cornerstone demo application, not a complete applicative viewer
- AMI (AMI Medical Imaging) — WebGL library for VR/AR of medical volumes
- Niivue (emerging) — browser-based viewer for neuroimaging
OHIF stands out for: modern stack (React/TypeScript), community activity, full DICOMweb support, integration with AI ecosystems.
In the Italian context
In Italy as of 2020 operational OHIF uptake is still limited but growing:
- IRCCS and universities managing clinical trials use it as consultation viewer
- Research projects in imaging use OHIF + Orthanc as light PACS
- Some hospitals have experimented with OHIF as complementary viewer for remote consultation
- Italian imaging startups embed OHIF as frontend for AI products
Italian commercial PACS vendors (Ebit, TeamOps, Dedalus) handle most daily reporting, but OHIF’s growth as embedded component in special solutions is tangible.
Outlook
Development directions in coming years:
- OHIF v3 stable release with Cornerstone3D — expected 2021-2022
- MONAI Label integration — AI-in-the-loop annotation tool mainstream
- Pathology — OHIF is gaining whole slide image capabilities, converging with digital pathology tools
- Structured reporting — DICOM SR integration for in-viewer reporting
- Accessibility — full support for visually impaired users, a growing regulatory requirement
- Federated viewing — viewing imaging from multiple sites in a single viewer, useful for second-opinion and multi-centre research
- FHIR integration — connecting the viewer to FHIR records for patient context fetching
OHIF Viewer in 2020 represents the most mature open source viewer for web-based scenarios. The emergence of Cornerstone3D as the next-generation rendering engine, combined with AI integrations (MONAI Label) and open source PACS (Orthanc), sets the stage for a complete open source clinical imaging ecosystem — from acquisition to rendering, from preservation to AI, from annotation to reporting — with permissive licences and growing maturity.
References: OHIF Viewer (ohif.org), Open Health Imaging Foundation. Massachusetts General Hospital / Harvard Medical School. Cornerstone.js (cornerstonejs.org). MIT licence. DICOMweb (DICOM PS3.18: QIDO-RS, WADO-RS, STOW-RS). Integration with Orthanc, dcm4chee, cloud healthcare APIs.