Inferensys

Glossary

WADO-RS

Web Access to DICOM Persistent Objects by RESTful Services (WADO-RS) is a DICOMweb standard enabling retrieval of DICOM instances, frames, or metadata over HTTP using RESTful protocols.
Stylish WeWork-like workspace with hot desks and document wall, professional searching through enterprise knowledge base on a mounted ultrawide display, warm industrial pendants overhead.
DICOMweb RESTful Retrieval

What is WADO-RS?

WADO-RS (Web Access to DICOM Persistent Objects - RESTful Services) is a DICOMweb standard that defines an HTTP-based protocol for retrieving DICOM instances, frames, and metadata using RESTful web services.

WADO-RS is a core component of the DICOMweb suite of standards, enabling the retrieval of medical imaging objects over HTTP/HTTPS using RESTful conventions. Unlike the legacy DIMSE C-MOVE protocol, WADO-RS uses standard GET requests with content negotiation via Accept headers to return DICOM data in application/dicom+xml, application/dicom+json, or raw binary formats.

A WADO-RS service exposes resources at a base URL endpoint and supports retrieving entire Study, Series, or Instance objects, as well as individual frames from multi-frame images. This stateless, cacheable architecture simplifies integration with web-based PACS viewers and VNA systems, allowing developers to use standard HTTP libraries instead of specialized DICOM networking stacks.

DICOMweb RESTful Services

Key Features of WADO-RS

WADO-RS (Web Access to DICOM Persistent Objects - RESTful Services) is a core DICOMweb protocol that enables the retrieval of DICOM instances, frames, and metadata over HTTP using standard RESTful conventions and media types.

01

RESTful Resource-Oriented Architecture

WADO-RS models DICOM data as uniquely addressable web resources using a hierarchical URI structure. Each Study, Series, and Instance is accessed via a predictable path, enabling standard HTTP caching, proxying, and load-balancing infrastructure to be used without custom DICOM networking stacks.

  • URI Pattern: {service}/studies/{StudyInstanceUID}/series/{SeriesInstanceUID}/instances/{SOPInstanceUID}
  • HTTP Methods: Uses GET for retrieval, supporting conditional requests via ETags and If-Modified-Since headers
  • Statelessness: Each request is independent, eliminating the need for persistent DICOM associations
02

Multiple Media Type Support

WADO-RS supports content negotiation through HTTP Accept headers, allowing clients to request DICOM data in the format most suitable for their processing pipeline. This decouples the storage format from the delivery format.

  • application/dicom: Returns the raw binary DICOM Part 10 file for full-fidelity transfer
  • application/dicom+xml: Returns native DICOM data encoded as XML per Part 19, ideal for XSLT processing
  • application/dicom+json: Returns DICOM data as JSON per Part 18 Annex F, optimized for web application consumption
  • Transfer Syntax Conversion: The server may transcode the pixel data to a requested transfer syntax, such as JPEG-LS Lossless to JPEG 2000 Lossless
03

Frame-Level Pixel Retrieval

A critical feature for multi-frame DICOM instances such as cine loops, whole slide images, and dynamic contrast-enhanced series. WADO-RS allows clients to request individual frames or ranges of frames without retrieving the entire multi-gigabyte object.

  • Frame URI: Append /frames/{frameNumber} to the instance URI to retrieve a single frame
  • Byte Range Support: Uses HTTP Range headers to request specific byte segments of a single frame for progressive rendering
  • Accept Header Control: Clients can request frames transcoded to a specific media type, e.g., image/jpeg for a thumbnail preview, reducing bandwidth for viewing clients
04

Metadata-Only Retrieval

WADO-RS enables the retrieval of DICOM metadata without the bulk pixel data, a critical optimization for study list queries, radiation dose reports, and structured report parsing. This is achieved by requesting the instance resource with an application/dicom+json or application/dicom+xml Accept header and the server omitting the 7FE0,0010 Pixel Data element.

  • Reduced Payload: A metadata-only response for a CT study may be kilobytes instead of hundreds of megabytes
  • Bulk Data URIs: When metadata is returned, large attributes like pixel data and waveforms are replaced with BulkDataURI references that can be fetched separately on demand
  • Use Case: Enables rapid dashboard population of patient demographics and acquisition parameters without saturating network bandwidth
05

Retrieve Transaction with Rendered Media

WADO-RS supports a dedicated endpoint at /rendered that allows a client to request a server-side rendered representation of a DICOM instance, frame, or thumbnail. The server performs the rendering and returns a consumer image format.

  • Rendered URI: {service}/studies/{StudyInstanceUID}/series/{SeriesInstanceUID}/instances/{SOPInstanceUID}/rendered
  • Query Parameters: Supports viewport for window center/width, annotation to toggle burned-in patient info, and quality for JPEG compression level
  • Thumbnail Endpoint: A dedicated /thumbnail sub-resource returns a small, fixed-size representation for efficient image strip navigation
06

Standard HTTP Security Integration

Unlike the legacy DIMSE protocol which requires complex IPSec or TLS tunneling configurations, WADO-RS inherits the mature security ecosystem of HTTPS. This simplifies compliance with HIPAA and GDPR mandates for data in transit.

  • Transport Security: Standard TLS 1.3 encryption with mutual certificate authentication
  • Authorization: Integrates with OAuth 2.0 and OpenID Connect bearer tokens for fine-grained access control to specific studies
  • Audit Logging: Standard HTTP server access logs capture every retrieval transaction, simplifying the DICOM Audit Trail (ATNA) integration profile requirements
DICOM RETRIEVAL PROTOCOL COMPARISON

WADO-RS vs. WADO-URI vs. DIMSE C-MOVE

A technical comparison of the three primary DICOM retrieval mechanisms: the RESTful WADO-RS, the legacy HTTP-based WADO-URI, and the traditional DIMSE C-MOVE network command.

FeatureWADO-RSWADO-URIDIMSE C-MOVE

Protocol

RESTful HTTP/HTTPS (DICOMweb)

HTTP/HTTPS GET

DICOM over TCP/IP (Port 104/11112)

DICOM Standard Reference

Part 18

Part 18 (Legacy)

Part 7

Request Method

GET

GET

C-MOVE Command

Response Media Types

application/dicom+xml, application/dicom+json, multipart/related

application/dicom

DICOM binary (native encoding)

Firewall Friendly

Requires Association Negotiation

Supports Bulk Data Retrieval

Supports Frame-Level Access

Supports Metadata-Only Retrieval

Requires Pre-Configured AE Titles

Typical Latency Overhead

< 50 ms

< 30 ms

100-500 ms (association setup)

Modern Web Integration

WADO-RS EXPLAINED

Frequently Asked Questions

Clear, technical answers to the most common questions about the DICOMweb WADO-RS protocol for retrieving medical imaging data over HTTP.

WADO-RS, or Web Access to DICOM Persistent Objects via RESTful Services, is a DICOMweb standard defined in DICOM Part 18 that enables the retrieval of DICOM instances, frames, and metadata over HTTP using RESTful principles. Unlike its predecessor WADO-URI, which relies on query-string parameters, WADO-RS uses clean, resource-oriented URLs to address specific studies, series, and instances. A client sends an HTTP GET request to a URI path like /studies/{StudyInstanceUID}/series/{SeriesInstanceUID}/instances/{SOPInstanceUID}, and the server responds with the requested DICOM object. The protocol supports multiple media types, including application/dicom+xml for metadata, application/dicom+json for JSON-formatted metadata, and multipart/related for bulk binary retrievals. WADO-RS is the cornerstone of modern, web-based medical imaging interoperability, replacing legacy DIMSE C-MOVE commands with a firewall-friendly, cacheable HTTP architecture.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.