Inferensys

Glossary

SCU (Service Class User)

The DICOM Application Entity (AE) that initiates a network operation by issuing a command request to a Service Class Provider (SCP) over an established association.
Strategy consultant facilitating AI use case discovery workshop, sticky notes on glass wall, casual corporate meeting.
DICOM NETWORKING ROLE

What is SCU (Service Class User)?

The SCU is the DICOM Application Entity that initiates a network operation by sending a service request to a peer entity, acting as the client in a DICOM association.

A Service Class User (SCU) is the DICOM Application Entity that actively initiates a network operation by transmitting a DIMSE or DICOMweb service request to a peer. It functions as the client in the DICOM client-server model, requesting operations such as storing images (C-STORE), querying a database (C-FIND), or retrieving prior studies (C-MOVE). A PACS workstation acting as an SCU queries a server for a patient's prior studies.

The SCU role is defined during Association Negotiation, where it proposes a set of SOP Classes and Transfer Syntaxes to the SCP. A single device can simultaneously serve as an SCU for one operation and an SCP for another. For example, a modality acts as an SCU when querying the Modality Worklist but as an SCP when receiving an MPPS confirmation.

SERVICE CLASS USER

Key Characteristics of a DICOM SCU

A DICOM Service Class User (SCU) is the Application Entity that initiates a network operation, playing the active client role in a DICOM association. Understanding the SCU's responsibilities is critical for integration engineers building imaging workflows.

01

The Active Initiator

The SCU is the client in a DICOM network transaction. It actively opens a TCP/IP Association to a listening SCP. The SCU's role is defined by the operation it requests: a PACS workstation acts as an SCU when it sends a C-FIND request to query a study, or a modality acts as an SCU when it uses a C-STORE command to push images to an archive. The SCU always sends the first command message after the association is established.

02

Association Negotiation

Before any operation, the SCU must propose a set of capabilities to the SCP. This handshake includes:

  • SOP Classes: The specific operations the SCU wants to perform (e.g., CT Image Storage).
  • Transfer Syntaxes: The encoding rules it supports (e.g., JPEG Lossless, Explicit VR Little Endian).
  • Maximum PDU Length: The largest data packet size it can receive. The SCP selects from these proposals, and the SCU must respect the negotiated terms for the session's duration.
03

Operation Semantics

The SCU's behavior is strictly defined by the DIMSE command it issues:

  • C-STORE SCU: Pushes a composite SOP Instance to an SCP. The SCU retains responsibility for the data until a successful response is received.
  • C-FIND SCU: Sends a query with matching keys and must be prepared to receive multiple, paginated responses from the SCP.
  • C-MOVE SCU: Requests the SCP to initiate a separate C-STORE sub-operation to a third-party Destination AE. The SCU here is instructing the SCP to act as a C-STORE SCU to another system.
  • C-ECHO SCU: Sends a simple verification request to test network connectivity and association capability.
04

Conformance Statement Mapping

An SCU's exact capabilities are documented in its DICOM Conformance Statement. This document is the definitive source for integration, detailing:

  • The complete list of SOP Class UIDs it can request as an SCU.
  • The specific Transfer Syntaxes it proposes for each SOP Class.
  • Any specializations or limitations, such as maximum query results or specific attribute handling. Integration engineers must compare the SCU's conformance statement with the SCP's to ensure a compatible intersection of supported services.
05

Real-World Example: Modality Worklist

A classic SCU scenario is a CT scanner querying a Modality Worklist SCP. The scanner (SCU) initiates a C-FIND request to the RIS broker (SCP) using the Scheduled Procedure Step SOP Class. It sends matching keys like Patient ID or Accession Number. The SCP returns the patient demographics and procedure details, which the SCU then uses to populate the DICOM header of the subsequent acquired images, eliminating manual technologist entry and reducing error rates.

06

SCU vs. SCP: Role Dynamics

A single Application Entity (AE) can simultaneously function as both an SCU and an SCP, depending on the operation. A PACS archive is an SCP for incoming C-STORE requests from modalities, but it acts as an SCU when it initiates a C-MOVE to forward studies to a reporting workstation. The roles are defined per-association, not per-system. This duality is fundamental to designing a DICOM network topology where data flows are initiated by the consumer.

DICOM NETWORKING

Frequently Asked Questions

Essential questions about the Service Class User (SCU) role in DICOM network communication, covering its function, configuration, and relationship with other Application Entities.

A DICOM Service Class User (SCU) is the Application Entity (AE) that actively initiates a network operation by sending a DICOM Message Service Element (DIMSE) command request to a peer Application Entity over an established TCP/IP association. The SCU acts as the client in the DICOM client-server model. Its operational lifecycle begins with Association Negotiation, where it proposes a set of supported SOP Classes and Transfer Syntaxes to the remote Service Class Provider (SCP). Once the association is accepted, the SCU transmits a specific DIMSE command—such as C-STORE to push images, C-FIND to query a database, or C-MOVE to retrieve studies—along with the relevant dataset. The SCU then waits for and processes the SCP's response, which includes a status code indicating success, failure, or a pending state with optional sub-operations. For example, a PACS workstation acting as an SCU sends a C-FIND-RQ at the Study level to query an archive for all prior chest CT exams for a specific patient, then parses the C-FIND-RSP messages to populate its local worklist. The SCU role is defined per operation; a single device can simultaneously function as an SCU for storage and an SCP for query requests.

DICOM ASSOCIATION ROLES

SCU vs. SCP: Client and Server Roles

A comparison of the initiating Service Class User and the responding Service Class Provider within a DICOM network association.

FeatureSCU (Service Class User)SCP (Service Class Provider)

Role Definition

Initiates the DICOM network operation

Listens for and performs the requested operation

Association Initiation

Sends the A-ASSOCIATE request

Responds with A-ASSOCIATE acceptance or rejection

Typical Deployment

Modality workstation, diagnostic viewer

PACS archive, VNA, modality server

C-STORE Operation

Pushes images to the archive

Receives and persists incoming images

C-FIND Operation

Queries the database for study lists

Searches the database and returns matches

C-MOVE Operation

Requests data transfer to a third AE

Retrieves data and initiates a C-STORE sub-operation

C-GET Operation

Requests direct data retrieval

Retrieves data and sends it over the same association

Modality Worklist Query

Scanner queries for patient schedule

HIS/RIS broker returns scheduled procedures

Association Release

Initiates the A-RELEASE request

Responds to and confirms the release

Error Handling

Receives and processes C-STORE responses

Generates status codes (Success, Warning, Failure)

Conformance Statement Role

Declared as SCU for each supported SOP Class

Declared as SCP for each supported SOP Class

Real-World Example

CT scanner sending images to PACS

PACS archive storing images from CT scanner

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.