DICOM — Digital Imaging and Communications in Medicine
Overview
DICOM is the standard for storing, transmitting, and displaying medical imaging data. Developed jointly by the ACR (American College of Radiology) and NEMA (National Electrical Manufacturers Association) from 1983, DICOM defines both the file format for medical images and the network protocol for communicating between imaging devices, PACS systems, and clinical applications. Imaging data at neuroimaging research centres typically originates as DICOM and is converted to NIfTI and organised into BIDS for research use.
DICOM Object Model
A DICOM dataset is organised hierarchically:
- Patient → Study → Series → Instance (Image)
- Each level has a unique identifier (Patient ID, Study Instance UID, Series Instance UID, SOP Instance UID)
- A DICOM file = one or more pixel data frames + a structured header of metadata tags
- Tags are identified by (Group, Element) pairs, e.g. (0010,0020) = Patient ID, (0008,0060) = Modality (MR, PT, CT…)
Key SOP Classes (Modalities)
DICOM defines Service-Object Pair (SOP) Classes for each imaging modality and service type:
| SOP Class | Description |
|---|---|
| MR Image Storage | Structural/functional MRI |
| Enhanced MR Image | Multi-frame MRI (fMRI, diffusion) |
| PET Image Storage | PET molecular imaging |
| Secondary Capture | Screenshots, processed images |
| CT Image Storage | Computed tomography |
| Digital X-Ray | Radiography |
| Encapsulated PDF | PDF reports embedded in DICOM |
| Structured Report (SR) | Machine-readable clinical reports |
| Segmentation | Label maps, segmentations |
| Parametric Map | Quantitative maps (ADC, T1 maps) |
DICOM Services (Network)
Beyond file storage, DICOM defines network services for clinical workflow:
- C-STORE sends images from a modality to a PACS system.
- C-FIND queries and retrieves metadata from a PACS.
- C-MOVE retrieves images from a remote PACS.
- C-GET retrieves images directly to the requesting system.
- WADO-RS (Web Access to DICOM Objects) is the modern RESTful image retrieval protocol used by the HL7 FHIR ImagingStudy resource.
DICOM to Research Pipeline
At research sites, DICOM files are typically converted to research formats via standardised tools:
Scanner → DICOM (raw) → dcm2niix → NIfTI + JSON sidecar → BIDS
→ heudiconv (alternative converter with automatic naming)
Key tools include dcm2niix (the primary DICOM-to-NIfTI converter), heudiconv (a flexible DICOM organiser for automated BIDS conversion), Orthanc (an open-source DICOM server for local PACS and research DICOM management), and XNAT (an imaging data management platform built on DICOM).
DICOM Structured Reports (SR)
DICOM SR allows machine-readable, coded clinical findings to be embedded directly in the DICOM object alongside images. Used for radiology reports with coded measurements, AI/CAD findings (lesion measurements, segmentation results), and HL7 FHIR DiagnosticReport which can link to DICOM SR objects via WADO-RS.
Connections
- Converted to: NIfTI (standard research image format)
- Compatible with: HL7 FHIR (ImagingStudy and DiagnosticReport resources via WADO-RS)
Resources
- https://www.dicomstandard.org
- https://www.dicomstandard.org/current (current standard — Part 1–20)
- https://github.com/rordenlab/dcm2niix (dcm2niix converter)
- https://heudiconv.readthedocs.io (heudiconv BIDS converter)
- https://www.orthanc-server.com (Orthanc DICOM server)