DICOM Structured Reporting (DICOM SR) and TID 1500 Guide

DICOM Structured Reporting (DICOM SR) is the part of the Digital Imaging and Communications in Medicine (DICOM) standard designed to encode, store, query, and transmit machine-readable clinical imaging reports, quantitative measurements, spatial coordinates, and computer-aided detection or radiology artificial intelligence outputs. Unlike traditional narrative radiology reports stored as unstructured text or static PDF attachments, a DICOM SR object is a self-contained DICOM SOP Instance containing a hierarchical tree of coded concepts linked directly to the exact pixel coordinates and SOP Instance UIDs of the source images.

Because DICOM SR objects share the same Patient, Study, and General Equipment header modules as standard CT, MR, or PET images, they can be archived directly in a PACS system or Vendor-Neutral Archive (VNA), queried and retrieved over DICOMweb (QIDO-RS and WADO-RS), and rendered as interactive overlays inside a DICOM viewer.

What Is DICOM Structured Reporting (DICOM SR) and How Is the Content Tree Organized?

Introduced into the DICOM standard via Supplement 23, DICOM SR represents clinical findings and quantitative measurements as a directed acyclic graph (or hierarchical tree) of Content Items rooted at the Content Sequence (0040,A730) attribute. Every node in the tree pairs a Concept Name Code Sequence (encoded as a triplicate of Code Value, Coding Scheme Designator such as SCT for SNOMED CT, LN for LOINC, or DCM for DICOM Controlled Terminology, and Code Meaning) with a typed value and a formal parent-child relationship (such as CONTAINS, HAS OBS CONTEXT, HAS CONCEPT MOD, or INFERRED FROM).

Core DICOM SR Value Types and SOP Classes

The DICOM standard defines escalating SR SOP Classes based on which Content Item value types they permit:

Value Type	Permitted in SOP Class	Encoded Data & Research Imaging Use Case
CONTAINER, TEXT, CODE	Basic Text SR, Enhanced SR, Comprehensive SR	Section headers, narrative observations, and coded anatomical locations or qualitative findings (SNOMED CT, RadLex, LOINC).
NUM	Enhanced SR, Comprehensive SR, Comprehensive 3D SR	Floating-point numerical measurements paired with Unified Code for Units of Measure (UCUM) units (e.g., lesion diameter in mm or SUVmax).
IMAGE, WAVEFORM, COMPOSITE	Basic Text SR, Enhanced SR, Comprehensive SR	Direct SOP Class / SOP Instance UID references to source slices, segmentation masks, or physiological waveforms.
SCOORD	Enhanced SR, Comprehensive SR, Comprehensive 3D SR	2D pixel coordinates (POINT, POLYLINE, CIRCLE, ELLIPSE) paired with an IMAGE reference via SELECTED FROM.
SCOORD3D	Comprehensive 3D SR	3D patient-space millimeter coordinates referenced directly to a DICOM Frame of Reference UID across volumetric series.

In addition to general reporting classes, DICOM defines specialized SR SOP Classes such as the X-Ray Radiation Dose SR and Radiopharmaceutical Radiation Dose SR, which imaging modalities generate automatically to record organ dose and CT Dose Index (CTDIvol / DLP) metrics for radiation safety registries.

How Does DICOM TID 1500 Encode AI Imaging Measurements and ROIs?

To prevent different imaging workstations and AI algorithms from inventing incompatible tree layouts, DICOM PS3.16 defines normative Template IDs (TIDs). For quantitative imaging biomarkers, oncology clinical trials (such as RECIST 1.1 target lesion tracking), and radiology AI inference pipelines, the primary standard is TID 1500 (Measurement Report).

A TID 1500 report organizes quantitative findings using a modular sub-template hierarchy:

• Procedure Context & Image Library (TID 1600): Catalogs the acquisition parameters (modality, slice thickness, contrast agent, Frame of Reference UID) of all source images analyzed by the reader or AI model.
• Observer Context (TID 1001–1004): Distinguishes human radiologist measurements from automated algorithm outputs by recording the Device Observer UID, software name, model version, and manufacturer.
• Imaging Measurements & Measurement Groups (TID 1410 / TID 1411): Groups all measurements belonging to a single anatomical lesion or Region of Interest (ROI). Each group assigns a Tracking Identifier (human-readable label, e.g., "Target Lesion 1") and a globally unique Tracking Unique Identifier (UID) so that the exact same lesion can be tracked longitudinally across baseline and follow-up timepoints.
• Spatial Definition of the ROI: Each measurement group links its derived numerical values (mean attenuation in Hounsfield Units, volume in mm3, or longest axial diameter) to either geometric 2D/3D coordinates (SCOORD / SCOORD3D) or a referenced DICOM Segmentation (SEG) binary/fractional mask SOP Instance.

Open-source toolkits such as highdicom (Python) and dcmqi (DICOM for Quantitative Imaging, developed under the NCI Quantitative Imaging Network) allow researchers to convert NIfTI or ITK segmentation masks and radiomics tables directly into standards-compliant TID 1500 DICOM SR and DICOM SEG files.

How Does DICOM SR Map to HL7 FHIR Observation and DiagnosticReport?

While PACS and imaging archives store DICOM SR natively, enterprise EHRs and clinical research warehouses typically consume clinical findings as HL7 FHIR resources. The HL7 / DICOM Imaging Integration Work Group (including the DICOM SR to FHIR Mapping Implementation Guide and IHE Integrated Reporting workflows) defines the canonical structural correspondence:

• Root DICOM SR Document → FHIR DiagnosticReport: The top-level SR instance metadata (Patient, Accession Number, Study Instance UID via ImagingStudy reference, completion/verification status, and verifying observer) maps to a FHIR DiagnosticReport resource.
• TID 1500 Measurement Group (TID 1410/1411) → FHIR Observation ( with bodySite, focus, and component / hasMember): Each tracked lesion or ROI Measurement Group maps to a parent FHIR Observation carrying the Tracking UID as an identifier, SNOMED CT anatomical site in Observation.bodySite, a reference to the source ImagingStudy in partOf or derivedFrom, and individual quantitative metrics (`NUM` items with UCUM units) encoded either as child Observation resources or Observation.component entries.

Frequently Asked Questions

What is the difference between DICOM Encapsulated PDF and DICOM Structured Reporting?

A DICOM Encapsulated PDF wraps a rendered visual document inside a DICOM header so it can be stored in PACS, but its internal measurements and findings remain opaque to software queries. A DICOM SR encodes every finding, anatomical site, numerical measurement, UCUM unit, and pixel coordinate as discrete coded Content Items that can be parsed, indexed, and mapped to FHIR or OMOP without natural language processing.

Why do radiology AI vendors use TID 1500 together with DICOM SEG?

DICOM SEG stores the pixel-level or voxel-level segmentation mask of an anatomical structure or lesion, whereas DICOM SR (TID 1500) stores the quantitative measurements (such as volume, probability score, or density) derived from that mask and links them back to the DICOM SEG SOP Instance UID via the INFERRED FROM relationship. Together, PACS viewers can render both the colored contour overlay and the structured measurement table.

How are longitudinal tumor measurements linked across serial scans in DICOM SR?

Within TID 1500 (specifically sub-template TID 1410/1411), every lesion is assigned a persistent Tracking Unique Identifier (0040,A124) (a DICOM UID) alongside a human-readable Tracking Identifier (0040,A123). When a follow-up scan is evaluated weeks later, reusing the same Tracking UID allows oncology workstations and trial databases to compute percent change in tumor burden automatically.