DICOM RT Structure Set

Stores regions of interest as sequences of 3D coordinate points defining closed planar contours on individual image slices. Each structure is composed of multiple contour sequences that collectively define a volumetric shape.

• Geometric Type: Points, open polylines, or closed planar contours
• Coordinate System: Patient-based 3D coordinates referenced to the image position
• Interpolation: Volumetric rendering requires interpolation between sparse slice-based contours

Each region of interest is identified by a unique ROI Number and described using a ROI Name and optional ROI Description. The ROI Observation Class specifies the semantic category.

• ROI Number: Unique integer identifier within the structure set
• ROI Name: Human-readable label such as 'GTV_Primary' or 'Spinal_Cord'
• Observation Class: Defines whether the structure is an organ-at-risk, target volume, or anatomical reference
• Generation Algorithm: Records whether contours were manually drawn, auto-segmented, or atlas-based

Each contour is spatially registered to a specific series of DICOM images through Referenced Frame of Reference and Referenced Image Sequence attributes. This ensures contours align precisely with the underlying voxel data.

• Frame of Reference UID: Uniquely identifies the coordinate system
• Contour Image Sequence: Lists the specific SOP Instance UIDs for each slice containing a contour
• Spatial Registration: Enables fusion with other imaging modalities via rigid or deformable registration objects

Optional attributes store derived quantitative metrics for each structure, supporting treatment planning calculations and clinical documentation.

• ROI Volume: Computed volume in cubic centimeters
• ROI Mean/Max/Min Intensity: Statistical descriptors of the underlying image values within the contour
• ROI Standard Deviation: Quantifies tissue heterogeneity
• ROI Generation Algorithm: Records whether contours were manually drawn, auto-segmented, or atlas-based

The Structure Set Label and Approval Status attributes track the clinical workflow state, indicating whether contours have been reviewed, approved, or are still in draft.

• Approval Status: Values include 'UNAPPROVED', 'APPROVED', or 'REJECTED'
• Structure Set Date/Time: Timestamp of creation or last modification
• Reviewer Name: Identifies the clinician responsible for contour validation
• Structure Set Description: Free-text field for workflow notes and clinical context

RT Structure Sets serve as the primary input for Treatment Planning Systems and Record and Verify Systems, enabling dose calculation, plan optimization, and machine delivery.

• DICOM RT Plan: References structure sets to define beam geometry and dose constraints
• DICOM RT Dose: Stores calculated dose distributions mapped to the same frame of reference
• DICOM RT Image: Portal images and setup verification referenced against planned structures
• Cross-Vendor Compatibility: Standardized encoding ensures contours transfer between systems from different manufacturers

The delineation of healthy anatomical structures surrounding a tumor that are sensitive to radiation dose. OARs are stored as individual regions of interest within the DICOM RT Structure Set and are critical for minimizing collateral damage during treatment planning.

• Common OARs: spinal cord, parotid glands, optic chiasm, heart
• Dose-volume histograms (DVHs) are computed directly from these contours
• Automatic OAR segmentation reduces inter-physicist variability

The macroscopic extent of a malignant tumor as visible on imaging or clinical examination. The GTV is the primary target volume defined in the DICOM RT Structure Set and serves as the foundation for expanding clinical and planning target volumes.

• Defined using CT, MRI, or PET fusion
• Stored as a 3D contour sequence in the structure set
• GTV delineation accuracy directly impacts tumor control probability

A statistical measure of spatial overlap between a predicted segmentation mask and the ground truth annotation. The Dice Score is the most common metric for evaluating automatic OAR and GTV segmentation algorithms before their contours are exported to a DICOM RT Structure Set.

• Formula: 2 * |A ∩ B| / (|A| + |B|)
• Ranges from 0 (no overlap) to 1 (perfect agreement)
• A Dice score above 0.7 is generally considered acceptable for clinical use

A boundary-based metric that measures the maximum distance from any point in one set to the nearest point in the other. The Hausdorff Distance quantifies the worst-case segmentation boundary error, which is critical for radiotherapy where a single large deviation can cause a geographic miss of the tumor.

• 95th percentile Hausdorff Distance (HD95) is used to exclude outliers
• Measured in millimeters on the original image grid
• Sensitive to small, spurious false positive predictions

A computer graphics algorithm that extracts a polygonal mesh of an isosurface from a 3D scalar field. Marching Cubes is commonly used to generate surface renderings from the binary volumetric masks stored in or derived from a DICOM RT Structure Set for visualization and 3D printing of patient-specific anatomy.

• Operates on a voxel-by-voxel basis using a lookup table
• Generates triangle meshes suitable for rendering
• Used in treatment planning systems for 3D dose visualization

The process of interpolating a volumetric medical image to achieve uniform voxel spacing in all three spatial dimensions. Isotropic resampling is a critical preprocessing step before automatic segmentation because DICOM RT Structure Sets require contours defined on the original acquisition grid, and networks perform best with consistent spatial context.

• Typical target resolution: 1mm x 1mm x 1mm
• Uses spline or linear interpolation
• Prevents anisotropic feature distortion in 3D CNNs