DICOM Height Map Segmentation is intended to be applicable to a broad variety of imaging domains, but its initial use case is for segmentation of retinal layer surfaces in ophthalmic tomography (OPT).

OPT generally creates multi-frame images with frames that are nominally perpendicular to the retinal surface, which is treated as if it were a flat baseline coronal plane for image rendering (see Section A.52.4.3.1 in PS3.3 ).

When OPT scans are acquired in a regular set of closely spaced rasters, they represent a complete volume and are characterized with the Ophthalmic Volumetric Properties Flag (0022,1622) value YES. This use may also typically have Scan Pattern Type Code Sequence (0022,1618) value (128279, DCM, "Cube B-scan pattern"). In this case, the height map segmentation for each surface may be a 2-D frame orthogonal to the OPT scan frames, and is analogous to an Ophthalmic Thickness Map image or a Corneal Topography Map image (which is also a type of height map). There will thus be one 2-D Height Map Segmentation frame for each segmented surface layer.

However, OPT scans may not be volumetric (see CID 4272 “OPT Scan Pattern Type” for non-cube patterns). In that case, the segmented surface layer in each OPT frame will have a corresponding Height Map Segmentation frame consisting of a single row. Each layer, i.e., segment, within a Height Map Segmentation SOP Instance may therefore be specified by a set of 1-D frames.

Height Map segmentations of OPT (or other) images may be used in a number of follow-on applications. The surfaces may be overlaid on renderings of the source images, or they may be used to select data to be further processed, e.g., to create en face images of individual retinal layers.