2D Point Segmentation

2D Point Segmentation annotations use the standard WKT POINT geometry type. Each annotation records an (x, y) coordinate and the DICOM frame on which the point was placed. Coordinates are scaled relative to the image dimensions, consistent with other Centaur Labs segmentation types.

See WKT Format for details on the coordinate system.

Result format

Each annotated point is exported as a WKT POINT string, for example:

POINT (0.4521 0.3187)

The coordinates represent the position on the image relative to its width and height (values between 0 and 1). The associated DICOM frame number is included in the results metadata.

Aggregation

Because each point marks a single precise location, aggregation for 2D Point Segmentation produces one consensus (x, y) coordinate per label class per case, derived from all qualified reads.

Availability

2D Point Segmentation is currently supported in DICOM Viewer tasks only.