ROSE Retinal OCTA Blood Vessel Segmentation Dataset

ROSE stands for Retinal OCTA SEgmentation, where OCTA stands for Optical Coherence Tomography Angiography. This dataset consists of 229 OCTA images with centerline and pixel-level blood vessel annotations. It has been released for public viewing to assist researchers in conducting research on eye disease-related topics.

A subset of this dataset:

• ROSE-1: The dataset was collected by the AngioVue software on the RTVue XR Avanti SD-OCT system of Optovue, USA. It contains 117 OCTA images of 39 subjects (26 with the disease and the rest as healthy controls), of which 90 were used for training and 27 for testing.
• ROSE-2: Data were acquired by the Heidelberg OCT2 system using Spectralis software, 112 OCTA images of 112 eyes, 90 for training and 22 for testing.