REVIEW: Retinal Vessel Image set for Estimation of Widths

DRIVE: Digital Retinal Images for Vessel Extraction

The DRIVE (Digital Retinal Images for Vessel Extraction) database was obtained from a diabetic retinopathy screening programme in the Netherlands. The screening population subjects range between 31 to 86 years of age.

The database consists of 40 images which are supplied as a training and test set, each containing 20 images; 7 contain pathology, namely exudates, hemorrhages and pigment epithelium changes. The training set has 3 images with pathology while the test set has 4 images with pathology.

They have been manually segmented by three observers (members of the DRIVE research group) who were computer science researchers trained by an ophthalmologist. The images in the training set were segmented once, while images in the test set were manually segmented twice, resulting in sets A and B. The observers of sets A and B produced similar segmentations.

Related Links

Some of the related websites:

STARE: STructured Analysis of the Retina

The STARE database contains 20 images for blood vessel segmentation; ten of these contain pathology. The digitized slides are captured by a TopCon TRV-50 fundus camera at 35^o  field of view. The slides were digitized to 605 x 700 pixels, 8 bits per colour channel. Two observers manually segmented all images.

The human observers are able to classify most of the pixels unambiguously. However, the human observers are uncertain about some pixels, such as those for vessels near pathology, those for small vessels and those on the vessel boundaries. The first observer took a more conservative view of the boundaries of vessels, and in the identification of small vessels than the second observer.