Table 2.
Description of databases.
| Dataset | Number of Image | Description | Camera | Resolution (Pixel) | Dataset Partition |
|---|---|---|---|---|---|
| DRIVE | 40 | Dataset was collected from 400 diabetic patients aged between 25 and 90 years. 40 photographs were randomly selected, 33 did not show any sign of DR, and 7 showed signs of mild early DR. Training set: Single manual segmentation Testing set: Two manual segmentation |
Canon CR5 nonmydriatic 3CCD camera with a 45° field of view (FOV) | 565 × 584 | Yes Training: 20 Testing: 20 |
| STARE | 20 | Images were collected from DR, PDR, ASR, HTR, etc. patients. Each image has pixel-level vessel annotation provided by two experts. Performance is computed with the segmentation of the first observer as ground truth. | TopCon TRV-50 fundus camera with a 35° FOV | 700 × 605 | No |
| CHASE_DB1 | 28 | Subset of retinal images of multiethnic children from the Child Heart and Healthy Study in England. (https://blogs.kingston.ac.uk/retinal/chasedb1/) | Nidek NM-200-D fundus camera with a 30° FOV | 1280 × 960 | Yes Training: 20 Testing: 8 |
| HRF | 45 | Data were collected from 15 healthy patients, 15 glaucomatous patients, and 15 diabetic retinopathy patients separately. It contains a binary gold standard vessel segmentation images that are determined by a group of experts (experience in retinal images analysis). | Canon CR-1 fundus camera with a field of view of 45° and different acquisition setting | 500 × 2500 | No |
| TONGRE | 30 | Images collected from 30 people at the Tongren Beijing Hospital, where five of these images show a pathological pattern (glaucoma). | NR | 1880 × 281 | Yes Training: 15 Test: 15 |
| DRIONS | 110 | Dataset contains high resolution images of blood vessels, 25 images were from patients with chronic glaucoma while the remaining 85 images were from hypertensive retinopathy patients. | Analogical fundus camera approximately centered on the ONH | 600 × 400 | Yes Training: 60 Test: 50 |
| REVIEW | 16 | The dataset includes retinal images with 193 vessel segments, demonstrating a variety of pathologies, and vessel types (8 high-resolution, 4 vascular diseases, 2 central light reflex, 2 kickpoint). It also contains 5066 manually marked profiles. It has been marked by three observers. | NR | 1360 × 1024 to 3584 × 2438 | No |
DRIVE = The Digital Retinal Images for Vessel Extraction database; STARE = The Structured Analysis of the Retina database; CHASE_DB1 = The Child Heart and Health Study in England database; HRF = High-Resolution Fundus; REVIEW = Retinal Vessel Image set for Estimation of Widths, DR = Diabetic retinopathy, PDR = Proliferative diabetic retinopathy, ASR = Arteriosclerotic Retinopathy, HTR = Hypertensive Retinopathy.