DR HAGIS: Diabetic Retinopathy, Hypertension, Age-related macular degeneration and Glacuoma ImageS

 

CREATOR: gd-jpeg v1.0 (using IJG JPEG v62), default quality

 

Introduction

 

The DR HAGIS database has been created to aid the development of vessel extraction algorithms suitable for retinal screening programmes. Researchers are encouraged to test their segmentation algorithms using this database.

 

Downloading the DR HAGIS database

 

The database can be downloaded here. The DR HAGIS database is free to download for research and educational purposes only. Unauthorized use, redistribution or copying of the entire, or any part of, the database is prohibited. Any research wishing to publish their results, which use this database, must acknowledge the DR HAGIS database by citing this publication:

 

S. Holm, G. Russell, V. Nourrit, N. McLoughlin, “DR HAGIS – A Novel Fundus Image Database for the Automatic Extraction of Retinal Surface Vessels”, SPIE Journal of Medical Imaging, 2017 (In press).

 

If you have any queries or feedback regarding the DR HAGIS database please contact Niall McLoughlin.

 

Data description

 

All thirty-nine fundus images were obtained from a diabetic retinopathy screening programme in the UK. Hence, all images were taken from diabetic patients. Since patients attending these screening programmes exhibit other co-morbidities, the DR HAGIS database consists of following four co-morbidity subgroups:

 

Š       Images 1-10: Glaucoma subgroup

Š       Images 11-20: Hypertension subgroup

Š       Images 21-30: Diabetic retinopathy subgroup

Š       Images 31-40: Age-related macular degeneration subgroup

 

It should be noted that image 24 and 32 are identical, as this fundus image was obtained from a patient exhibiting both diabetic retinopathy and age-related macular degeneration.

 

Besides the fundus images, the manual segmentation of the retinal surface vessels is provided by an expert grader. These manually segmented images can be used as the ground truth to compare and assess the automatic vessel extraction algorithms. Masks of the FOV are provided as well to quantify the accuracy of vessel extraction within the FOV only.

 

The images were acquired in different screening centers, therefore reflecting the range of image resolutions, digital cameras and fundus cameras used in the clinic. The fundus images were captured using a Topcon TRC-NW6s, Topcon TRC-NW8 or a Canon CR DGi fundus camera with a horizontal 45 degree field-of-view (FOV). The images are 4752x3168 pixels, 3456x2304 pixels, 3126x2136 pixels, 2896x1944 pixels or 2816x1880 pixels in size.

 

The fundus images are saved as compressed JPEG files with 8 bits per colour plane. The ground truth and mask images are saved as binary PNG files.

 

We would like to thank Health Intelligence, Sandbach, UK for making these fundus images available.

 

Data Analysis

 

The mask and ground truth images are provided to calculate the accuracy, sensitivity and specificity within the FOV of the automatic vessel extraction. Since image 24 and 32 are identical, please exclude one of them from analysis when reporting the results of the entire DR HAGIS database. However, both images should be included in the analysis when reporting the quality of vessel extraction for the individual subgroups.

 

 

© 2016- Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK