Skip to main content

Datasets

Standard Dataset

1450 fundus images with 899 glaucoma data and 551 normal data.

Tag image
Citation Author(s):
wheyming song
Submitted by:
Wheyming Tina Song
Last updated:
DOI:
10.21227/4bcp-2z21
Data Format:
*.png
Research Article Link:
A Statistical Robust Glaucoma Detection Framework Combining Retinex CNN and DO…
6484 views
Categories:
Keywords:
fundus images
glaucoma
normal data
Average: 5 (1 vote)
ACCESS DATASET CITE SHARE/EMBED

Abstract

This set contains 1450 fundus images with 899 glaucoma data and 551 normal data.

All text about patient information and the date that the associated images were collected are replaced by 0, which is black.

Instructions:

This set contains 1450 fundus images with 899 glaucoma data and 551 normal data.

All text about patient information and the date that the associated images were collected are replaced by 0, which is black.

I was very much excited in beginning when I came with this dataset, but somehow this disappoints me that many images are directly redundant (taking repeatedly from the same eye of the same patient) i.e., G-740.png, G-741 is a redundant pair of images, and so if G-742 and G-743.

Also, some redundancy comes from the color transformation of some images, i.e., G-2 is just a Greenish version of G-1.png. Such color transformations are not commonly observed in Fundus images.

Also, someone working in the domain could have obtained augmented images with transformations like shifts and rotations. So I would highly suggest authors remove all such redundancies from the dataset and report actual original images from different eyes of different patients in the heading of the dataset. Also, it would be of great help if you can at least quote original source of this dataset, so that people working can be sure ahead in time, that the original images are not from some publicly available dataset

 

   
ACHINT CHAUDHARY Tue, 06/09/2020 - 14:35 Permalink
I believe the green images are red-free images taken by placing a filter in front of the camera. This type of filter is used to highlight RNFL defects, which is a finding in eyes with glaucoma. Thus they are unique images and not generated for the purpose of data augmentation.
Renaud Duval Tue, 09/01/2020 - 15:07 Permalink
Hi, thanks for the dataset. Is there an official paper that goes with this dataset for citation purposes?
Michael Mayo Tue, 03/23/2021 - 21:44 Permalink

I am not able to download the dataset. I am a student and researcher.please provide these dataset.

Tanmoy Mondal Wed, 01/08/2025 - 05:18 Permalink

Dataset Files

LOGIN TO ACCESS DATASET FILES

QUESTIONS?