├── .github └── FUNDING.yml ├── README.md ├── benchmark-eyepacs-airogs-light-v1.md ├── benchmark-eyepacs-airogs-light-v2.md ├── data-availability.md ├── origin.md └── summary.md /.github/FUNDING.yml: -------------------------------------------------------------------------------- 1 | # These are supported funding model platforms 2 | 3 | github: TheBeastCoding 4 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | ## New Updates 2 | * 08/03/24 : A new abstract is available to the public, introducing a novel approach to glaucoma detection: Assessment of Retinal Vasculature for Glaucoma Detection: A Comparative Analysis of Human Expertise and Deep Learning Algorithms. 3 | * 05/15/24 : The EyePACS-light-V2 code folder on Kaggle has a new PyTorch template for quick and easy glaucoma detection setup. This model uses the lightweight MobileNetV3 and achieved a test accuracy of 92.6%: https://www.kaggle.com/code/deathtrooper/pytorch-easy-setup-for-glaucoma-detection-92-6 4 | * 03/09/24 : EyePACS-light-V2 now has a 10.0 Kaggle usability score: supplemental metadata.csv file added to dataset. 5 | * 01/20/24 : EyePACS-light-V2 preprocess high-level overview is now available on the Kaggle dataset link in the about section (scroll all the way down) if you are curious on how the dataset was derived 6 | * 12/28/23 : EyePACS-light-V2 94.94% test accuracy benchmark using ConvNeXtTiny: https://www.kaggle.com/code/deathtrooper/benchmark-94-94-convnexttiny 7 | * 12/12/23 : EyePACS-light-V2 is HERE!!! Be the first to benchmark your model with this improved dataset! Download from kaggle: https://www.kaggle.com/datasets/deathtrooper/glaucoma-dataset-eyepacs-airogs-light-v2/data 8 | 9 | ## Citation Note 10 | If you found this catalog helpful, please consider citing the following: 11 | - Riley Kiefer, Muhammad Abid, Jessica Steen, Mahsa Raeisi Ardali, and Ehsan Amjadian. 2023. A Catalog of Public Glaucoma Datasets for Machine Learning Applications: A detailed description and analysis of public glaucoma datasets available to machine learning engineers tackling glaucoma-related problems using retinal fundus images and OCT images. In Proceedings of the 2023 7th International Conference on Information System and Data Mining (ICISDM '23). Association for Computing Machinery, New York, NY, USA, 24–31. https://doi.org/10.1145/3603765.3603779 12 | 13 | 14 | # Public Glaucoma Dataset Catalog 15 | [Help expand this repository by providing links/publications to new glaucoma datasets!] 16 | 17 | ## Repository Table of Contents 18 | - README.md : Glaucoma overview, relevant research, and dataset access links 19 | - benchmark-eyepacs-airogs-light.md : Leaderboard for the test set evaluation using the train/val sets of the EyePACS-AIROGS-light dataset 20 | - summary.md : Dataset class breakdown, image types, and glaucoma types 21 | - data-availability.md : Dataset image and segmentation availability. 22 | - origin.md : Dataset collection origin and collection years. 23 | 24 | ## Glaucoma Overview 25 | According to AAO, "Glaucoma is a disease that damages your eye’s optic nerve. It usually happens when fluid builds up in the front part of your eye. That extra fluid increases the pressure in your eye, damaging the optic nerve". It is a leading cause of blindness and worsens over time if left untreated. Optometrists diagnose glaucoma through fundus images (a 2D image of the eye) or ocular coherence tomography (OCT) images (a 3D image of the eye). In fundus images, optometrists typically look for optic cup or optic disc damage. In OCT images, optometrists typically look for layer atrophy. To automate the detection of glaucoma, datasets are curated for machine learning. Fundus image datasets are typically designed for either glaucoma classification to distinguish healthy for glaucoma or optic nerve head segmentation to extract and analyze cup or disc damage. High-quality benchmark datasets like EyePACS-AIROGS-light have predetermined train, validation, and test sets for reproducibility. 26 | 27 | ## Relevant Glaucoma Datasets (by me) 28 | - SMDG-19 [Dataset], "Standardized Multi-Channel Dataset for Glaucoma, Version 19", https://www.kaggle.com/datasets/deathtrooper/multichannel-glaucoma-benchmark-dataset 29 | - EyePACS-AIROGS-light-v1 [Dataset], "EyePACS Artificial Intelligence for Robust Glaucoma Screening Challenge, Lightweight Version 1", https://www.kaggle.com/datasets/deathtrooper/eyepacs-airogs-light 30 | - EyePACS-AIROGS-light-v2 [Dataset], "EyePACS Artificial Intelligence for Robust Glaucoma Screening Challenge, Lightweight Version 2", https://www.kaggle.com/datasets/deathtrooper/glaucoma-dataset-eyepacs-airogs-light-v2 31 | 32 | ## Relevant Glaucoma Research (by me) 33 | - ARVO 2024 [Abstract], "Assessment of Retinal Vasculature for Glaucoma Detection: A Comparative Analysis of Human Expertise and Deep Learning Algorithms", https://iovs.arvojournals.org/article.aspx?articleid=2794846 34 | - AAO 2022 [Abstract], "A Comprehensive Survey of Publicly Available Glaucoma Datasets for Automated Glaucoma Detection", https://aaopt.org/past-meeting-abstract-archives/?SortBy=ArticleYear&ArticleType=&ArticleYear=2022&Title=&Abstract=&Authors=&Affiliation=&PROGRAMNUMBER=225129 35 | - AAO 2023 [Abstract], "The Predictive Power of Fundus Blood Vessels in Glaucoma Detection", accepted as a presentation for AAO 2023 36 | - AAO 2023 [Abstract], "EyePACS-light: A Lightweight Balanced Dataset for Automated Glaucoma Classification Modeling", accepted as a poster for AAO 2023 37 | - ARVO 2023 [Abstract], "Ground truth validation of publicly available datasets utilized in artificial intelligence models for glaucoma detection", https://iovs.arvojournals.org/article.aspx?articleid=2791017 38 | - ARVO 2023 [Abstract], "Standardized and Open-Access Glaucoma Dataset for Artificial Intelligence Applications", https://iovs.arvojournals.org/article.aspx?articleid=2790420 39 | - IEEE-IEMCOM 2022 [Full Paper], "A Survey of Glaucoma Detection Algorithms using Fundus and OCT Images", https://ieeexplore.ieee.org/abstract/document/9946629 40 | - IEEE-ICIVC 2023 [Full Paper], "Automated Fundus Image Standardization Using a Dynamic Global Foreground Threshold Algorithm", https://ieeexplore.ieee.org/abstract/document/10270429 41 | - ICISDM 2023 [Full Paper], "A Catalog of Public Glaucoma Datasets for Machine Learning Applications", https://dl.acm.org/doi/abs/10.1145/3603765.3603779 42 | 43 | ## Example data 44 | Drishti-GS | G1020 | ORIGA-light | REFUGE1-VAL | PAPILA 45 | --- | --- | --- | --- | --- 46 | | | | | 47 | 48 | ## Use Case Acronyms 49 | - Classification 50 | - BGC = Binary Glaucoma Classification (healthy vs. glaucoma or non-glaucoma vs. glaucoma) 51 | - MGC = Multi Glaucoma Classification (at least 2 glaucoma types, including suspect) 52 | - Segmentation 53 | - ODS = Optic Disc Segmentation 54 | - OCS = Optic Cup Segmentation 55 | - BVS = Blood Vessel Segmentation 56 | - OLS = OCT Layer Segmentation 57 | - RNFLS = Retinal Nerve Fiber Layer Segmentation 58 | - Other 59 | - LT = Localization Task 60 | - IQA = Image Quality Assesment 61 | - MIDI = Multi Image Domain Input 62 | - CDR = Cup-to-Disc Ratio Estimation 63 | - N = Notching 64 | - VF = Visual Field Information/Segmentation 65 | 66 | ## Public Glaucoma Image Datasets 67 | | Dataset | Access Link | Accessibility | Glaucoma Labels? | Use Case | 68 | | ------------- | ------------- | ------------- | ------------- | ------------- | 69 | | ACRIMA | https://figshare.com/s/c2d31f850af14c5b5232 | open | Y | BGC | 70 | | AGE | https://age.grand-challenge.org/Download/ | registration | Y | MGC, LT | 71 | | BEH (Bangladesh Eye Hospital) | https://github.com/mirtanvirislam/Deep-Learning-Based-Glaucoma-Detection-with-Cropped-Optic-Cup-and-Disc-and-Blood-Vessel-Segmentation/tree/master/Dataset | open | Y | BGC | 72 | | BIOMISA | https://data.mendeley.com/datasets/2rnnz5nz74/2 | open | Y | MGC, BGC, MIDI, OLS, CDR | 73 | | Chaksu-IMAGE | https://doi.org/10.6084/m9.figshare.20123135 | open | Y | BGC | 74 | | CRFO-v4 | https://data.mendeley.com/datasets/trghs22fpg/4 | open | Y | BGC, MDI, ODS, OCS | 75 | | DR-HAGIS | https://personalpages.manchester.ac.uk/staff/niall.p.mcloughlin/ | open | Y | BGC, BVS | 76 | | DRIONS-DB | https://www.researchgate.net/publication/326460478_Glaucoma_dataset_-_DRIONS-DB | open | N | ODS | 77 | | DRISHTI-GS1 | https://cvit.iiit.ac.in/projects/mip/drishti-gs/mip-dataset2/Home.php | open | Y | BGC, ODS, OCS, CDR, N | 78 | | EyePACS-AIROGS | https://airogs.grand-challenge.org/data-and-challenge/ | open | Y | BGC, IQA | 79 | | EyePACS-AIROGS-light (v1) | https://www.kaggle.com/datasets/deathtrooper/eyepacs-airogs-light | registration | Y | BGC | 80 | | EyePACS-AIROGS-light (v2) | https://www.kaggle.com/datasets/deathtrooper/glaucoma-dataset-eyepacs-airogs-light-v2 | registration | Y | BGC | 81 | | FIVES | https://figshare.com/articles/figure/FIVES_A_Fundus_Image_Dataset_for_AI-based_Vessel_Segmentation/19688169/1 | open | Y | BGC, BVS | 82 | | G1020 | https://www.kaggle.com/datasets/arnavjain1/glaucoma-datasets | registration | Y | BGC, ODS, OCS | 83 | | GAMMA | https://gamma.grand-challenge.org/ | registration | Y | BGC?, ODS, OCS, OLS?, LT, MIDI | 84 | | GOALS | https://ichallenges.grand-challenge.org/iChallenge-GON3/ | registration | Y | BGC, OLS | 85 | | GRAPE | https://springernature.figshare.com/collections/GRAPE_A_multi-modal_glaucoma_dataset_of_follow-up_visual_field_and_fundus_images_for_glaucoma_management/6406319/1 | open | Y | MGC, VF | 86 | | Harvard-GF | https://ophai.hms.harvard.edu/datasets/harvard-glaucoma-fairness-3300-samples/ | request | N | BGC, VF, RNFLS, MIDI | 87 | | HRF | https://www5.cs.fau.de/research/data/fundus-images/ | open | Y | 88 | | INSPIRE-AVR-test | https://medicine.uiowa.edu/eye/inspire-datasets | open | N | 89 | | INSPIRE-STEREO | https://medicine.uiowa.edu/eye/inspire-datasets | open | N | 90 | | JSIEC-1000 | https://www.kaggle.com/datasets/linchundan/fundusimage1000 | registration | Y | 91 | | KEH (Kim's Eye Hospital) | https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/1YRRAC | open | Y | 92 | | LAG | https://github.com/smilell/AG-CNN | request | Y | 93 | | LES-AV | https://figshare.com/articles/dataset/LES-AV_dataset/11857698/1 | open | Y | 94 | | Leuven-Haifa HRF | https://rdr.kuleuven.be/dataset.xhtml?persistentId=doi:10.48804/Z7SHGO | request | Y | MGC, BVS | 95 | | MSHF | https://www.nature.com/articles/s41597-023-02188-x#ref-CR17 | open | Y | BGC, IQA | 96 | | OCTV | https://zenodo.org/record/1481223#.Y20g3XbMIuV | open | Y | 97 | | OIA-ODIR | https://www.kaggle.com/datasets/andrewmvd/ocular-disease-recognition-odir5k | registration | Y | 98 | | ONHSD | https://aldiri.info/Image%20Datasets/ONHSD.aspx | inaccessible | Y | 99 | | ORIGA-light | https://www.kaggle.com/datasets/sshikamaru/glaucoma-detection | registration | Y | 100 | | PAPILA | https://doi.org/10.6084/m9.figshare.14798004.v1 | open | Y | 101 | | REFUGE1 | https://refuge.grand-challenge.org/REFUGE2Download/ | registration | Y | 102 | | REFUGE2 | https://refuge.grand-challenge.org/REFUGE2Download/ | registration | Y | 103 | | RIGA-BIN-RUSHED | https://deepblue.lib.umich.edu/data/concern/data_sets/3b591905z | open | N | 104 | | RIGA-MAGRABI | https://deepblue.lib.umich.edu/data/concern/data_sets/3b591905z | open | N | 105 | | RIGA-MESSIDOR | https://deepblue.lib.umich.edu/data/concern/data_sets/3b591905z | open | N | 106 | | RIM-ONE-r1 | http://medimrg.webs.ull.es/research/retinal-imaging/rim-one/ | open | Y | 107 | | RIM-ONE-r2 | http://medimrg.webs.ull.es/research/retinal-imaging/rim-one/ | open | Y | 108 | | RIM-ONE-r3 | http://medimrg.webs.ull.es/research/retinal-imaging/rim-one/ | open | Y | 109 | | RIM-ONE-DL | http://medimrg.webs.ull.es/research/retinal-imaging/rim-one/ | open | Y | 110 | | SIGF | https://github.com/XiaofeiWang2018/DeepGF | request | Y | 111 | | SMDG | https://www.kaggle.com/datasets/deathtrooper/multichannel-glaucoma-benchmark-dataset | registration | Y | 112 | | sjchoi86-HRF | https://github.com/yiweichen04/retina_dataset | open | Y | 113 | | VEIRC | https://github.com/ProfMKD/Glaucoma-dataset | open | Y | 114 | -------------------------------------------------------------------------------- /benchmark-eyepacs-airogs-light-v1.md: -------------------------------------------------------------------------------- 1 | ## EyePACS-AIROGS-light (v1) Benchmark Leaderboard 2 | Dataset Found Here: https://www.kaggle.com/datasets/deathtrooper/eyepacs-airogs-light 3 | 4 | | Rank | Method | Author | Date | Test Accuracy | Link | 5 | | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | 6 | | 1 | Xception-Preprocessed | KEREM KARABACAK | Aug, 2023 | 93.5% | https://www.kaggle.com/code/keremkarabacak/xception-93-5 | 7 | | 2 | MobileNetV3-Regularized | Self | Sept, 2023 | 92.1% | https://www.kaggle.com/code/deathtrooper/92-mobilenetv3-glaucoma-detection | 8 | | 3 | MobileNetV3-Preprocessed | Self | Aug, 2023 | 91.7% | https://www.kaggle.com/code/deathtrooper/91-7-glaucoma-detection-with-preprocess-noise | 9 | | 4 | MobileNetV3-Baseline | Self | May, 2023 | 88.9% | https://www.kaggle.com/code/deathtrooper/glaucoma-classification-easy-setup-88-9-baseline | 10 | -------------------------------------------------------------------------------- /benchmark-eyepacs-airogs-light-v2.md: -------------------------------------------------------------------------------- 1 | ## EyePACS-AIROGS-light (v2) Benchmark Leaderboard 2 | Dataset Found Here: https://www.kaggle.com/datasets/deathtrooper/glaucoma-dataset-eyepacs-airogs-light-v2/data 3 | 4 | | Rank | Method | Author | Date | Test Accuracy | Link | 5 | | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | 6 | | 1 | ConvNeXtTiny-(Noise, Regularization, Preprocessed) | Riley Kiefer | Dec, 2023 | 94.94% | https://www.kaggle.com/code/deathtrooper/benchmark-94-94-convnexttiny | 7 | | 2 | Xception-(Noise, Regularization, Preprocessed) | Riley Kiefer | Dec, 2023 | 94.16% | https://www.kaggle.com/code/deathtrooper/benchmark-94-16-xception-noise-and-preprocess | 8 | | 3 | MobileNetV3-Baseline | Riley Kiefer | Dec, 2023 | 87.66% | https://www.kaggle.com/code/deathtrooper/glaucoma-classification-template-with-baseline | 9 | 10 | -------------------------------------------------------------------------------- /data-availability.md: -------------------------------------------------------------------------------- 1 | ## Overview 2 | This table presents the dataset image and segmentation type availability. 3 | 4 | ## Key 5 | - F1 : Cropped Fundus 6 | - F2 : Full Fundus 7 | - F-OD : Fundus Optic Disc Segmentation 8 | - F-OC : Fundus Optic Cup Segmentation 9 | - F-BV : Fundus Blood Vessel Segmentation 10 | - F-V : Fundus Vein Segmentation 11 | - F-A : Fundus Artery Segmentation 12 | - OCT1 : Ocular Coherence Tomography Variant 13 | - OCT-OD : OCT Optic Disc Segmentation 14 | - OCT-OC : OCT Optic Cup Segmentation 15 | - OCT-L : OCT Layer Segmentation 16 | - M1 : Miscellaneous Segmentation 17 | 18 | ## Data Table 19 | | Dataset | F1 | F2 | F-OD | F-OC | F-BV | F-V | F-A | OCT1 | OCT-OD | OCT-OC | OCT-L | M1 | 20 | | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | 21 | | AGE | | | | | | | | X | | | | | 22 | | ACRIMA | X | | | | | | | | | | | | 23 | | BEH | | X | | | | | | | | | | | 24 | | BIOMISA | | X | | | | | | X | | X | X | | 25 | | Chaksu-IMAGE | | X | X | X | | | | | | | | | 26 | | CRFO-v4 | | X | X | X | | | | X | X | | | | 27 | | DRISHTI-GS1 | | X | X | X | | | | | | | | | 28 | | DR-HAGIS | | X | | | X | | | | | | | | 29 | | EyePACS-AIROGS| | X | | | | | | | | | | | 30 | | FIVES | | X | | | X | | | | | | | | 31 | | G1020 | | X | X | X | | | | | | | | | 32 | | GRAPE | X | X | X | X | | | | | | | | X | 33 | | HARVARD-GF | | | | | | | | X | | | | X | 34 | | HRF | | X | | | X | | | | | | | | 35 | | INSPIRE-AVR | | X | | | | | | | | | | | 36 | | INSPIRE-S | X | | X | | | | | | | | | | 37 | | JSIEC-1000 | | X | | | | | | | | | | | 38 | | KEH | X | | | | | | | | | | | | 39 | | LAG | | X | | | | | | | | | | X | 40 | | LES-AV | | X | | | X | X | X | | | | | | 41 | | Leuven-Haifa | | X? | | | | X | X | | | | | | 42 | | MSHF | | X | | | | | | | | | | | 43 | | OCTV | | | | | | | | X | | | | | 44 | | OIA-ODIR | | X | | | | | | | | | | | 45 | | ORIGA-light | | X | X | X | | | | | | | | | 46 | | PAPILA | | X | X | X | | | | | | | | | 47 | | REFUGE | | X | X | X | | | | | | | | | 48 | | RIM-ONE-r1 | X | | X | | | | | | | | | | 49 | | RIM-ONE-r2 | X | | X | | | | | | | | | | 50 | | RIM-ONE-r3 | X | | X | X | | | | | | | | | 51 | | RIM-ONE-DL | X | | | | | | | | | | | | 52 | | SIGF | X | | | | | | | | | | | | 53 | | sjchoi86-HRF | | X | | | | | | | | | | | 54 | | VEIRC | | X | | | | | | | | | | | 55 | -------------------------------------------------------------------------------- /origin.md: -------------------------------------------------------------------------------- 1 | ## Overview 2 | This table presents the dataset collection origin (city/state/region/country) and the years of collection. 3 | 4 | ## Key 5 | - NR = Not Reported 6 | 7 | ## Data Table 8 | | Dataset | Collection City/State/Region | Collection Country | Collection Years | 9 | | ------------- | ------------- | ------------- | ------------- | 10 | | AGE | Guangzhou | China | 2016 - 2019 | 11 | | ACRIMA | Valencia | Spain | NR | 12 | | BEH | Dhaka | Bangladesh | 2019 - 2020 | 13 | | BIOMISA | Manipal, Karnataka | India | NR | 14 | | Chaksu-IMAGE | NR | NR | 2-year period, unknown | 15 | | CRFO-v4 | NR | NR | NR | 16 | | DRISHTI-GS1 | Madurai | India | NR | 17 | | DR-HAGIS | Sandbach | United Kingdom | NR | 18 | | EyePACS-AIROGS | mix | United States | NR | 19 | | FIVES | Hangzhou, Zhejiang | China | 2016 - 2021 | 20 | | G1020 | Kaiserslautern | Germany | 2005 - 2017 | 21 | | GRAPE | Hangzhou | China | 2015 - 2022 | 22 | | HARVARD-GF | Cambridge, Massachusetts | United States | NR | 23 | | HRF | NR | Germany, Czech Republic | NR | 24 | | INSPIRE-AVR | Iowa | United States | NR | 25 | | INSPIRE-S | Iowa | United States | NR | 26 | | JSIEC-1000 | Guangdong | China | 2009 - 2018 | 27 | | KEH | Seoul | Korea | NR | 28 | | LAG | Bejing | China | NR | 29 | | LES-AV | NR | NR | NR | 30 | | Leuven-Haifa HRF | Leuven | Belgium | 2010 - 2019 | 31 | | MSHF | Hangzhou | China | NR | 32 | | OCTV | NR | United States | 2008 - 2016 | 33 | | OIA-ODIR | mix | China | NR | 34 | | ORIGA-light | southwestern | Singapore | Full Fundus | 2004 - 2007 | 35 | | PAPILA | Murcia | Spain | 2018 - 2020 | 36 | | REFUGE | Guangzhou | China | NR | 37 | | RIM-ONE-r1 | Tenerife, Zaragoza, and Madrid | Spain | NR | 38 | | RIM-ONE-r2 | Tenerife, Zaragoza | Spain | NR | 39 | | RIM-ONE-r3 | Tenerife | Spain | NR | 40 | | RIM-ONE-DL | Tenerife, Zaragoza, and Madrid | Spain | NR | 41 | | SIGF | NR | NR | 1986 - 2018 | 42 | | sjchoi86-HRF | NR | NR | NR | 43 | | SMDG (v19) | mix | mix | mix | 44 | | VEIRC | New Delhi | India | NR | 45 | -------------------------------------------------------------------------------- /summary.md: -------------------------------------------------------------------------------- 1 | ## Overview 2 | This table presents the class breakdown (healthy and glaucoma), the image type, and the glaucoma types. 3 | 4 | ## Key 5 | - UG = Unspecified Glaucoma 6 | - OAG = Open Angle Glaucoma 7 | - CAG = Closed Angle Glaucoma 8 | - CSG = Chronic Simple Glaucoma 9 | - RG = Referable Glaucoma 10 | - NTG = Normal Tension Glaucoma 11 | - HTG = High Tension Glaucoma 12 | - EG = Early Glaucoma 13 | - MG = Moderate Glaucoma 14 | - DG = Deep/Severe/Advanced Glaucoma 15 | - SG = Suspect Glaucoma 16 | - OCT = Ocular Coherence Tomography 17 | - AS-OCT = Anterior Segment OCT 18 | - SD-OCT = Spectral Domain OCT 19 | - *Contains other non-glaucoma diseases. 20 | 21 | ## Data Table 22 | | Dataset | Total Healthy | Total Glaucoma | Image Type | Glaucoma Types | 23 | | ------------- | ------------- | ------------- | ------------- | ------------- | 24 | | AGE | 0 | 4800 | AS-OCT | CAG, OAG | 25 | | ACRIMA | 309 | 396 | Cropped Fundus | UG | 26 | | BEH | 463 | 171 | Full Fundus | UG | 27 | | BIOMISA (fundus) | 18 | 32 | Full Fundus | SG, UG | 28 | | BIOMISA (OCT) | 18 | 32 | SD-OCT | SG, UG | 29 | | Chaksu-IMAGE | 1157 | 188 | Full Fundus | SG, UG | 30 | | CRFO-v4 (fundus) | 30 | 48 | Full Fundus | UG | 31 | | CRFO-v4 (OCT) | 13 | 28 | SD-OCT | UG | 32 | | DRISHTI-GS1 (test) | 13 | 38 | Full Fundus | UG | 33 | | DRISHTI-GS1 (train) | 18 | 32 | Full Fundus | UG | 34 | | DR-HAGIS | 0 | 10 | Full Fundus | UG | 35 | | EyePACS-AIROGS (train) | 98172 | 3270 | Full Fundus | RG | 36 | | EyePACS-AIROGS-light (train) | 2500 | 2500 | Full Fundus | RG | 37 | | EyePACS-AIROGS-light (val) | 270 | 270 | Full Fundus | RG | 38 | | EyePACS-AIROGS-light (test) | 500 | 500 | Full Fundus | RG | 39 | | FIVES | 200 | 200 | Full Fundus | CAG, OAG | 40 | | G1020 | 724 | 296 | Full Fundus | UG | 41 | | GRAPE | 0 | 631 | Cropped and Full Fundus | OAG, CAG | 42 | | HARVARD-GF-TRAIN | unknown (2100 total) | unknown (2100 total) | 3D OCT B-scans | UG | 43 | | HARVARD-GF-VAL | unknown (300 total) | unknown (300 total) | 3D OCT B-scans | UG | 44 | | HARVARD-GF-TEST | unknown (900 total) | unknown (900 total) | 3D OCT B-scans | UG | 45 | | HRF | 15 | 15 | Full Fundus | UG | 46 | | INSPIRE-AVR (test) | 0 | 40 | Full Fundus | OAG | 47 | | INSPIRE-S (fundus) | 0 | 30 | Stereo Fundus | UG | 48 | | INSPIRE-S (OCT) | 0 | 30 | Depth OCT | UG | 49 | | JSIEC-1000 | 38 | 13 | Full Fundus | SG | 50 | | KEH | 788 | 756 | Cropped Fundus | DG, EG | 51 | | LAG | 3882 | 4875 | Full Fundus | UG | 52 | | LES-AV | 11 | 11 | Full Fundus | NTG, OAG | 53 | | Leuven-Haifa HRF | 56? | 138? | Full Fundus | NTG, HTG | 54 | | MSHF | 26 | 52 | Full Fundus | UG | 55 | | OCTV | 263 | 847 | OCT Volume | OAG | 56 | | OIA-ODIR (test-offline)| 435 | 36 | Full Fundus | UG | 57 | | OIA-ODIR (test-online)| 843 | 58 | Full Fundus | UG | 58 | | OIA-ODIR (train)| 3038 | 200 | Full Fundus | UG | 59 | | ORIGA-light | 482 | 168 | Full Fundus | UG | 60 | | PAPILA | 333 | 155 | Full Fundus | CSG, SG | 61 | | REFUGE1 | 1080* | 120 | Full Fundus | NTG, OAG | 62 | | REFUGE2 | 1720* | 280 | Full Fundus | NTG, OAG | 63 | | RIM-ONE-r1 | 118 | 40 | Cropped Fundus | DG, EG, MG | 64 | | RIM-ONE-r2 | 255 | 200 | Cropped Fundus | UG | 65 | | RIM-ONE-DL (hospital-train) | 195 | 116 | Cropped Fundus | DG, EG, MG, SG, UG | 66 | | RIM-ONE-DL (hospital-test) | 118 | 56 | Cropped Fundus | DG, EG, MG, SG, UG | 67 | | SIGF | 3303 | 368 | Cropped Fundus | UG | 68 | | sjchoi86-HRF | 300 | 101 | Full Fundus | UG | 69 | | SMDG (v19) | 7499 | 4946 | Full Fundus | many | 70 | | VEIRC | 225 | 139 | Full Fundus | UG | 71 | --------------------------------------------------------------------------------