OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers

Mustafa Arikan; James Willoughby; Sevim Ongun; Ferenc Sallo; Andrea Montesel; Hend Ahmed; Ahmed Hagag; Marius Book; Henrik Faatz; Maria Vittoria Cicinelli; Amani A. Fawzi; Dominika Podkowinski; Marketa Cilkova; Diana Morais De Almeida; Moussa Zouache; Ganesham Ramsamy; Watjana Lilaonitkul; Adam M. Dubis

doi:10.1038/s41597-024-04259-z

OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers

Mustafa Arikan, James Willoughby, Sevim Ongun, Ferenc Sallo, Andrea Montesel, Hend Ahmed, Ahmed Hagag, Marius Book, Henrik Faatz, Maria Vittoria Cicinelli, Amani A. Fawzi, Dominika Podkowinski, Marketa Cilkova, Diana Morais De Almeida, Moussa Zouache, Ganesham Ramsamy, Watjana Lilaonitkul, Adam M. Dubis^*

^*Corresponding author for this work

Ophthalmology

Research output: Contribution to journal › Article › peer-review

3 Scopus citations

Abstract

Publicly available open-access OCT datasets for retinal layer segmentation have been limited in scope, often being small in size, specific to a single disease, or containing only one grading. This dataset improves upon this with multi-grader and multi-disease labels for training machine learning-based algorithms. The proposed dataset covers three subsets of scans (Age-related Macular Degeneration, Diabetic Macular Edema, and healthy) and annotations for two types of tasks (semantic segmentation and object detection). This dataset compiled 5016 pixel-wise manual labels for 1672 OCT scans featuring 5 layer boundaries for three different disease classes to support development of automatic techniques. A subset of data (566 scans across 9 classes of disease biomarkers) was subsequently labeled for disease features for 4698 bounding box annotations. To minimize bias, images were shuffled and distributed among graders. Retinal layers were corrected, and outliers identified using the interquartile range (IQR). This step was iterated three times, improving layer annotations’ quality iteratively, ensuring a reliable dataset for automated retinal image analysis.

Original language	English (US)
Article number	267
Journal	Scientific Data
Volume	12
Issue number	1
DOIs	https://doi.org/10.1038/s41597-024-04259-z
State	Published - Dec 2025

Funding

The authors acknowledge the use of the Joint Academic Data science Endeavour (JADE) Tier 2 computing facility funded by the Engineering and Physical Sciences Research Council (EPSRC), UK. This research was supported by the National Institute for Health and Care Research (NIHR) Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust, UCL Institute of Ophthalmology and UCL Institute of Healthcare Engineering. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. This work was supported by National Institutes of Health Care Grant (EY014800), and an Unrestricted Grant from Research to Prevent Blindness, New York, NY, to the Department of Ophthalmology & Visual Sciences, University of Utah.

ASJC Scopus subject areas

Statistics and Probability
Information Systems
Education
Computer Science Applications
Statistics, Probability and Uncertainty
Library and Information Sciences

Access to Document

10.1038/s41597-024-04259-z

Cite this

Arikan, M., Willoughby, J., Ongun, S., Sallo, F., Montesel, A., Ahmed, H., Hagag, A., Book, M., Faatz, H., Cicinelli, M. V., Fawzi, A. A., Podkowinski, D., Cilkova, M., De Almeida, D. M., Zouache, M., Ramsamy, G., Lilaonitkul, W., & Dubis, A. M. (2025). OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers. Scientific Data, 12(1), Article 267. https://doi.org/10.1038/s41597-024-04259-z

@article{6155bf0f043448749c218cc2b7c1c250,

title = "OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers",

abstract = "Publicly available open-access OCT datasets for retinal layer segmentation have been limited in scope, often being small in size, specific to a single disease, or containing only one grading. This dataset improves upon this with multi-grader and multi-disease labels for training machine learning-based algorithms. The proposed dataset covers three subsets of scans (Age-related Macular Degeneration, Diabetic Macular Edema, and healthy) and annotations for two types of tasks (semantic segmentation and object detection). This dataset compiled 5016 pixel-wise manual labels for 1672 OCT scans featuring 5 layer boundaries for three different disease classes to support development of automatic techniques. A subset of data (566 scans across 9 classes of disease biomarkers) was subsequently labeled for disease features for 4698 bounding box annotations. To minimize bias, images were shuffled and distributed among graders. Retinal layers were corrected, and outliers identified using the interquartile range (IQR). This step was iterated three times, improving layer annotations{\textquoteright} quality iteratively, ensuring a reliable dataset for automated retinal image analysis.",

author = "Mustafa Arikan and James Willoughby and Sevim Ongun and Ferenc Sallo and Andrea Montesel and Hend Ahmed and Ahmed Hagag and Marius Book and Henrik Faatz and Cicinelli, \{Maria Vittoria\} and Fawzi, \{Amani A.\} and Dominika Podkowinski and Marketa Cilkova and \{De Almeida\}, \{Diana Morais\} and Moussa Zouache and Ganesham Ramsamy and Watjana Lilaonitkul and Dubis, \{Adam M.\}",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2025.",

year = "2025",

month = dec,

doi = "10.1038/s41597-024-04259-z",

language = "English (US)",

volume = "12",

journal = "Scientific Data",

issn = "2052-4463",

publisher = "Nature Research",

number = "1",

}

Arikan, M, Willoughby, J, Ongun, S, Sallo, F, Montesel, A, Ahmed, H, Hagag, A, Book, M, Faatz, H, Cicinelli, MV, Fawzi, AA, Podkowinski, D, Cilkova, M, De Almeida, DM, Zouache, M, Ramsamy, G, Lilaonitkul, W & Dubis, AM 2025, 'OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers', Scientific Data, vol. 12, no. 1, 267. https://doi.org/10.1038/s41597-024-04259-z

TY - JOUR

T1 - OCT5k

T2 - A dataset of multi-disease and multi-graded annotations for retinal layers

AU - Arikan, Mustafa

AU - Willoughby, James

AU - Ongun, Sevim

AU - Sallo, Ferenc

AU - Montesel, Andrea

AU - Ahmed, Hend

AU - Hagag, Ahmed

AU - Book, Marius

AU - Faatz, Henrik

AU - Cicinelli, Maria Vittoria

AU - Fawzi, Amani A.

AU - Podkowinski, Dominika

AU - Cilkova, Marketa

AU - De Almeida, Diana Morais

AU - Zouache, Moussa

AU - Ramsamy, Ganesham

AU - Lilaonitkul, Watjana

AU - Dubis, Adam M.

PY - 2025/12

Y1 - 2025/12

N2 - Publicly available open-access OCT datasets for retinal layer segmentation have been limited in scope, often being small in size, specific to a single disease, or containing only one grading. This dataset improves upon this with multi-grader and multi-disease labels for training machine learning-based algorithms. The proposed dataset covers three subsets of scans (Age-related Macular Degeneration, Diabetic Macular Edema, and healthy) and annotations for two types of tasks (semantic segmentation and object detection). This dataset compiled 5016 pixel-wise manual labels for 1672 OCT scans featuring 5 layer boundaries for three different disease classes to support development of automatic techniques. A subset of data (566 scans across 9 classes of disease biomarkers) was subsequently labeled for disease features for 4698 bounding box annotations. To minimize bias, images were shuffled and distributed among graders. Retinal layers were corrected, and outliers identified using the interquartile range (IQR). This step was iterated three times, improving layer annotations’ quality iteratively, ensuring a reliable dataset for automated retinal image analysis.

AB - Publicly available open-access OCT datasets for retinal layer segmentation have been limited in scope, often being small in size, specific to a single disease, or containing only one grading. This dataset improves upon this with multi-grader and multi-disease labels for training machine learning-based algorithms. The proposed dataset covers three subsets of scans (Age-related Macular Degeneration, Diabetic Macular Edema, and healthy) and annotations for two types of tasks (semantic segmentation and object detection). This dataset compiled 5016 pixel-wise manual labels for 1672 OCT scans featuring 5 layer boundaries for three different disease classes to support development of automatic techniques. A subset of data (566 scans across 9 classes of disease biomarkers) was subsequently labeled for disease features for 4698 bounding box annotations. To minimize bias, images were shuffled and distributed among graders. Retinal layers were corrected, and outliers identified using the interquartile range (IQR). This step was iterated three times, improving layer annotations’ quality iteratively, ensuring a reliable dataset for automated retinal image analysis.

UR - https://www.scopus.com/pages/publications/85218817114

UR - https://www.scopus.com/inward/citedby.url?scp=85218817114&partnerID=8YFLogxK

U2 - 10.1038/s41597-024-04259-z

DO - 10.1038/s41597-024-04259-z

M3 - Article

C2 - 39952954

AN - SCOPUS:85218817114

SN - 2052-4463

VL - 12

JO - Scientific Data

JF - Scientific Data

IS - 1

M1 - 267

ER -

OCT5k: A dataset of multi-disease and multi-graded annotations for retinal layers

Abstract

Funding

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this