A foundation model for generalizable disease detection from retinal images
© 2023. The Author(s)..
Medical artificial intelligence (AI) offers great potential for recognizing signs of health conditions in retinal images and expediting the diagnosis of eye diseases and systemic disorders1. However, the development of AI models requires substantial annotation and models are usually task-specific with limited generalizability to different clinical applications2. Here, we present RETFound, a foundation model for retinal images that learns generalizable representations from unlabelled retinal images and provides a basis for label-efficient model adaptation in several applications. Specifically, RETFound is trained on 1.6 million unlabelled retinal images by means of self-supervised learning and then adapted to disease detection tasks with explicit labels. We show that adapted RETFound consistently outperforms several comparison models in the diagnosis and prognosis of sight-threatening eye diseases, as well as incident prediction of complex systemic disorders such as heart failure and myocardial infarction with fewer labelled data. RETFound provides a generalizable solution to improve model performance and alleviate the annotation workload of experts to enable broad clinical AI applications from retinal imaging.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:622 |
|---|---|
| Enthalten in: |
Nature - 622(2023), 7981 vom: 13. Okt., Seite 156-163 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Zhou, Yukun [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
|---|
| Anmerkungen: |
Date Completed 06.10.2023 Date Revised 20.03.2024 published: Print-Electronic Citation Status MEDLINE |
|---|
| doi: |
10.1038/s41586-023-06555-x |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM362020205 |
|---|
| LEADER | 01000caa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM362020205 | ||
| 003 | DE-627 | ||
| 005 | 20240320233432.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231226s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1038/s41586-023-06555-x |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1337.xml |
| 035 | |a (DE-627)NLM362020205 | ||
| 035 | |a (NLM)37704728 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Zhou, Yukun |e verfasserin |4 aut | |
| 245 | 1 | 2 | |a A foundation model for generalizable disease detection from retinal images |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Completed 06.10.2023 | ||
| 500 | |a Date Revised 20.03.2024 | ||
| 500 | |a published: Print-Electronic | ||
| 500 | |a Citation Status MEDLINE | ||
| 520 | |a © 2023. The Author(s). | ||
| 520 | |a Medical artificial intelligence (AI) offers great potential for recognizing signs of health conditions in retinal images and expediting the diagnosis of eye diseases and systemic disorders1. However, the development of AI models requires substantial annotation and models are usually task-specific with limited generalizability to different clinical applications2. Here, we present RETFound, a foundation model for retinal images that learns generalizable representations from unlabelled retinal images and provides a basis for label-efficient model adaptation in several applications. Specifically, RETFound is trained on 1.6 million unlabelled retinal images by means of self-supervised learning and then adapted to disease detection tasks with explicit labels. We show that adapted RETFound consistently outperforms several comparison models in the diagnosis and prognosis of sight-threatening eye diseases, as well as incident prediction of complex systemic disorders such as heart failure and myocardial infarction with fewer labelled data. RETFound provides a generalizable solution to improve model performance and alleviate the annotation workload of experts to enable broad clinical AI applications from retinal imaging | ||
| 650 | 4 | |a Journal Article | |
| 700 | 1 | |a Chia, Mark A |e verfasserin |4 aut | |
| 700 | 1 | |a Wagner, Siegfried K |e verfasserin |4 aut | |
| 700 | 1 | |a Ayhan, Murat S |e verfasserin |4 aut | |
| 700 | 1 | |a Williamson, Dominic J |e verfasserin |4 aut | |
| 700 | 1 | |a Struyven, Robbert R |e verfasserin |4 aut | |
| 700 | 1 | |a Liu, Timing |e verfasserin |4 aut | |
| 700 | 1 | |a Xu, Moucheng |e verfasserin |4 aut | |
| 700 | 1 | |a Lozano, Mateo G |e verfasserin |4 aut | |
| 700 | 1 | |a Woodward-Court, Peter |e verfasserin |4 aut | |
| 700 | 1 | |a Kihara, Yuka |e verfasserin |4 aut | |
| 700 | 0 | |a UK Biobank Eye & Vision Consortium |e verfasserin |4 aut | |
| 700 | 1 | |a Altmann, Andre |e verfasserin |4 aut | |
| 700 | 1 | |a Lee, Aaron Y |e verfasserin |4 aut | |
| 700 | 1 | |a Topol, Eric J |e verfasserin |4 aut | |
| 700 | 1 | |a Denniston, Alastair K |e verfasserin |4 aut | |
| 700 | 1 | |a Alexander, Daniel C |e verfasserin |4 aut | |
| 700 | 1 | |a Keane, Pearse A |e verfasserin |4 aut | |
| 700 | 1 | |a Allen, Naomi |e investigator |4 oth | |
| 700 | 1 | |a Gallacher, John E J |e investigator |4 oth | |
| 700 | 1 | |a Littlejohns, Thomas |e investigator |4 oth | |
| 700 | 1 | |a Aslam, Tariq |e investigator |4 oth | |
| 700 | 1 | |a Bishop, Paul |e investigator |4 oth | |
| 700 | 1 | |a Black, Graeme |e investigator |4 oth | |
| 700 | 1 | |a Sergouniotis, Panagiotis |e investigator |4 oth | |
| 700 | 1 | |a Atan, Denize |e investigator |4 oth | |
| 700 | 1 | |a Dick, Andrew D |e investigator |4 oth | |
| 700 | 1 | |a Williams, Cathy |e investigator |4 oth | |
| 700 | 1 | |a Barman, Sarah |e investigator |4 oth | |
| 700 | 1 | |a Barrett, Jenny H |e investigator |4 oth | |
| 700 | 1 | |a Mackie, Sarah |e investigator |4 oth | |
| 700 | 1 | |a Braithwaite, Tasanee |e investigator |4 oth | |
| 700 | 1 | |a Carare, Roxana O |e investigator |4 oth | |
| 700 | 1 | |a Ennis, Sarah |e investigator |4 oth | |
| 700 | 1 | |a Gibson, Jane |e investigator |4 oth | |
| 700 | 1 | |a Lotery, Andrew J |e investigator |4 oth | |
| 700 | 1 | |a Self, Jay |e investigator |4 oth | |
| 700 | 1 | |a Chakravarthy, Usha |e investigator |4 oth | |
| 700 | 1 | |a Hogg, Ruth E |e investigator |4 oth | |
| 700 | 1 | |a Paterson, Euan |e investigator |4 oth | |
| 700 | 1 | |a Woodside, Jayne |e investigator |4 oth | |
| 700 | 1 | |a Peto, Tunde |e investigator |4 oth | |
| 700 | 1 | |a Mckay, Gareth |e investigator |4 oth | |
| 700 | 1 | |a Mcguinness, Bernadette |e investigator |4 oth | |
| 700 | 1 | |a Foster, Paul J |e investigator |4 oth | |
| 700 | 1 | |a Balaskas, Konstantinos |e investigator |4 oth | |
| 700 | 1 | |a Khawaja, Anthony P |e investigator |4 oth | |
| 700 | 1 | |a Pontikos, Nikolas |e investigator |4 oth | |
| 700 | 1 | |a Rahi, Jugnoo S |e investigator |4 oth | |
| 700 | 1 | |a Lascaratos, Gerassimos |e investigator |4 oth | |
| 700 | 1 | |a Patel, Praveen J |e investigator |4 oth | |
| 700 | 1 | |a Chan, Michelle |e investigator |4 oth | |
| 700 | 1 | |a Chua, Sharon Y L |e investigator |4 oth | |
| 700 | 1 | |a Day, Alexander |e investigator |4 oth | |
| 700 | 1 | |a Desai, Parul |e investigator |4 oth | |
| 700 | 1 | |a Egan, Cathy |e investigator |4 oth | |
| 700 | 1 | |a Fruttiger, Marcus |e investigator |4 oth | |
| 700 | 1 | |a Garway-Heath, David F |e investigator |4 oth | |
| 700 | 1 | |a Hardcastle, Alison |e investigator |4 oth | |
| 700 | 1 | |a Khaw, Sir Peng T |e investigator |4 oth | |
| 700 | 1 | |a Moore, Tony |e investigator |4 oth | |
| 700 | 1 | |a Sivaprasad, Sobha |e investigator |4 oth | |
| 700 | 1 | |a Strouthidis, Nicholas |e investigator |4 oth | |
| 700 | 1 | |a Thomas, Dhanes |e investigator |4 oth | |
| 700 | 1 | |a Tufail, Adnan |e investigator |4 oth | |
| 700 | 1 | |a Viswanathan, Ananth C |e investigator |4 oth | |
| 700 | 1 | |a Dhillon, Bal |e investigator |4 oth | |
| 700 | 1 | |a Macgillivray, Tom |e investigator |4 oth | |
| 700 | 1 | |a Sudlow, Cathie |e investigator |4 oth | |
| 700 | 1 | |a Vitart, Veronique |e investigator |4 oth | |
| 700 | 1 | |a Doney, Alexander |e investigator |4 oth | |
| 700 | 1 | |a Trucco, Emanuele |e investigator |4 oth | |
| 700 | 1 | |a Guggeinheim, Jeremy A |e investigator |4 oth | |
| 700 | 1 | |a Morgan, James E |e investigator |4 oth | |
| 700 | 1 | |a Hammond, Chris J |e investigator |4 oth | |
| 700 | 1 | |a Williams, Katie |e investigator |4 oth | |
| 700 | 1 | |a Hysi, Pirro |e investigator |4 oth | |
| 700 | 1 | |a Harding, Simon P |e investigator |4 oth | |
| 700 | 1 | |a Zheng, Yalin |e investigator |4 oth | |
| 700 | 1 | |a Luben, Robert |e investigator |4 oth | |
| 700 | 1 | |a Luthert, Phil |e investigator |4 oth | |
| 700 | 1 | |a Sun, Zihan |e investigator |4 oth | |
| 700 | 1 | |a McKibbin, Martin |e investigator |4 oth | |
| 700 | 1 | |a O'Sullivan, Eoin |e investigator |4 oth | |
| 700 | 1 | |a Oram, Richard |e investigator |4 oth | |
| 700 | 1 | |a Weedon, Mike |e investigator |4 oth | |
| 700 | 1 | |a Owen, Chris G |e investigator |4 oth | |
| 700 | 1 | |a Rudnicka, Alicja R |e investigator |4 oth | |
| 700 | 1 | |a Sattar, Naveed |e investigator |4 oth | |
| 700 | 1 | |a Steel, David |e investigator |4 oth | |
| 700 | 1 | |a Stratton, Irene |e investigator |4 oth | |
| 700 | 1 | |a Tapp, Robyn |e investigator |4 oth | |
| 700 | 1 | |a Yates, Max M |e investigator |4 oth | |
| 700 | 1 | |a Petzold, Axel |e investigator |4 oth | |
| 700 | 1 | |a Madhusudhan, Savita |e investigator |4 oth | |
| 773 | 0 | 8 | |i Enthalten in |t Nature |d 1945 |g 622(2023), 7981 vom: 13. Okt., Seite 156-163 |w (DE-627)NLM000008257 |x 1476-4687 |7 nnns |
| 773 | 1 | 8 | |g volume:622 |g year:2023 |g number:7981 |g day:13 |g month:10 |g pages:156-163 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1038/s41586-023-06555-x |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 622 |j 2023 |e 7981 |b 13 |c 10 |h 156-163 | ||