Loading…
Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images
The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospec...
Saved in:
| Published in: | American journal of ophthalmology 2020-09, Vol.217, p.121-130 |
|---|---|
| Main Authors: | , , , , , , , , , , , , , |
| Format: | Article |
| Language: | English |
| Subjects: | |
| Citations: | Items that this one cites Items that cite this one |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| cited_by | cdi_FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203 |
|---|---|
| cites | cdi_FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203 |
| container_end_page | 130 |
| container_issue | |
| container_start_page | 121 |
| container_title | American journal of ophthalmology |
| container_volume | 217 |
| creator | Chang, Jooyoung Ko, Ahryoung Park, Sang Min Choi, Seulggie Kim, Kyuwoong Kim, Sung Min Yun, Jae Moon Kang, Uk Shin, Il Hyung Shin, Joo Young Ko, Taehoon Lee, Jinho Oh, Baek-Lok Park, Ki Ho |
| description | The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospective cohort analysis.
Retrospective cohort study.
The database at the Health Promotion Center of Seoul National University Hospital (HPC-SNUH) was used. The deep learning model was trained using 15,408 images to predict carotid artery atherosclerosis, which was named the deep-learning funduscopic atherosclerosis score (DL-FAS). A retrospective cohort was constructed of participants 30-80 years old who had completed elective health examinations at HPC-SNUH. Using DL-FAS as the main exposure, participants were followed for the primary outcome of death due to CVD until Dec. 31, 2017.
For predicting carotid artery atherosclerosis among subjects, the model achieved an area under receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), accuracy, sensitivity, specificity, positive and negative predictive values of 0.713, 0.569, 0.583, 0.891, 0.404, 0.465, and 0.865 respectively. The cohort consisted of 32,227 participants, 78 cardiovascular disease (CVD) deaths, and 7.6-year median follow-up visits. Those with DL-FAS greater than 0.66 had an increased risk of CVD deaths compared to those with DL-FAS |
| doi_str_mv | 10.1016/j.ajo.2020.03.027 |
| format | article |
| fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2384495070</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0002939420301276</els_id><sourcerecordid>2441250087</sourcerecordid><originalsourceid>FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203</originalsourceid><addsrcrecordid>eNp9kU-L1DAYh4Mo7rj6AbxIwIuX1jdJ2zR4GmbddWFE8M85pMnbNaVtxqQd2Iuf3Qyz68GDOSQEnt-PN3kIec2gZMCa90NphlBy4FCCKIHLJ2TDWqkK1ir2lGwAgBdKqOqCvEhpyNdGVvI5uRA8LyFhQ35vUwrWm8WHmYae7kx0PhxNsutoIv0c4mJGv9xTMzt6hXigezRx9vNdcb3Obk02HLyl2-UnxpDseNp9ot9siEgdRn9ER_sYJvoVFz-bkZ5j9HYyd5hekme9GRO-ejgvyY_rj993n4r9l5vb3XZf2KqSS6HQsq5h0LEOEFTVChTWQs2Za0zXcdX0TWtk7aStBZNKCiW54G3LeycVB3FJ3p17DzH8WjEtevLJ4jiaGcOaNBdtVaka5Al9-w86hDXmyTNVVYzXAK3MFDtTNj84Rez1IfrJxHvNQJ_k6EFnOfokR4PQWU7OvHloXrsJ3d_Eo40MfDgDmL_i6DHqZD3OFp2PaBftgv9P_R-XUp-Z</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2441250087</pqid></control><display><type>article</type><title>Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images</title><source>ScienceDirect Journals</source><creator>Chang, Jooyoung ; Ko, Ahryoung ; Park, Sang Min ; Choi, Seulggie ; Kim, Kyuwoong ; Kim, Sung Min ; Yun, Jae Moon ; Kang, Uk ; Shin, Il Hyung ; Shin, Joo Young ; Ko, Taehoon ; Lee, Jinho ; Oh, Baek-Lok ; Park, Ki Ho</creator><creatorcontrib>Chang, Jooyoung ; Ko, Ahryoung ; Park, Sang Min ; Choi, Seulggie ; Kim, Kyuwoong ; Kim, Sung Min ; Yun, Jae Moon ; Kang, Uk ; Shin, Il Hyung ; Shin, Joo Young ; Ko, Taehoon ; Lee, Jinho ; Oh, Baek-Lok ; Park, Ki Ho</creatorcontrib><description>The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospective cohort analysis.
Retrospective cohort study.
The database at the Health Promotion Center of Seoul National University Hospital (HPC-SNUH) was used. The deep learning model was trained using 15,408 images to predict carotid artery atherosclerosis, which was named the deep-learning funduscopic atherosclerosis score (DL-FAS). A retrospective cohort was constructed of participants 30-80 years old who had completed elective health examinations at HPC-SNUH. Using DL-FAS as the main exposure, participants were followed for the primary outcome of death due to CVD until Dec. 31, 2017.
For predicting carotid artery atherosclerosis among subjects, the model achieved an area under receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), accuracy, sensitivity, specificity, positive and negative predictive values of 0.713, 0.569, 0.583, 0.891, 0.404, 0.465, and 0.865 respectively. The cohort consisted of 32,227 participants, 78 cardiovascular disease (CVD) deaths, and 7.6-year median follow-up visits. Those with DL-FAS greater than 0.66 had an increased risk of CVD deaths compared to those with DL-FAS <0.33 (hazard ratio: 8.33; 95% confidence interval [CI], 3.16-24.7). Risk association was significant among intermediate and high Framingham risk score (FRS) subgroups. The DL-FAS improved the concordance by 0.0266 (95% CI, 0.0043-0.0489) over the FRS-only model. The relative integrated discrimination index was 20.45% and net reclassification index was 29.5%.
A deep learning model was developed which could predict atherosclerosis from retinal fundus images. The resulting DL-FAS was an independent predictor of CVD deaths when adjusted for FRS and added predictive value over FRS.
•Retinal fundus imaging and deep learning may be used for stratification of CVD risk.•Deep learning added predictive value compared with conventional CVD risk scoring methods.•The developed model was verified in a large cohort of 30,000 Koreans.</description><identifier>ISSN: 0002-9394</identifier><identifier>EISSN: 1879-1891</identifier><identifier>DOI: 10.1016/j.ajo.2020.03.027</identifier><identifier>PMID: 32222370</identifier><language>eng</language><publisher>United States: Elsevier Inc</publisher><subject>Atherosclerosis ; Carotid arteries ; Deep learning ; Diabetes ; Hypertension ; Mortality ; Patients ; Risk factors ; Ultrasonic imaging ; Veins & arteries</subject><ispartof>American journal of ophthalmology, 2020-09, Vol.217, p.121-130</ispartof><rights>2020 Elsevier Inc.</rights><rights>Copyright © 2020 Elsevier Inc. All rights reserved.</rights><rights>2020. Elsevier Inc.</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203</citedby><cites>FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/32222370$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Chang, Jooyoung</creatorcontrib><creatorcontrib>Ko, Ahryoung</creatorcontrib><creatorcontrib>Park, Sang Min</creatorcontrib><creatorcontrib>Choi, Seulggie</creatorcontrib><creatorcontrib>Kim, Kyuwoong</creatorcontrib><creatorcontrib>Kim, Sung Min</creatorcontrib><creatorcontrib>Yun, Jae Moon</creatorcontrib><creatorcontrib>Kang, Uk</creatorcontrib><creatorcontrib>Shin, Il Hyung</creatorcontrib><creatorcontrib>Shin, Joo Young</creatorcontrib><creatorcontrib>Ko, Taehoon</creatorcontrib><creatorcontrib>Lee, Jinho</creatorcontrib><creatorcontrib>Oh, Baek-Lok</creatorcontrib><creatorcontrib>Park, Ki Ho</creatorcontrib><title>Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images</title><title>American journal of ophthalmology</title><addtitle>Am J Ophthalmol</addtitle><description>The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospective cohort analysis.
Retrospective cohort study.
The database at the Health Promotion Center of Seoul National University Hospital (HPC-SNUH) was used. The deep learning model was trained using 15,408 images to predict carotid artery atherosclerosis, which was named the deep-learning funduscopic atherosclerosis score (DL-FAS). A retrospective cohort was constructed of participants 30-80 years old who had completed elective health examinations at HPC-SNUH. Using DL-FAS as the main exposure, participants were followed for the primary outcome of death due to CVD until Dec. 31, 2017.
For predicting carotid artery atherosclerosis among subjects, the model achieved an area under receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), accuracy, sensitivity, specificity, positive and negative predictive values of 0.713, 0.569, 0.583, 0.891, 0.404, 0.465, and 0.865 respectively. The cohort consisted of 32,227 participants, 78 cardiovascular disease (CVD) deaths, and 7.6-year median follow-up visits. Those with DL-FAS greater than 0.66 had an increased risk of CVD deaths compared to those with DL-FAS <0.33 (hazard ratio: 8.33; 95% confidence interval [CI], 3.16-24.7). Risk association was significant among intermediate and high Framingham risk score (FRS) subgroups. The DL-FAS improved the concordance by 0.0266 (95% CI, 0.0043-0.0489) over the FRS-only model. The relative integrated discrimination index was 20.45% and net reclassification index was 29.5%.
A deep learning model was developed which could predict atherosclerosis from retinal fundus images. The resulting DL-FAS was an independent predictor of CVD deaths when adjusted for FRS and added predictive value over FRS.
•Retinal fundus imaging and deep learning may be used for stratification of CVD risk.•Deep learning added predictive value compared with conventional CVD risk scoring methods.•The developed model was verified in a large cohort of 30,000 Koreans.</description><subject>Atherosclerosis</subject><subject>Carotid arteries</subject><subject>Deep learning</subject><subject>Diabetes</subject><subject>Hypertension</subject><subject>Mortality</subject><subject>Patients</subject><subject>Risk factors</subject><subject>Ultrasonic imaging</subject><subject>Veins & arteries</subject><issn>0002-9394</issn><issn>1879-1891</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNp9kU-L1DAYh4Mo7rj6AbxIwIuX1jdJ2zR4GmbddWFE8M85pMnbNaVtxqQd2Iuf3Qyz68GDOSQEnt-PN3kIec2gZMCa90NphlBy4FCCKIHLJ2TDWqkK1ir2lGwAgBdKqOqCvEhpyNdGVvI5uRA8LyFhQ35vUwrWm8WHmYae7kx0PhxNsutoIv0c4mJGv9xTMzt6hXigezRx9vNdcb3Obk02HLyl2-UnxpDseNp9ot9siEgdRn9ER_sYJvoVFz-bkZ5j9HYyd5hekme9GRO-ejgvyY_rj993n4r9l5vb3XZf2KqSS6HQsq5h0LEOEFTVChTWQs2Za0zXcdX0TWtk7aStBZNKCiW54G3LeycVB3FJ3p17DzH8WjEtevLJ4jiaGcOaNBdtVaka5Al9-w86hDXmyTNVVYzXAK3MFDtTNj84Rez1IfrJxHvNQJ_k6EFnOfokR4PQWU7OvHloXrsJ3d_Eo40MfDgDmL_i6DHqZD3OFp2PaBftgv9P_R-XUp-Z</recordid><startdate>20200901</startdate><enddate>20200901</enddate><creator>Chang, Jooyoung</creator><creator>Ko, Ahryoung</creator><creator>Park, Sang Min</creator><creator>Choi, Seulggie</creator><creator>Kim, Kyuwoong</creator><creator>Kim, Sung Min</creator><creator>Yun, Jae Moon</creator><creator>Kang, Uk</creator><creator>Shin, Il Hyung</creator><creator>Shin, Joo Young</creator><creator>Ko, Taehoon</creator><creator>Lee, Jinho</creator><creator>Oh, Baek-Lok</creator><creator>Park, Ki Ho</creator><general>Elsevier Inc</general><general>Elsevier Limited</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>K9.</scope><scope>NAPCQ</scope><scope>7X8</scope></search><sort><creationdate>20200901</creationdate><title>Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images</title><author>Chang, Jooyoung ; Ko, Ahryoung ; Park, Sang Min ; Choi, Seulggie ; Kim, Kyuwoong ; Kim, Sung Min ; Yun, Jae Moon ; Kang, Uk ; Shin, Il Hyung ; Shin, Joo Young ; Ko, Taehoon ; Lee, Jinho ; Oh, Baek-Lok ; Park, Ki Ho</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Atherosclerosis</topic><topic>Carotid arteries</topic><topic>Deep learning</topic><topic>Diabetes</topic><topic>Hypertension</topic><topic>Mortality</topic><topic>Patients</topic><topic>Risk factors</topic><topic>Ultrasonic imaging</topic><topic>Veins & arteries</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Chang, Jooyoung</creatorcontrib><creatorcontrib>Ko, Ahryoung</creatorcontrib><creatorcontrib>Park, Sang Min</creatorcontrib><creatorcontrib>Choi, Seulggie</creatorcontrib><creatorcontrib>Kim, Kyuwoong</creatorcontrib><creatorcontrib>Kim, Sung Min</creatorcontrib><creatorcontrib>Yun, Jae Moon</creatorcontrib><creatorcontrib>Kang, Uk</creatorcontrib><creatorcontrib>Shin, Il Hyung</creatorcontrib><creatorcontrib>Shin, Joo Young</creatorcontrib><creatorcontrib>Ko, Taehoon</creatorcontrib><creatorcontrib>Lee, Jinho</creatorcontrib><creatorcontrib>Oh, Baek-Lok</creatorcontrib><creatorcontrib>Park, Ki Ho</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>ProQuest Health & Medical Complete (Alumni)</collection><collection>Nursing & Allied Health Premium</collection><collection>MEDLINE - Academic</collection><jtitle>American journal of ophthalmology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Chang, Jooyoung</au><au>Ko, Ahryoung</au><au>Park, Sang Min</au><au>Choi, Seulggie</au><au>Kim, Kyuwoong</au><au>Kim, Sung Min</au><au>Yun, Jae Moon</au><au>Kang, Uk</au><au>Shin, Il Hyung</au><au>Shin, Joo Young</au><au>Ko, Taehoon</au><au>Lee, Jinho</au><au>Oh, Baek-Lok</au><au>Park, Ki Ho</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images</atitle><jtitle>American journal of ophthalmology</jtitle><addtitle>Am J Ophthalmol</addtitle><date>2020-09-01</date><risdate>2020</risdate><volume>217</volume><spage>121</spage><epage>130</epage><pages>121-130</pages><issn>0002-9394</issn><eissn>1879-1891</eissn><abstract>The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospective cohort analysis.
Retrospective cohort study.
The database at the Health Promotion Center of Seoul National University Hospital (HPC-SNUH) was used. The deep learning model was trained using 15,408 images to predict carotid artery atherosclerosis, which was named the deep-learning funduscopic atherosclerosis score (DL-FAS). A retrospective cohort was constructed of participants 30-80 years old who had completed elective health examinations at HPC-SNUH. Using DL-FAS as the main exposure, participants were followed for the primary outcome of death due to CVD until Dec. 31, 2017.
For predicting carotid artery atherosclerosis among subjects, the model achieved an area under receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), accuracy, sensitivity, specificity, positive and negative predictive values of 0.713, 0.569, 0.583, 0.891, 0.404, 0.465, and 0.865 respectively. The cohort consisted of 32,227 participants, 78 cardiovascular disease (CVD) deaths, and 7.6-year median follow-up visits. Those with DL-FAS greater than 0.66 had an increased risk of CVD deaths compared to those with DL-FAS <0.33 (hazard ratio: 8.33; 95% confidence interval [CI], 3.16-24.7). Risk association was significant among intermediate and high Framingham risk score (FRS) subgroups. The DL-FAS improved the concordance by 0.0266 (95% CI, 0.0043-0.0489) over the FRS-only model. The relative integrated discrimination index was 20.45% and net reclassification index was 29.5%.
A deep learning model was developed which could predict atherosclerosis from retinal fundus images. The resulting DL-FAS was an independent predictor of CVD deaths when adjusted for FRS and added predictive value over FRS.
•Retinal fundus imaging and deep learning may be used for stratification of CVD risk.•Deep learning added predictive value compared with conventional CVD risk scoring methods.•The developed model was verified in a large cohort of 30,000 Koreans.</abstract><cop>United States</cop><pub>Elsevier Inc</pub><pmid>32222370</pmid><doi>10.1016/j.ajo.2020.03.027</doi><tpages>10</tpages></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 0002-9394 |
| ispartof | American journal of ophthalmology, 2020-09, Vol.217, p.121-130 |
| issn | 0002-9394 1879-1891 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_2384495070 |
| source | ScienceDirect Journals |
| subjects | Atherosclerosis Carotid arteries Deep learning Diabetes Hypertension Mortality Patients Risk factors Ultrasonic imaging Veins & arteries |
| title | Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images |
| url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-28T06%3A38%3A20IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Association%20of%20Cardiovascular%20Mortality%20and%20Deep%20Learning-Funduscopic%20Atherosclerosis%20Score%20derived%20from%20Retinal%20Fundus%20Images&rft.jtitle=American%20journal%20of%20ophthalmology&rft.au=Chang,%20Jooyoung&rft.date=2020-09-01&rft.volume=217&rft.spage=121&rft.epage=130&rft.pages=121-130&rft.issn=0002-9394&rft.eissn=1879-1891&rft_id=info:doi/10.1016/j.ajo.2020.03.027&rft_dat=%3Cproquest_cross%3E2441250087%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c447t-9ec1b610b1b0e09483e3cc0521d6abb296f68a75d7c531797397232882fd79203%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2441250087&rft_id=info:pmid/32222370&rfr_iscdi=true |