Artificial intelligence for detecting keratoconus
Background Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends...
Gespeichert in:
| Veröffentlicht in: | Cochrane database of systematic reviews 2023-11, Vol.2023 (11), p.CD014911 |
|---|---|
| Hauptverfasser: | , , , , , , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | 11 |
| container_start_page | CD014911 |
| container_title | Cochrane database of systematic reviews |
| container_volume | 2023 |
| creator | Dickman, Mor M Vandevenne, Magali MS Favuzza, Eleonora Veta, Mitko Lucenteforte, Ersilia Berendschot, Tos TJM Mencucci, Rita Nuijts, Rudy MMA Virgili, Gianni Dickman, Mor M |
| description | Background
Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends on the interpretation of corneal imaging (e.g. topography and tomography) by trained cornea specialists. Using artificial intelligence (AI) to analyse the corneal images and detect cases of keratoconus could help prevent visual acuity loss and even corneal transplantation. However, a missed diagnosis in people seeking refractive surgery could lead to weakening of the cornea and keratoconus‐like ectasia. There is a need for a reliable overview of the accuracy of AI for detecting keratoconus and the applicability of this automated method to the clinical setting.
Objectives
To assess the diagnostic accuracy of artificial intelligence (AI) algorithms for detecting keratoconus in people presenting with refractive errors, especially those whose vision can no longer be fully corrected with glasses, those seeking corneal refractive surgery, and those suspected of having keratoconus. AI could help ophthalmologists, optometrists, and other eye care professionals to make decisions on referral to cornea specialists.
Secondary objectives
To assess the following potential causes of heterogeneity in diagnostic performance across studies.
• Different AI algorithms (e.g. neural networks, decision trees, support vector machines)
• Index test methodology (preprocessing techniques, core AI method, and postprocessing techniques)
• Sources of input to train algorithms (topography and tomography images from Placido disc system, Scheimpflug system, slit‐scanning system, or optical coherence tomography (OCT); number of training and testing cases/images; label/endpoint variable used for training)
• Study setting
• Study design
• Ethnicity, or geographic area as its proxy
• Different index test positivity criteria provided by the topography or tomography device
• Reference standard, topography or tomography, one or two cornea specialists
• Definition of keratoconus
• Mean age of participants
• Recruitment of participants
• Severity of keratoconus (clinically manifest or subclinical)
Search methods
We searched CENTRAL (which contains the Cochrane Eyes and Vision Trials Register), Ovid MEDLINE, Ovid Embase, OpenGrey, the ISRCTN registry, ClinicalTrials.gov, and the World Health Organizatio |
| doi_str_mv | 10.1002/14651858.CD014911.pub2 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_pubme</sourceid><recordid>TN_cdi_pubmedcentral_primary_oai_pubmedcentral_nih_gov_10646985</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2890363722</sourcerecordid><originalsourceid>FETCH-LOGICAL-c4212-56cd1931af21359d243d3d6e3c8b288cc89ef1b23ba3d7a75c3c76dc5e74daf03</originalsourceid><addsrcrecordid>eNqFkElPwzAQhS0EomX5C1WOXFK8JE58QqWsUiUucLacsdMa0rjYCaj_nkRdVLhwsqX35r2ZD6ERwWOCMb0mCU9Jnubj6R0miSBkvGoLeoSGvRD3yvHBf4DOQnjHmHFBs1M0YJngqeB4iMjEN7a0YFUV2boxVWXnpgYTlc5H2jQGGlvPow_jVePA1W24QCelqoK53L7n6O3h_nX6FM9eHp-nk1kMCSU0TjloIhhRJSUsFZomTDPNDYO8oHkOkAtTkoKyQjGdqSwFBhnXkJos0arE7BzdbHK7w5ZGg6kbryq58nap_Fo6ZeVvpbYLOXdfkmCecJGnXcLVNsG7z9aERi5tgO5EVRvXBklz0RFhGaWdlW-s4F0I3pT7HoJlD1zugMsd8L69Hxwdbrkf2xHuDLcbw7etzFqCg4Xv-v_J_dPyAy-okdc</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2890363722</pqid></control><display><type>article</type><title>Artificial intelligence for detecting keratoconus</title><source>MEDLINE</source><source>EZB-FREE-00999 freely available EZB journals</source><source>Alma/SFX Local Collection</source><source>Cochrane Library</source><creator>Dickman, Mor M ; Vandevenne, Magali MS ; Favuzza, Eleonora ; Veta, Mitko ; Lucenteforte, Ersilia ; Berendschot, Tos TJM ; Mencucci, Rita ; Nuijts, Rudy MMA ; Virgili, Gianni ; Dickman, Mor M</creator><creatorcontrib>Dickman, Mor M ; Vandevenne, Magali MS ; Favuzza, Eleonora ; Veta, Mitko ; Lucenteforte, Ersilia ; Berendschot, Tos TJM ; Mencucci, Rita ; Nuijts, Rudy MMA ; Virgili, Gianni ; Dickman, Mor M</creatorcontrib><description>Background
Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends on the interpretation of corneal imaging (e.g. topography and tomography) by trained cornea specialists. Using artificial intelligence (AI) to analyse the corneal images and detect cases of keratoconus could help prevent visual acuity loss and even corneal transplantation. However, a missed diagnosis in people seeking refractive surgery could lead to weakening of the cornea and keratoconus‐like ectasia. There is a need for a reliable overview of the accuracy of AI for detecting keratoconus and the applicability of this automated method to the clinical setting.
Objectives
To assess the diagnostic accuracy of artificial intelligence (AI) algorithms for detecting keratoconus in people presenting with refractive errors, especially those whose vision can no longer be fully corrected with glasses, those seeking corneal refractive surgery, and those suspected of having keratoconus. AI could help ophthalmologists, optometrists, and other eye care professionals to make decisions on referral to cornea specialists.
Secondary objectives
To assess the following potential causes of heterogeneity in diagnostic performance across studies.
• Different AI algorithms (e.g. neural networks, decision trees, support vector machines)
• Index test methodology (preprocessing techniques, core AI method, and postprocessing techniques)
• Sources of input to train algorithms (topography and tomography images from Placido disc system, Scheimpflug system, slit‐scanning system, or optical coherence tomography (OCT); number of training and testing cases/images; label/endpoint variable used for training)
• Study setting
• Study design
• Ethnicity, or geographic area as its proxy
• Different index test positivity criteria provided by the topography or tomography device
• Reference standard, topography or tomography, one or two cornea specialists
• Definition of keratoconus
• Mean age of participants
• Recruitment of participants
• Severity of keratoconus (clinically manifest or subclinical)
Search methods
We searched CENTRAL (which contains the Cochrane Eyes and Vision Trials Register), Ovid MEDLINE, Ovid Embase, OpenGrey, the ISRCTN registry, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP). There were no date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 29 November 2022.
Selection criteria
We included cross‐sectional and diagnostic case‐control studies that investigated AI for the diagnosis of keratoconus using topography, tomography, or both. We included studies that diagnosed manifest keratoconus, subclinical keratoconus, or both. The reference standard was the interpretation of topography or tomography images by at least two cornea specialists.
Data collection and analysis
Two review authors independently extracted the study data and assessed the quality of studies using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS‐2) tool. When an article contained multiple AI algorithms, we selected the algorithm with the highest Youden's index. We assessed the certainty of evidence using the GRADE approach.
Main results
We included 63 studies, published between 1994 and 2022, that developed and investigated the accuracy of AI for the diagnosis of keratoconus. There were three different units of analysis in the studies: eyes, participants, and images. Forty‐four studies analysed 23,771 eyes, four studies analysed 3843 participants, and 15 studies analysed 38,832 images.
Fifty‐four articles evaluated the detection of manifest keratoconus, defined as a cornea that showed any clinical sign of keratoconus. The accuracy of AI seems almost perfect, with a summary sensitivity of 98.6% (95% confidence interval (CI) 97.6% to 99.1%) and a summary specificity of 98.3% (95% CI 97.4% to 98.9%). However, accuracy varied across studies and the certainty of the evidence was low.
Twenty‐eight articles evaluated the detection of subclinical keratoconus, although the definition of subclinical varied. We grouped subclinical keratoconus, forme fruste, and very asymmetrical eyes together. The tests showed good accuracy, with a summary sensitivity of 90.0% (95% CI 84.5% to 93.8%) and a summary specificity of 95.5% (95% CI 91.9% to 97.5%). However, the certainty of the evidence was very low for sensitivity and low for specificity.
In both groups, we graded most studies at high risk of bias, with high applicability concerns, in the domain of patient selection, since most were case‐control studies. Moreover, we graded the certainty of evidence as low to very low due to selection bias, inconsistency, and imprecision.
We could not explain the heterogeneity between the studies. The sensitivity analyses based on study design, AI algorithm, imaging technique (topography versus tomography), and data source (parameters versus images) showed no differences in the results.
Authors' conclusions
AI appears to be a promising triage tool in ophthalmologic practice for diagnosing keratoconus. Test accuracy was very high for manifest keratoconus and slightly lower for subclinical keratoconus, indicating a higher chance of missing a diagnosis in people without clinical signs. This could lead to progression of keratoconus or an erroneous indication for refractive surgery, which would worsen the disease.
We are unable to draw clear and reliable conclusions due to the high risk of bias, the unexplained heterogeneity of the results, and high applicability concerns, all of which reduced our confidence in the evidence.
Greater standardization in future research would increase the quality of studies and improve comparability between studies.</description><identifier>ISSN: 1465-1858</identifier><identifier>ISSN: 1469-493X</identifier><identifier>EISSN: 1465-1858</identifier><identifier>EISSN: 1469-493X</identifier><identifier>DOI: 10.1002/14651858.CD014911.pub2</identifier><identifier>PMID: 37965960</identifier><language>eng</language><publisher>Chichester, UK: John Wiley & Sons, Ltd</publisher><subject>Artificial Intelligence ; Case-Control Studies ; Cornea ; Cross-Sectional Studies ; Eyes & vision ; Humans ; Keratoconus ; Keratoconus - diagnostic imaging ; Medicine General & Introductory Medical Sciences ; Outer eye ; Physical Examination</subject><ispartof>Cochrane database of systematic reviews, 2023-11, Vol.2023 (11), p.CD014911</ispartof><rights>Copyright © 2023 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c4212-56cd1931af21359d243d3d6e3c8b288cc89ef1b23ba3d7a75c3c76dc5e74daf03</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,314,776,780,881,27901,27902</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/37965960$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Dickman, Mor M</creatorcontrib><creatorcontrib>Vandevenne, Magali MS</creatorcontrib><creatorcontrib>Favuzza, Eleonora</creatorcontrib><creatorcontrib>Veta, Mitko</creatorcontrib><creatorcontrib>Lucenteforte, Ersilia</creatorcontrib><creatorcontrib>Berendschot, Tos TJM</creatorcontrib><creatorcontrib>Mencucci, Rita</creatorcontrib><creatorcontrib>Nuijts, Rudy MMA</creatorcontrib><creatorcontrib>Virgili, Gianni</creatorcontrib><creatorcontrib>Dickman, Mor M</creatorcontrib><title>Artificial intelligence for detecting keratoconus</title><title>Cochrane database of systematic reviews</title><addtitle>Cochrane Database Syst Rev</addtitle><description>Background
Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends on the interpretation of corneal imaging (e.g. topography and tomography) by trained cornea specialists. Using artificial intelligence (AI) to analyse the corneal images and detect cases of keratoconus could help prevent visual acuity loss and even corneal transplantation. However, a missed diagnosis in people seeking refractive surgery could lead to weakening of the cornea and keratoconus‐like ectasia. There is a need for a reliable overview of the accuracy of AI for detecting keratoconus and the applicability of this automated method to the clinical setting.
Objectives
To assess the diagnostic accuracy of artificial intelligence (AI) algorithms for detecting keratoconus in people presenting with refractive errors, especially those whose vision can no longer be fully corrected with glasses, those seeking corneal refractive surgery, and those suspected of having keratoconus. AI could help ophthalmologists, optometrists, and other eye care professionals to make decisions on referral to cornea specialists.
Secondary objectives
To assess the following potential causes of heterogeneity in diagnostic performance across studies.
• Different AI algorithms (e.g. neural networks, decision trees, support vector machines)
• Index test methodology (preprocessing techniques, core AI method, and postprocessing techniques)
• Sources of input to train algorithms (topography and tomography images from Placido disc system, Scheimpflug system, slit‐scanning system, or optical coherence tomography (OCT); number of training and testing cases/images; label/endpoint variable used for training)
• Study setting
• Study design
• Ethnicity, or geographic area as its proxy
• Different index test positivity criteria provided by the topography or tomography device
• Reference standard, topography or tomography, one or two cornea specialists
• Definition of keratoconus
• Mean age of participants
• Recruitment of participants
• Severity of keratoconus (clinically manifest or subclinical)
Search methods
We searched CENTRAL (which contains the Cochrane Eyes and Vision Trials Register), Ovid MEDLINE, Ovid Embase, OpenGrey, the ISRCTN registry, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP). There were no date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 29 November 2022.
Selection criteria
We included cross‐sectional and diagnostic case‐control studies that investigated AI for the diagnosis of keratoconus using topography, tomography, or both. We included studies that diagnosed manifest keratoconus, subclinical keratoconus, or both. The reference standard was the interpretation of topography or tomography images by at least two cornea specialists.
Data collection and analysis
Two review authors independently extracted the study data and assessed the quality of studies using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS‐2) tool. When an article contained multiple AI algorithms, we selected the algorithm with the highest Youden's index. We assessed the certainty of evidence using the GRADE approach.
Main results
We included 63 studies, published between 1994 and 2022, that developed and investigated the accuracy of AI for the diagnosis of keratoconus. There were three different units of analysis in the studies: eyes, participants, and images. Forty‐four studies analysed 23,771 eyes, four studies analysed 3843 participants, and 15 studies analysed 38,832 images.
Fifty‐four articles evaluated the detection of manifest keratoconus, defined as a cornea that showed any clinical sign of keratoconus. The accuracy of AI seems almost perfect, with a summary sensitivity of 98.6% (95% confidence interval (CI) 97.6% to 99.1%) and a summary specificity of 98.3% (95% CI 97.4% to 98.9%). However, accuracy varied across studies and the certainty of the evidence was low.
Twenty‐eight articles evaluated the detection of subclinical keratoconus, although the definition of subclinical varied. We grouped subclinical keratoconus, forme fruste, and very asymmetrical eyes together. The tests showed good accuracy, with a summary sensitivity of 90.0% (95% CI 84.5% to 93.8%) and a summary specificity of 95.5% (95% CI 91.9% to 97.5%). However, the certainty of the evidence was very low for sensitivity and low for specificity.
In both groups, we graded most studies at high risk of bias, with high applicability concerns, in the domain of patient selection, since most were case‐control studies. Moreover, we graded the certainty of evidence as low to very low due to selection bias, inconsistency, and imprecision.
We could not explain the heterogeneity between the studies. The sensitivity analyses based on study design, AI algorithm, imaging technique (topography versus tomography), and data source (parameters versus images) showed no differences in the results.
Authors' conclusions
AI appears to be a promising triage tool in ophthalmologic practice for diagnosing keratoconus. Test accuracy was very high for manifest keratoconus and slightly lower for subclinical keratoconus, indicating a higher chance of missing a diagnosis in people without clinical signs. This could lead to progression of keratoconus or an erroneous indication for refractive surgery, which would worsen the disease.
We are unable to draw clear and reliable conclusions due to the high risk of bias, the unexplained heterogeneity of the results, and high applicability concerns, all of which reduced our confidence in the evidence.
Greater standardization in future research would increase the quality of studies and improve comparability between studies.</description><subject>Artificial Intelligence</subject><subject>Case-Control Studies</subject><subject>Cornea</subject><subject>Cross-Sectional Studies</subject><subject>Eyes & vision</subject><subject>Humans</subject><subject>Keratoconus</subject><subject>Keratoconus - diagnostic imaging</subject><subject>Medicine General & Introductory Medical Sciences</subject><subject>Outer eye</subject><subject>Physical Examination</subject><issn>1465-1858</issn><issn>1469-493X</issn><issn>1465-1858</issn><issn>1469-493X</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>RWY</sourceid><sourceid>EIF</sourceid><recordid>eNqFkElPwzAQhS0EomX5C1WOXFK8JE58QqWsUiUucLacsdMa0rjYCaj_nkRdVLhwsqX35r2ZD6ERwWOCMb0mCU9Jnubj6R0miSBkvGoLeoSGvRD3yvHBf4DOQnjHmHFBs1M0YJngqeB4iMjEN7a0YFUV2boxVWXnpgYTlc5H2jQGGlvPow_jVePA1W24QCelqoK53L7n6O3h_nX6FM9eHp-nk1kMCSU0TjloIhhRJSUsFZomTDPNDYO8oHkOkAtTkoKyQjGdqSwFBhnXkJos0arE7BzdbHK7w5ZGg6kbryq58nap_Fo6ZeVvpbYLOXdfkmCecJGnXcLVNsG7z9aERi5tgO5EVRvXBklz0RFhGaWdlW-s4F0I3pT7HoJlD1zugMsd8L69Hxwdbrkf2xHuDLcbw7etzFqCg4Xv-v_J_dPyAy-okdc</recordid><startdate>20231115</startdate><enddate>20231115</enddate><creator>Dickman, Mor M</creator><creator>Vandevenne, Magali MS</creator><creator>Favuzza, Eleonora</creator><creator>Veta, Mitko</creator><creator>Lucenteforte, Ersilia</creator><creator>Berendschot, Tos TJM</creator><creator>Mencucci, Rita</creator><creator>Nuijts, Rudy MMA</creator><creator>Virgili, Gianni</creator><creator>Dickman, Mor M</creator><general>John Wiley & Sons, Ltd</general><scope>7PX</scope><scope>RWY</scope><scope>ZYTZH</scope><scope>CGR</scope><scope>CUY</scope><scope>CVF</scope><scope>ECM</scope><scope>EIF</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7X8</scope><scope>5PM</scope></search><sort><creationdate>20231115</creationdate><title>Artificial intelligence for detecting keratoconus</title><author>Dickman, Mor M ; Vandevenne, Magali MS ; Favuzza, Eleonora ; Veta, Mitko ; Lucenteforte, Ersilia ; Berendschot, Tos TJM ; Mencucci, Rita ; Nuijts, Rudy MMA ; Virgili, Gianni ; Dickman, Mor M</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c4212-56cd1931af21359d243d3d6e3c8b288cc89ef1b23ba3d7a75c3c76dc5e74daf03</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Artificial Intelligence</topic><topic>Case-Control Studies</topic><topic>Cornea</topic><topic>Cross-Sectional Studies</topic><topic>Eyes & vision</topic><topic>Humans</topic><topic>Keratoconus</topic><topic>Keratoconus - diagnostic imaging</topic><topic>Medicine General & Introductory Medical Sciences</topic><topic>Outer eye</topic><topic>Physical Examination</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Dickman, Mor M</creatorcontrib><creatorcontrib>Vandevenne, Magali MS</creatorcontrib><creatorcontrib>Favuzza, Eleonora</creatorcontrib><creatorcontrib>Veta, Mitko</creatorcontrib><creatorcontrib>Lucenteforte, Ersilia</creatorcontrib><creatorcontrib>Berendschot, Tos TJM</creatorcontrib><creatorcontrib>Mencucci, Rita</creatorcontrib><creatorcontrib>Nuijts, Rudy MMA</creatorcontrib><creatorcontrib>Virgili, Gianni</creatorcontrib><creatorcontrib>Dickman, Mor M</creatorcontrib><collection>Wiley-Blackwell Cochrane Library</collection><collection>Cochrane Library</collection><collection>Cochrane Library (Open Aceess)</collection><collection>Medline</collection><collection>MEDLINE</collection><collection>MEDLINE (Ovid)</collection><collection>MEDLINE</collection><collection>MEDLINE</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>MEDLINE - Academic</collection><collection>PubMed Central (Full Participant titles)</collection><jtitle>Cochrane database of systematic reviews</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Dickman, Mor M</au><au>Vandevenne, Magali MS</au><au>Favuzza, Eleonora</au><au>Veta, Mitko</au><au>Lucenteforte, Ersilia</au><au>Berendschot, Tos TJM</au><au>Mencucci, Rita</au><au>Nuijts, Rudy MMA</au><au>Virgili, Gianni</au><au>Dickman, Mor M</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Artificial intelligence for detecting keratoconus</atitle><jtitle>Cochrane database of systematic reviews</jtitle><addtitle>Cochrane Database Syst Rev</addtitle><date>2023-11-15</date><risdate>2023</risdate><volume>2023</volume><issue>11</issue><spage>CD014911</spage><pages>CD014911-</pages><issn>1465-1858</issn><issn>1469-493X</issn><eissn>1465-1858</eissn><eissn>1469-493X</eissn><abstract>Background
Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends on the interpretation of corneal imaging (e.g. topography and tomography) by trained cornea specialists. Using artificial intelligence (AI) to analyse the corneal images and detect cases of keratoconus could help prevent visual acuity loss and even corneal transplantation. However, a missed diagnosis in people seeking refractive surgery could lead to weakening of the cornea and keratoconus‐like ectasia. There is a need for a reliable overview of the accuracy of AI for detecting keratoconus and the applicability of this automated method to the clinical setting.
Objectives
To assess the diagnostic accuracy of artificial intelligence (AI) algorithms for detecting keratoconus in people presenting with refractive errors, especially those whose vision can no longer be fully corrected with glasses, those seeking corneal refractive surgery, and those suspected of having keratoconus. AI could help ophthalmologists, optometrists, and other eye care professionals to make decisions on referral to cornea specialists.
Secondary objectives
To assess the following potential causes of heterogeneity in diagnostic performance across studies.
• Different AI algorithms (e.g. neural networks, decision trees, support vector machines)
• Index test methodology (preprocessing techniques, core AI method, and postprocessing techniques)
• Sources of input to train algorithms (topography and tomography images from Placido disc system, Scheimpflug system, slit‐scanning system, or optical coherence tomography (OCT); number of training and testing cases/images; label/endpoint variable used for training)
• Study setting
• Study design
• Ethnicity, or geographic area as its proxy
• Different index test positivity criteria provided by the topography or tomography device
• Reference standard, topography or tomography, one or two cornea specialists
• Definition of keratoconus
• Mean age of participants
• Recruitment of participants
• Severity of keratoconus (clinically manifest or subclinical)
Search methods
We searched CENTRAL (which contains the Cochrane Eyes and Vision Trials Register), Ovid MEDLINE, Ovid Embase, OpenGrey, the ISRCTN registry, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP). There were no date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 29 November 2022.
Selection criteria
We included cross‐sectional and diagnostic case‐control studies that investigated AI for the diagnosis of keratoconus using topography, tomography, or both. We included studies that diagnosed manifest keratoconus, subclinical keratoconus, or both. The reference standard was the interpretation of topography or tomography images by at least two cornea specialists.
Data collection and analysis
Two review authors independently extracted the study data and assessed the quality of studies using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS‐2) tool. When an article contained multiple AI algorithms, we selected the algorithm with the highest Youden's index. We assessed the certainty of evidence using the GRADE approach.
Main results
We included 63 studies, published between 1994 and 2022, that developed and investigated the accuracy of AI for the diagnosis of keratoconus. There were three different units of analysis in the studies: eyes, participants, and images. Forty‐four studies analysed 23,771 eyes, four studies analysed 3843 participants, and 15 studies analysed 38,832 images.
Fifty‐four articles evaluated the detection of manifest keratoconus, defined as a cornea that showed any clinical sign of keratoconus. The accuracy of AI seems almost perfect, with a summary sensitivity of 98.6% (95% confidence interval (CI) 97.6% to 99.1%) and a summary specificity of 98.3% (95% CI 97.4% to 98.9%). However, accuracy varied across studies and the certainty of the evidence was low.
Twenty‐eight articles evaluated the detection of subclinical keratoconus, although the definition of subclinical varied. We grouped subclinical keratoconus, forme fruste, and very asymmetrical eyes together. The tests showed good accuracy, with a summary sensitivity of 90.0% (95% CI 84.5% to 93.8%) and a summary specificity of 95.5% (95% CI 91.9% to 97.5%). However, the certainty of the evidence was very low for sensitivity and low for specificity.
In both groups, we graded most studies at high risk of bias, with high applicability concerns, in the domain of patient selection, since most were case‐control studies. Moreover, we graded the certainty of evidence as low to very low due to selection bias, inconsistency, and imprecision.
We could not explain the heterogeneity between the studies. The sensitivity analyses based on study design, AI algorithm, imaging technique (topography versus tomography), and data source (parameters versus images) showed no differences in the results.
Authors' conclusions
AI appears to be a promising triage tool in ophthalmologic practice for diagnosing keratoconus. Test accuracy was very high for manifest keratoconus and slightly lower for subclinical keratoconus, indicating a higher chance of missing a diagnosis in people without clinical signs. This could lead to progression of keratoconus or an erroneous indication for refractive surgery, which would worsen the disease.
We are unable to draw clear and reliable conclusions due to the high risk of bias, the unexplained heterogeneity of the results, and high applicability concerns, all of which reduced our confidence in the evidence.
Greater standardization in future research would increase the quality of studies and improve comparability between studies.</abstract><cop>Chichester, UK</cop><pub>John Wiley & Sons, Ltd</pub><pmid>37965960</pmid><doi>10.1002/14651858.CD014911.pub2</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 1465-1858 |
| ispartof | Cochrane database of systematic reviews, 2023-11, Vol.2023 (11), p.CD014911 |
| issn | 1465-1858 1469-493X 1465-1858 1469-493X |
| language | eng |
| recordid | cdi_pubmedcentral_primary_oai_pubmedcentral_nih_gov_10646985 |
| source | MEDLINE; EZB-FREE-00999 freely available EZB journals; Alma/SFX Local Collection; Cochrane Library |
| subjects | Artificial Intelligence Case-Control Studies Cornea Cross-Sectional Studies Eyes & vision Humans Keratoconus Keratoconus - diagnostic imaging Medicine General & Introductory Medical Sciences Outer eye Physical Examination |
| title | Artificial intelligence for detecting keratoconus |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-10T08%3A04%3A48IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_pubme&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Artificial%20intelligence%20for%20detecting%20keratoconus&rft.jtitle=Cochrane%20database%20of%20systematic%20reviews&rft.au=Dickman,%20Mor%20M&rft.date=2023-11-15&rft.volume=2023&rft.issue=11&rft.spage=CD014911&rft.pages=CD014911-&rft.issn=1465-1858&rft.eissn=1465-1858&rft_id=info:doi/10.1002/14651858.CD014911.pub2&rft_dat=%3Cproquest_pubme%3E2890363722%3C/proquest_pubme%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2890363722&rft_id=info:pmid/37965960&rfr_iscdi=true |