FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4376875196> ?p ?o ?g. }

• W4376875196 endingPage "100331" @default.
• W4376875196 startingPage "100331" @default.
• W4376875196 abstract "ObjectiveTo investigate the impact of corneal photograph quality on convolutional neural network (CNN) predictions.DesignA CNN trained to classify bacterial and fungal keratitis was evaluated using photographs of ulcers labeled according to 5 corneal image quality parameters: eccentric gaze direction, abnormal eyelid position, over/under-exposure, inadequate focus, and malpositioned light reflection.ParticipantsAll eligible subjects with culture and stain-proven bacterial and/or fungal ulcers presenting to Aravind Eye Hospital in Madurai, India, between January 1, 2021 and December 31, 2021.MethodsConvolutional neural network classification performance was compared for each quality parameter, and gradient class activation heatmaps were generated to visualize regions of highest influence on CNN predictions.Main Outcome MeasuresArea under the receiver operating characteristic and precision recall curves were calculated to quantify model performance. Bootstrapped confidence intervals were used for statistical comparisons. Logistic loss was calculated to measure individual prediction accuracy.ResultsIndividual presence of either light reflection or eyelids obscuring the corneal surface was associated with significantly higher CNN performance. No other quality parameter significantly influenced CNN performance. Qualitative review of gradient class activation heatmaps generally revealed the infiltrate as having the highest diagnostic relevance.ConclusionsThe CNN demonstrated expert-level performance regardless of image quality. Future studies may investigate use of smartphone cameras and image sets with greater variance in image quality to further explore the influence of these parameters on model performance.Financial Disclosure(s)Proprietary or commercial disclosure may be found after the references. To investigate the impact of corneal photograph quality on convolutional neural network (CNN) predictions. A CNN trained to classify bacterial and fungal keratitis was evaluated using photographs of ulcers labeled according to 5 corneal image quality parameters: eccentric gaze direction, abnormal eyelid position, over/under-exposure, inadequate focus, and malpositioned light reflection. All eligible subjects with culture and stain-proven bacterial and/or fungal ulcers presenting to Aravind Eye Hospital in Madurai, India, between January 1, 2021 and December 31, 2021. Convolutional neural network classification performance was compared for each quality parameter, and gradient class activation heatmaps were generated to visualize regions of highest influence on CNN predictions. Area under the receiver operating characteristic and precision recall curves were calculated to quantify model performance. Bootstrapped confidence intervals were used for statistical comparisons. Logistic loss was calculated to measure individual prediction accuracy. Individual presence of either light reflection or eyelids obscuring the corneal surface was associated with significantly higher CNN performance. No other quality parameter significantly influenced CNN performance. Qualitative review of gradient class activation heatmaps generally revealed the infiltrate as having the highest diagnostic relevance. The CNN demonstrated expert-level performance regardless of image quality. Future studies may investigate use of smartphone cameras and image sets with greater variance in image quality to further explore the influence of these parameters on model performance." @default.
• W4376875196 created "2023-05-18" @default.
• W4376875196 creator A5006652347 @default.
• W4376875196 creator A5023831605 @default.
• W4376875196 creator A5031651070 @default.
• W4376875196 creator A5033034428 @default.
• W4376875196 creator A5035227516 @default.
• W4376875196 creator A5037444025 @default.
• W4376875196 creator A5055729842 @default.
• W4376875196 creator A5065241414 @default.
• W4376875196 creator A5075952432 @default.
• W4376875196 creator A5087014948 @default.
• W4376875196 date "2023-12-01" @default.
• W4376875196 modified "2023-10-14" @default.
• W4376875196 title "Assessing the Impact of Image Quality on Deep Learning Classification of Infectious Keratitis" @default.
• W4376875196 cites W1976071803 @default.
• W4376875196 cites W1976480045 @default.
• W4376875196 cites W2048679677 @default.
• W4376875196 cites W2098947340 @default.
• W4376875196 cites W2469369919 @default.
• W4376875196 cites W2589074029 @default.
• W4376875196 cites W2752747624 @default.
• W4376875196 cites W2761604622 @default.
• W4376875196 cites W2800012666 @default.
• W4376875196 cites W2905636906 @default.
• W4376875196 cites W2962858109 @default.
• W4376875196 cites W2981369174 @default.
• W4376875196 cites W3035455206 @default.
• W4376875196 cites W3043058905 @default.
• W4376875196 cites W3082939062 @default.
• W4376875196 cites W3109971553 @default.
• W4376875196 cites W3161310076 @default.
• W4376875196 cites W3177778205 @default.
• W4376875196 cites W3193478640 @default.
• W4376875196 cites W3203861427 @default.
• W4376875196 cites W3210824773 @default.
• W4376875196 cites W4210638541 @default.
• W4376875196 cites W4226325157 @default.
• W4376875196 cites W4237275920 @default.
• W4376875196 cites W4281488668 @default.
• W4376875196 cites W4296098989 @default.
• W4376875196 doi "https://doi.org/10.1016/j.xops.2023.100331" @default.
• W4376875196 hasPublicationYear "2023" @default.
• W4376875196 type Work @default.
• W4376875196 citedByCount "0" @default.
• W4376875196 crossrefType "journal-article" @default.
• W4376875196 hasAuthorship W4376875196A5006652347 @default.
• W4376875196 hasAuthorship W4376875196A5023831605 @default.
• W4376875196 hasAuthorship W4376875196A5031651070 @default.
• W4376875196 hasAuthorship W4376875196A5033034428 @default.
• W4376875196 hasAuthorship W4376875196A5035227516 @default.
• W4376875196 hasAuthorship W4376875196A5037444025 @default.
• W4376875196 hasAuthorship W4376875196A5055729842 @default.
• W4376875196 hasAuthorship W4376875196A5065241414 @default.
• W4376875196 hasAuthorship W4376875196A5075952432 @default.
• W4376875196 hasAuthorship W4376875196A5087014948 @default.
• W4376875196 hasBestOaLocation W43768751961 @default.
• W4376875196 hasConcept C108583219 @default.
• W4376875196 hasConcept C115961682 @default.
• W4376875196 hasConcept C119857082 @default.
• W4376875196 hasConcept C153180895 @default.
• W4376875196 hasConcept C154945302 @default.
• W4376875196 hasConcept C41008148 @default.
• W4376875196 hasConcept C55020928 @default.
• W4376875196 hasConcept C71924100 @default.
• W4376875196 hasConcept C81363708 @default.
• W4376875196 hasConceptScore W4376875196C108583219 @default.
• W4376875196 hasConceptScore W4376875196C115961682 @default.
• W4376875196 hasConceptScore W4376875196C119857082 @default.
• W4376875196 hasConceptScore W4376875196C153180895 @default.
• W4376875196 hasConceptScore W4376875196C154945302 @default.
• W4376875196 hasConceptScore W4376875196C41008148 @default.
• W4376875196 hasConceptScore W4376875196C55020928 @default.
• W4376875196 hasConceptScore W4376875196C71924100 @default.
• W4376875196 hasConceptScore W4376875196C81363708 @default.
• W4376875196 hasFunder F4320306080 @default.
• W4376875196 hasFunder F4320306811 @default.
• W4376875196 hasFunder F4320306915 @default.
• W4376875196 hasFunder F4320309233 @default.
• W4376875196 hasFunder F4320332161 @default.
• W4376875196 hasFunder F4320337350 @default.
• W4376875196 hasIssue "4" @default.
• W4376875196 hasLocation W43768751961 @default.
• W4376875196 hasOpenAccess W4376875196 @default.
• W4376875196 hasPrimaryLocation W43768751961 @default.
• W4376875196 hasRelatedWork W2731899572 @default.
• W4376875196 hasRelatedWork W2999805992 @default.
• W4376875196 hasRelatedWork W3116150086 @default.
• W4376875196 hasRelatedWork W3133861977 @default.
• W4376875196 hasRelatedWork W4200173597 @default.
• W4376875196 hasRelatedWork W4223943233 @default.
• W4376875196 hasRelatedWork W4291897433 @default.
• W4376875196 hasRelatedWork W4312417841 @default.
• W4376875196 hasRelatedWork W4321369474 @default.
• W4376875196 hasRelatedWork W4380075502 @default.
• W4376875196 hasVolume "3" @default.
• W4376875196 isParatext "false" @default.
• W4376875196 isRetracted "false" @default.

• next