{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,10]],"date-time":"2026-02-10T13:57:35Z","timestamp":1770731855520,"version":"3.49.0"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T00:00:00Z","timestamp":1753833600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Sci."],"abstract":"<jats:sec><jats:title>Introduction<\/jats:title><jats:p>Uncorrected refractive errors are a leading cause of preventable vision impairment globally, particularly affecting individuals in low-resource regions where timely diagnosis and screening access remain significant challenges despite the availability of economical treatments.<\/jats:p><\/jats:sec><jats:sec><jats:title>Aim<\/jats:title><jats:p>This study introduces a novel deep learning-based system for automated refractive error classification using photorefractive images acquired via a standard smartphone camera.<\/jats:p><\/jats:sec><jats:sec><jats:title>Methods<\/jats:title><jats:p>A multi-branch convolutional neural network (CNN) was developed and trained on a dataset of 2,139 corneal images collected from an Indonesian public eye hospital. The model was designed to classify refractive errors into four categories: significant myopia, significant hypermetropia, insignificant refractive error, and not applicable to classified. Grad-CAM visualization was employed to provide insights into the model\u2019s interpretability.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>The 3-branch CNN architecture demonstrated superior performance, achieving an overall test accuracy of 91%, precision of 96%, and recall of 98%, with an area under the curve (AUC) score of 0.9896. Its multi-scale feature extraction pathways were pivotal in effectively addressing overlapping red reflex patterns and subtle variations between classes.<\/jats:p><\/jats:sec><jats:sec><jats:title>Conclusion<\/jats:title><jats:p>This study establishes the feasibility of smartphone-based photorefractive assessment integrated with artificial intelligence for scalable and cost-effective vision screening. By training the CNN model with a real-world dataset representative of Southeast Asian populations, this system offers a reliable solution for early refractive error detection with significant implications for improving accessibility to eye care services in resource-limited settings.<\/jats:p><\/jats:sec>","DOI":"10.3389\/fcomp.2025.1576958","type":"journal-article","created":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T05:31:03Z","timestamp":1753853463000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Deep learning for vision screening in resource-limited settings: development of multi-branch CNN for refractive error detection based on smartphone image"],"prefix":"10.3389","volume":"7","author":[{"given":"Muhammad","family":"Syauqie","sequence":"first","affiliation":[]},{"given":"Harry","family":"Patria","sequence":"additional","affiliation":[]},{"given":"Sutanto Priyo","family":"Hastono","sequence":"additional","affiliation":[]},{"given":"Kemal Nazaruddin","family":"Siregar","sequence":"additional","affiliation":[]},{"given":"Nila Djuwita Farieda","family":"Moeloek","sequence":"additional","affiliation":[]}],"member":"1965","published-online":{"date-parts":[[2025,7,30]]},"reference":[{"key":"ref1","first-page":"265","article-title":"Tensor flow: a system for large-scale machine learning","author":"Abadi","year":"2016"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"1145","DOI":"10.1016\/S0031-3203(96)00142-2","article-title":"Use of the area under the ROC curve in the evaluation of machine learning algorithms","volume":"30","author":"Bradley","year":"1997","journal-title":"Pattern Recogn."},{"key":"ref3","doi-asserted-by":"publisher","first-page":"379","DOI":"10.1097\/ICU.0b013e328244dfed","article-title":"Diagnosis and treatment of refractive errors in the pediatric population","volume":"18","author":"Braverman","year":"2007","journal-title":"Curr. Opin. Ophthalmol."},{"key":"ref4","doi-asserted-by":"publisher","first-page":"e0268800","DOI":"10.1371\/journal.pone.0268800","article-title":"The prevalence and causes of pediatric uncorrected refractive error: pooled data from population studies for global burden of disease (GBD) sub-regions","volume":"17","author":"Cao","year":"2022","journal-title":"PLoS One"},{"key":"ref5","volume-title":"Deep learning with Python","author":"Chollet","year":"2017"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"e16225","DOI":"10.2196\/16225","article-title":"A deep-learning-based prediction of refractive error using smartphone-captured photorefraction images: model development and validation study","volume":"8","author":"Chun","year":"2019","journal-title":"JMIR Med. Inform."},{"key":"ref7","doi-asserted-by":"publisher","first-page":"c1711","DOI":"10.1136\/bmj.c1711","article-title":"The management of refractive errors","volume":"340","author":"Cochrane","year":"2010","journal-title":"BMJ"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"103","DOI":"10.1119\/1.4905810","article-title":"Photorefraction of the eye","volume":"53","author":"Colicchia","year":"2015","journal-title":"Phys. Teach."},{"key":"ref9","doi-asserted-by":"publisher","first-page":"347","DOI":"10.1111\/j.1442-9071.1996.tb01606.x","article-title":"Photoscreener instrument evaluation in a childhood population. 1. The Otago photoscreener and Dortmans videophotorefractor","volume":"24","author":"Cooper","year":"1996","journal-title":"Aust. N. Z. J. Ophthalmol."},{"key":"ref10","first-page":"2819 \u2013 A0149","article-title":"Using a smartphone camera to detect refractive error among schoolchildren: a simple model to address cost-related barriers for refractive error detection","volume":"63","author":"Do","year":"2022","journal-title":"Invest. Ophthalmol. Vis. Sci."},{"key":"ref11","doi-asserted-by":"crossref","first-page":"61","DOI":"10.1007\/978-3-540-68632-3_6","article-title":"Screening for pediatric ophthalmologic disorders","volume-title":"Pediatric ophthalmology: Current thought and a practical guide","author":"Donahue","year":"2009"},{"key":"ref12","first-page":"2775","article-title":"Exploiting active learning in novel refractive error detection with smartphones","author":"Fu","year":"2020"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"e144","DOI":"10.1016\/S2214-109X(20)30489-7","article-title":"Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: the right to sight: an analysis for the global burden of disease study","volume":"9","year":"2021","journal-title":"Lancet Glob. Health"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"1737354","DOI":"10.1080\/2331205X.2020.1737354","article-title":"Prevalence and associated factors of uncorrected refractive errors among school children in suburban areas in Bandung, Indonesia","volume":"7","author":"Halim","year":"2020","journal-title":"Power. Med."},{"key":"ref15","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1016\/j.joco.2017.08.009","article-title":"Global and regional estimates of refractive error prevalence: systematic review and meta-analysis","volume":"30","author":"Hashemi","year":"2018","journal-title":"J. Curr. Ophthalmol."},{"key":"ref16","first-page":"305","volume-title":"Deep learning. Genetic Programming and Evolvable Machines","author":"Heaton","year":"2018"},{"key":"ref17","first-page":"37","article-title":"Uncorrected refractive error: the major and most easily avoidable cause of vision loss","volume":"20","author":"Holden","year":"2007","journal-title":"Community Eye Health"},{"key":"ref18","first-page":"24","article-title":"Mobile cataract screening app using a smartphone","author":"Ik","year":"2015"},{"key":"ref19","first-page":"448","article-title":"Batch normalization: accelerating deep network training by reducing internal covariate shift","author":"Ioffe","year":"2015"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"37","DOI":"10.4274\/tjo.galenos.2020.78989","article-title":"Artificial intelligence and ophthalmology","volume":"50","author":"Keskinbora","year":"2020","journal-title":"Turk. J. Ophthalmol."},{"key":"ref21","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014"},{"key":"ref22","first-page":"608","article-title":"Enhancing myopia screening in the community: exploring the potential of smartphone photorefraction","volume":"65","author":"Kiu","year":"2024","journal-title":"Invest. Ophthalmol. Vis. Sci."},{"key":"ref23","first-page":"1","article-title":"Anomaly detection in red reflex images using deep learning approaches","author":"Kriangsakdachai","year":"2022"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"1097","DOI":"10.1145\/3065386","article-title":"Imagenet classification with deep convolutional neural networks","volume":"25","author":"Krizhevsky","year":"2012","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref25","first-page":"1","article-title":"Medical imaging using machine learning and deep learning algorithms: a review","author":"Latif","year":"2019"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"LeCun","year":"2015","journal-title":"Nature"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"174","DOI":"10.1097\/ICU.0000000000000846","article-title":"Novel technical and privacy-preserving technology for artificial intelligence in ophthalmology","volume":"33","author":"Lim","year":"2022","journal-title":"Curr. Opin. Ophthalmol."},{"key":"ref28","doi-asserted-by":"publisher","first-page":"2810","DOI":"10.3390\/diagnostics13172810","article-title":"Automatic refractive error estimation using deep learning-based analysis of red reflex images","volume":"13","author":"Linde","year":"2023","journal-title":"Diagnostics (Basel)"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"e18226","DOI":"10.2196\/18226","article-title":"A one-step and streamlined children vision screening solution based on smartphone imaging for resource-limited areas: design and evaluation","volume":"8","author":"Ma","year":"2020","journal-title":"JMIR Mhealth Uhealth"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"1771","DOI":"10.18240\/ijo.2017.11.21","article-title":"Prevalence of uncorrected refractive error in urban, suburban, exurban and rural primary school children in Indonesian population","volume":"10","author":"Mahayana","year":"2017","journal-title":"Int. J. Ophthalmol."},{"key":"ref31","doi-asserted-by":"publisher","first-page":"S82","DOI":"10.1016\/j.optom.2022.06.005","article-title":"Application of artificial intelligence to ophthalmology and optometry: a citation network analysis","volume":"15","author":"Martinez-Perez","year":"2022","journal-title":"J. Optom."},{"key":"ref32","doi-asserted-by":"publisher","first-page":"2357","DOI":"10.1042\/CS20210207","article-title":"Updates in deep learning research in ophthalmology","volume":"135","author":"Ng","year":"2021","journal-title":"Clin. Sci. (Lond.)"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"2590","DOI":"10.1167\/iovs.13-11725","article-title":"Prevalence of refractive errors in a multiethnic Asian population: the Singapore epidemiology of eye disease study","volume":"54","author":"Pan","year":"2013","journal-title":"Invest. Ophthalmol. Vis. Sci."},{"key":"ref34","first-page":"195","volume-title":"Application of artificial intelligence in medical imaging","author":"Panda","year":"2022"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"2229","DOI":"10.48550\/arXiv.2010.16061","article-title":"Evaluation: from precision, recall, and F-measure to ROC, informedness, markedness, and correlation","volume":"2","author":"Powers","year":"2011","journal-title":"J. Mach. Learn. Technol."},{"key":"ref36","doi-asserted-by":"publisher","first-page":"55","DOI":"10.1007\/3-540-49430-8_3","article-title":"Early stopping\u2014but when?","volume":"1524","author":"Prechelt","year":"2000","journal-title":"Lect. Notes Comput. Sci"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"g5991","DOI":"10.1136\/bmj.g5991","article-title":"Uncorrected refractive error and education","volume":"349","author":"Rahi","year":"2014","journal-title":"BMJ"},{"key":"ref38","doi-asserted-by":"publisher","first-page":"63","DOI":"10.2471\/BLT.07.041210","article-title":"Global magnitude of visual impairment caused by uncorrected refractive errors in 2004","volume":"86","author":"Resnikoff","year":"2008","journal-title":"Bull. World Health Organ."},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-019-01228-7","article-title":"Grad-CAM: visual explanations from deep networks via gradient-based localization","volume":"618","author":"Rs","year":"2020","journal-title":"Int. J. Comput. Vis."},{"key":"ref40","doi-asserted-by":"publisher","first-page":"60","DOI":"10.1186\/s40537-019-0197-0","article-title":"A survey on image data augmentation for deep learning","volume":"6","author":"Shorten","year":"2019","journal-title":"J. Big Data"},{"key":"ref41","author":"Simonyan","year":"2014"},{"key":"ref42","first-page":"1519","article-title":"Revealing the inherent properties of pupillary red reflex in refractive errors by smartphone imaging: a study in Indonesia\u2019s population","volume-title":"Optics Continuum","author":"Syauqie","year":"2025"},{"key":"ref43","doi-asserted-by":"publisher","first-page":"2861","DOI":"10.1167\/iovs.18-23887","article-title":"Deep learning for predicting refractive error from retinal fundus images","volume":"59","author":"Varadarajan","year":"2018","journal-title":"Invest. Ophthalmol. Vis. Sci."},{"key":"ref44","doi-asserted-by":"publisher","first-page":"e55270","DOI":"10.2196\/55270","article-title":"Using smartphones to enhance vision screening in rural areas: pilot study","volume":"8","author":"Wang","year":"2024","journal-title":"JMIR Form. Res."},{"key":"ref45","doi-asserted-by":"crossref","first-page":"363","DOI":"10.1038\/eye.1999.92","article-title":"Photoscreening for refractive errors in children and young adults with severe learning disabilities using the MTI photoscreener","volume":"13","author":"Watts","year":"1999","journal-title":"Eye (Lond.)"},{"key":"ref46","doi-asserted-by":"crossref","DOI":"10.1145\/3428690.3429175","article-title":"Screening for refractive error with low-quality smartphone images","author":"Yang","year":"2020"}],"container-title":["Frontiers in Computer Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1576958\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T05:31:05Z","timestamp":1753853465000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1576958\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,7,30]]},"references-count":46,"alternative-id":["10.3389\/fcomp.2025.1576958"],"URL":"https:\/\/doi.org\/10.3389\/fcomp.2025.1576958","relation":{},"ISSN":["2624-9898"],"issn-type":[{"value":"2624-9898","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,7,30]]},"article-number":"1576958"}}