Research on Eye Disease Recognition Algorithm based on Deep Learning
With the popularization of digital devices, the myopia rate among teenagers has surged due to excessive eye use. Among them, rational myopia, characterized by abnormal axial growth, has become an important cause of blindness in China, but the public awareness is insufficient. Traditional diagnosis relies on genetic testing, mydriatic refraction, and OCT imaging, which have pain points such as insufficient primary medical resources, low diagnostic efficiency, and high misdiagnosis rate. This study is based on deep learning technology to construct an intelligent diagnostic system, designed with a dual-mode architecture: using MobileNetV3 lightweight network to achieve rapid pathological myopia screening of fundus color photos, and combining YOLOV5 algorithm to accurately locate the lesion area of positive cases. The system is developed under the PyTorch framework and has been clinically validated to have two major advantages: firstly, it improves classification accuracy on small sample data through transfer learning, and secondly, it achieves millisecond level image processing speed, with lesion detection accuracy reaching the level of professional physicians. This solution can effectively alleviate the problem of uneven distribution of medical resources, assist grassroots doctors in reducing the risk of missed diagnosis, and is suitable for large-scale screening and remote medical scenarios. The research provides an intelligent solution for early screening and diagnosis of blinding eye diseases, which has significant clinical application value and social benefits. In the future, multi center data fusion can be used to further optimize the model generalization ability and assist in the construction of an eye health management system.
References
- KermanyDS,GoldbaumM,CaiW,etal.Identifyingmedicaldiagnosesandtreatablediseasesbyimagebaseddeeplearning[J].Cell,2018,172(5):1122-1131.
- GargeyaR,LengT.Automatedidentificationofdiabeticretinopathyusingdeeplearning[J].Ophthalmology,2017,124(7):962-969.
- OgurtsovaK,GuariguataL,BarengoNC,etal.IDFdiabetesAtlas:Globalestimatesofundiagnoseddiabetesinadultsfor2021.DiabetesResClinPract.2022Jan;183:109-118.
- Gargeya R, Leng T. Automated Identification of Diabetic Retinopathy Using Deep Learning. Ophthalmology. 2017 Jul;124(7):962-969. doi: 10.1016/j.ophtha.2017.02.008. Epub 2017 Mar 27. PMID: 28359545.
- Elif Setenay Cutur,and Neslihan Gokmen Inan."Multi-class Classification of Retinal Eye Diseases from Ophthalmoscopy Images Using Transfer Learning-Based Vision Transformers."Journal of Imaging Informatics in Medicine prepublish(2025):1-15.
- G Divya Deepak,and Subraya Krishna Bhat."Deep learning-based CNN for multiclassification of ocular diseases using transfer learning."Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 12.1(2024):
- A. Anitha Rani,C. Karthikeyini,and C. Reshma Ravi."Eye Disease Prediction Using Deep Learning and Attention on Oct Scans."SN Computer Science 5.8(2024):1065-1065.
- Shayla Sharmin, et al."A dataset of color fundus images for the detection and classification of eye diseases."Data in Brief 57.(2024):110979-110979.
- Anjar Wanto, Yuhandri Yuhandri,andOkfalisa Okfalisa."RetMobileNet: A New Deep Learning Approach for Multi-Class Eye Disease Identification."Revue d'Intelligence Artificielle 38.4(2024):
- Uyar K ,Yurdakul M ,Taşdemir Ş .Abc-based weighted voting deep ensemble learning model for multiple eye disease detection[J].Biomedical Signal Processing and Control,2024,96(PB):106617-106617.