AI Based System for Remote Eye Testing and Prescription Determination

Abstract

This research explores the development of an AI model capable of detecting the refractive index (RI) of a person\'s eyes using a smartphone camera. Traditional methods for measuring refractive errors require specialized ophthalmic equipment, which may not be accessible to all. The proposed method leverages deep learning and computer vision techniques to analyse eye images and predict the refractive index, providing a cost-effective and widely accessible alternative. Additionally, this paper discusses the advantages, challenges, and potential future improvements of the proposed approach.

Introduction

1. Introduction

Vision plays a crucial role in human life, yet over 2.2 billion people worldwide suffer from visual impairment or blindness (WHO), with 1 billion cases preventable through timely diagnosis and affordable care. The most common and correctable causes are refractive errors—myopia, hyperopia, and astigmatism—resulting from improper light focusing on the retina, leading to blurred vision.
Uncorrected vision issues negatively affect education, productivity, and overall quality of life, particularly in rural and low-resource regions lacking ophthalmic facilities.

2. Limitations of Traditional Vision Testing

Conventional eye exams depend on optical clinics and costly instruments like autorefractors and retinoscopes. These methods, while accurate, are expensive, time-consuming, and unsuitable for mass or remote screenings, restricting accessibility in developing areas.

3. Motivation and Problem Statement

The increasing global burden of vision disorders and the need for home-based, low-cost testing drive the research.
Key problems:

• Lack of accessibility to optometrists

• High cost of eye exams

• Manual, subjective testing

• Limited scalability and personalization

4. Proposed Solution

The study proposes an AI-powered mobile application that enables automated vision testing and digital spectacle prescription generation using only a smartphone camera.
It leverages computer vision, photorefraction techniques, and a lightweight convolutional neural network (CNN) to:

• Analyze eye images

• Detect refractive errors (myopia, hyperopia, emmetropia)

• Predict prescription parameters (SPH, CYL, Axis)

• Provide instant digital prescriptions in PDF or email formats

This approach removes the need for expensive equipment, making vision care affordable, scalable, and accessible for remote communities, schools, and home users.

5. Literature Review

Prior AI studies in ophthalmology show success in detecting diabetic retinopathy, glaucoma, and retinal diseases using CNNs:

• Gulshan et al. (2016): AI matched ophthalmologists in retinal diagnosis.

• Ting et al. (2017): AI-based glaucoma detection from fundus images.

• Chaurasia et al. (2023): Lightweight CNNs feasible for mobile health apps.

• Zhang et al. (2022): Smartphone optics for refractive index detection.

Existing tools (Essilor Eye-Ruler, Zeiss App, EyeQue) lack full AI integration or require extra hardware. The proposed system bridges this gap by delivering AI-driven, prescription-grade accuracy on smartphones.

6. System Architecture

The proposed system follows a multi-layered architecture:

1. User Interface Layer: Interactive app for test selection and result viewing.

2. Image Processing Module: Uses OpenCV for grayscale conversion, contrast enhancement, region-of-interest detection, and edge detection to extract pupil features.

3. AI-Based Analysis: CNN model interprets visual patterns (crescent shapes, light reflections) to detect refractive errors and compute prescription values.

4. Cloud Database: Stores user data securely for historical analysis.

5. AR/VR Integration: Adds immersive, interactive eye testing options.

7. Methodology

The research follows four phases:

1. Requirement Analysis: Input from optometrists and users.

2. Design: Developing accessible UI and model architecture.

3. Implementation: Integrating AI algorithms with image processing.

4. Testing & Validation: Comparing AI predictions with clinical exams and real-world trials.

The CNN model uses residual learning (ResNet) and Grad-CAM heatmaps to visualize decision-making. Training data consists of simulated and real photorefraction images, showing distinctive light crescents for different refractive errors.

8. Expected Outcomes

• Accurate and automated spectacle prescription generation

• Improved accessibility and affordability for vision testing

• Reduced dependence on manual eye exams

• Scalability for teleophthalmology and remote healthcare

Conclusion

This paper introduces an innovative mobile application for spectacles prescription determination using AI and computer vision. The system enhances accessibility, reduces costs, and enables early detection of vision issues. Future work will focus on refining AI models, addressing challenges in edge cases, and expanding global usability by incorporating multi-language support and regulatory compliance [8].The proposed AI-based mobile application enables users to conduct remote eye tests and obtain prescriptions accurately. The system integrates computer vision, deep learning, and real-time image processing to enhance accessibility and cost-effectiveness in vision care [6], [9], [14].

References

[1] K. Lee et al., \"Advancements in AI-based Optical Diagnostics,\" IEEE Transactions on Biomedical Engineering, vol. 66, no. 3, pp. 567-579, 2021. [2] A. Gomez, \"Computer Vision for Healthcare Applications,\" Journal of Artificial Intelligence in Medicine, vol. 32, no. 1, pp. 89-102, 2022. [3] B. Williams et al., \"Validation of AI-based Eye Testing Models,\" International Journal of Ophthalmology, vol. 45, no. 6, pp. 300-312, 2020. [4] C. Johnson, \"Regulatory Challenges in AI-powered Healthcare Applications,\" Journal of Medical Law and Ethics, vol. 29, no. 4, pp. 220-234, 2023. [5] World Health Organization, \"Global Report on Vision,\" WHO, 2019. [6] J. Smith et al., \"AI in Optometry: Trends and Applications,\" IEEE Transactions on Medical Imaging, vol. 39, no. 4, pp. 1234-1245, 2022. [7] P. Brown, \"Smartphone-Based Vision Tests: A Comparative Study,\" Journal of Telemedicine and eHealth, vol. 28, no. 3, pp. 211-223, 2021. [8] R. Patel et al., \"Machine Learning in Ophthalmology: A Review,\" IEEE Transactions on Biomedical Engineering, vol. 67, no. 5, pp. 1405-1418, 2023. [9] M. R. Hossain et al., “AI-based system for eye diagnosis using convolutional neural networks,” Journal of Imaging, vol. 9, no. 4, p. 84, Apr. 2023. [Online]. Available: https://www.mdpi.com/2313-433X/9/4/84 [10] Gulshan V., et al. \"Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs.\" JAMA, 2016. [11] Ting D. S. W., et al. \"Artificial Intelligence and Deep Learning in Ophthalmology.\" British Journal of Ophthalmology, 2017. [12] He K., et al. \"Deep Residual Learning for Image Recognition.\" IEEE Conference on Computer Vision and Pattern Recognition, 2016. [13] Kingma D., Ba J. \"Adam: A Method for Stochastic Optimization.\" arXiv preprint arXiv:1412.6980, 2014. [14] Chaurasia, A., et al. \"Mobile Health Monitoring with Lightweight CNNs.\" Elsevier, 2023. [15] Zhang, M., et al. \"Refractive Index Sensing Using Smartphone Optics.\" Sensors Journal, 2022. [16] [Image] \"AI-Based Eye Disease Model Diagram.\" Jordi Corbilla. [Online image]. Available: https://jordicorbilla.github.io/ocular-disease-intelligent-recognition-deep-learning/images/deeplearningdesign.png [17] [Image] \"Retinal Imaging Flowchart.\" MDPI. [Online image]. Available: https://pub.mdpi-res.com/jimaging/jimaging-09-00084/article_deploy/html/images/jimaging-09-00084-g001.png [18] [Image] \"CNN Eye Structure Diagram.\" Elsevier. [Online image]. Available: https://ars.els-cdn.com/content/image/1-s2.0-S1566253523003755-gr2.jpg [19] [Image] \"Feature Extraction in Eye Imaging.\" Springer Nature. [Online image]. Available: https://media.springernature.com/lw685/springer-static/image/art%3A10.1038%2Fs41598-023-34212-w/MediaObjects/41598_2023_34212_Fig2_HTML.png

Copyright

Copyright © 2025 Shrey Gupta, Shraddha Sharma, Shreya Chaturvedi, Shreyansh Sharma, Soumya Jain. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.