Optimising Optometry Education Through AI

Biswajit Kashyap, M.Optom

Student, Sri Sankaradeva Nethralaya, Guwahati, India

 

Artificial Intelligence (AI), particularly deep learning (DL), has revolutionised numerous sectors, including healthcare, by improving efficiency and driving innovation. In the field of optometry, DL models such as Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs) present significant opportunities to enhance both clinical practice and educational frameworks. However, the effective adoption of these advanced technologies requires a solid understanding of foundational AI concepts. Developing this foundational knowledge equips students as well as professionals with the necessary tools to integrate AI into diagnostic processes and optimise learning outcomes.⁽¹⁾

What are CNN and GAN?

Both of them are models of Deep learning, a subset of AI that involves machine learning inspired by the structure and function of the brain called artificial neural networks.

Convolutional Neural Networks (CNN): A Convolutional Neural Network is a specialised type of deep learning algorithm mainly designed for tasks that necessitate object recognition, including image classification, detection, and segmentation.⁽¹⁾

Generative Adversarial Networks (GANs): Generative Adversarial Networks use two neural networks, and together, they produce synthetic instances of original data. ⁽¹⁾

How can they be useful in optometry?

Implementation of these DL models can open a new path of learning and education in Optometry:

• GANs can be used to create a diverse set of retinal images that include rare or advanced disease stages, which students might not frequently encounter in clinical settings, thus exposing them to a wide spectrum of cases and preparing them for real-world clinical challenges.⁽²⁾
• Furthermore, GANs can be implemented to create patient-specific visual simulations. These simulations can be used to educate both students and also the general public about the impact of various ocular diseases on vision. For example, a GAN could generate a series of images that replicate how a patient with macular degeneration might perceive their environment.⁽²⁾
• CNNs are highly effective in image recognition and analysis, making them ideal for optometry, a field heavily reliant on imaging technologies like retinal photography and Optical Coherence Tomography (OCT). By incorporating CNNs into teaching modules, educators can provide students with automated, real-time analysis of retinal images. This not only aids in the early detection and diagnosis of conditions such as diabetic retinopathy and glaucoma but also allows students to learn by interacting with advanced diagnostic tools. Students can compare their manual assessments with those generated by CNNs, enabling them to refine their diagnostic skills.⁽³⁾
• Moreover, CNNs (specifically 2D CNN models) can be utilised to create interactive learning platforms where students can explore different disease manifestations through segmented and annotated images further deepening their understanding of ocular anatomy and pathology, making theoretical concepts more tangible.⁽³⁾

Conclusion

The integration of deep learning technologies such as CNNs and GANs in optometric education offers significant potential for both enhancing student learning and increasing public awareness. By integrating these advanced tools, educators can provide more dynamic, interactive, and comprehensive training, preparing future optometrists for the complexities of modern clinical practice.⁽⁴⁾

Simultaneously, these technologies can serve as powerful instruments not only for students but also for public health education.

 

References

1. Tom, T. (2019). Artificial Intelligence Basics: A Non-Technical Introduction. Monrovia, CA, USA: Appres.
2. Kugelman, J., Alonso-Caneiro, D., Read, S. A., & Collins, M. J. (2021). A review of generative adversarial network applications in optical coherence tomography image analysis. Eye and Vision, 8(1), 1-12.
3. Rasel, R. K., Wu, F., Chiariglione, M., Choi, S. S., Doble, N., & Gao, X. R. (2024). Assessing the efficacy of 2D and 3D CNN algorithms in OCT-based glaucoma detection. Scientific Reports, 14(1), 11758.
4. Scanzera, A. C., Shorter, E., Kinnaird, C., Valikodath, N., Al-Khaled, T., Cole, E., Kravets, S., Hallak, J. A., & McMahon, T. (2022). Optometrist’s perspectives of artificial intelligence in eye care. R.V. Paul Chan, 15(2), 215-227.