Varun Ranganathan, MCOptom

Clinical Optometrist

An OCULAR Interface Exclusive

Keywords: Computer Vision, Vision Sciences, Visual Pathway, Visual Processing, Imaging

Introduction

Computer vision (CV) is a field of computer science which deals with image recognition and processing of objects and images of people. It is used to replicate how humans see and how our brain analyses information. Being a branch of Artificial Intelligence (AI), it uses Machine Learning (ML) and Deep Learning algorithms to achieve this ¹. Large datasets are used to train CV applications to recognise patterns in an image and gather valuable information.

How The Visual Processing System Works

When light enters our eyes and falls on the retina, the process of vision begins. The cornea and lens combine to produce a sharp and reversed image on the photoreceptors. The information is then transmitted to the visual cortex in our brain through the optic nerve where the image is reversed again. This happens very quickly and seamlessly, and it takes the brain only around 13 milliseconds to do that ².

Although the visual processing systems in humans are not fully understood, studies using monkeys have identified three further separate systems; one is for shape, second for colour and the third about movement ³. All these systems primarily depend on the intensity of light. Since one of the foundational aspects of CV is image processing, we can better understand how visual processing works.

CV Uses in Brain Imaging

fMRI measures slight changes in blood flow which happens during brain activity ⁴. It is used to understand how the brain’s critical functions work and how certain conditions like stroke and traumatic brain injury affect the brain. Large datasets are necessary to train algorithms and they can assist in the field of neuro-ophthalmology to help in research, can detect lesions in the brain and can also map the visual pathways for deeper understanding.

Assistive Technology

CV applications for the visually impaired can improve accessibility in their daily lives. Its uses are not only limited to text-to-speech functions and navigation aids but can also help them to use the Internet and social media ⁵. Navigating the Internet can be especially hard due to large-scale use of un-captioned images.

Developing Visual Prosthetics

Visual prosthetics are different from prosthetic shells as they improve a person’s functional vision ⁶. This field is evolving to great lengths thanks to the advent of Artificial Intelligence. CV being a subsect of AI is being incorporated into bionic eyes and retinal implants by using visual processing algorithms.

Medical Imaging

Image segmentation is commonly used in medical imaging and is one of the key aspects of a CV. In image segmentation, an image is split into multiple segments which make it easier to analyse. Deep Learning, a sub-field of AI is incorporated into Optical Coherence Tomography and fundus images making retinal segmentation easier to identify pathologies ⁷.

Conclusion

The integration of computer vision with vision sciences has many avenues waiting to be explored. It can create a synergistic relationship with vision sciences and as it continues to evolve, it holds a promising future and can deepen our understanding of visual pathways which can unravel deeper mysteries of the brain.

References

1. https://azure.microsoft.com/en-gb/resources/what-is-computer-vision#objectclassification.
2. In the blink of an eye | MIT News | Massachusetts Institute of Technology.
3. Goodale, M.A. (2013), Separate visual systems for perception and action: a framework for understanding cortical visual impairment. Dev Med Child Neurol, 55: 9-12. https://doi.org/10.1111/dmcn.12299.
4. Marc M. Himmelberg, Justin L. Gardner, Jonathan Winawer (2022), What has vision scienc taught us about functional MRI? NeuroImage, Volume 261-119536, ISSN 1053-8119, https://doi.org/10.1016/j.neuroimage.2022.119536.
5. Valipoor, M.M., de Antonio, A. Recent trends in computer vision-driven scene understanding for VI/blind users: a systematic mapping. Univ Access Inf Soc 22, 983–1005 (2023). https://doi.org/10.1007/s10209-022-00868-w.
6. Wang J, Zhu H, Liu J, Li H, Han Y, Zhou R, Zhang Y. The application of computer vision to visual prosthesis. Artif Organs. 2021 Oct;45(10):1141-1154. doi: 10.1111/aor.14022. Epub 2021 Jul 27. PMID: 34318520.
7. Ignacio A. Viedma, David Alonso-Caneiro, Scott A. Read, Michael J. Collins. Deep learning in retinal optical coherence tomography (OCT) (2022): A comprehensive survey. Neurocomputing, Volume 507, Pages 247-264, ISSN 0925-2312, https://doi.org/10.1016/j.neucom.2022.08.021.