“Innovation is more often than not driven by unmet need”

Optos chief technology officer, Derek Swan, outlined the potential of artificial intelligence technology for the analysis of retinal images at 100% Optical (25–27 February).

Outlining the history behind ophthalmic imaging, Swan shared that the first Helmholtz ophthalmoscope was developed in 1851, while retinal photography was first published in 1886. By 1907, retinal images were routinely being captured.

Swan shared that the founder of Optos became involved in developing imaging technology after his son lost vision in an eye due to an undiagnosed retinal detachment.

“Innovation is more often than not driven by unmet need,” Swan observed.

He shared that fundus imaging remained largely unchanged over the course of the 20^th century, until the creation of the Optomap and ultra-wide field (UWF) imaging. Swan added that UWF had a long period of development.

“This has been a journey rather than an overnight success,” he said

In 1996, the first commercially available optical coherence tomography (OCT) device was launched. By 2004, around 10 million OCT scans were performed each year.

Swan highlighted that the catalyst behind the adoption of OCT has been its ability to both quantify and analyse the presence and extent of disease.

Developments in OCT have aimed to develop higher speed devices, with a greater depth and field of view, and enhanced diagnostics.

Swan described artificial intelligence (AI) as computer systems that are capable of performing tasks that usually require human intelligence to perform – for example, decision making and object detection.

Improvements in treatment, an ageing population and advances in medical devices mean that data around eye health is proliferating, Swan shared.

“The key to this lies around the data and how we change the data into something that helps us make decisions,” he said.

Swan observed that there is a sense among some people that AI is “taking over” or making people redundant.

“AI doesn’t replace the person who makes the decision,” he said.

To manage the potential risks associated with AI, proper data management and governance is important, Swan stressed, alongside an awareness of the bias that can be embedded in systems.

Swan highlighted that an online poll of 150 UK optometrists found that 54% described themselves as “not very knowledgeable” about AI, while 20% said they were “not at all knowledgeable” on the topic.

However, most of those surveyed expected they would use the technology in the near future.

“We have to bridge the gap between the understanding and knowledge base, and the appetite for adoption,” Swan emphasised.