Thoughts about artificial intelligence (AI) may conjure images of sentient robots rising up and bringing about the end of humanity, à la the Terminator films. The reality, in optics at least, is mercifully not as dramatic and largely focuses on the collection of data to inform an improved lens design.
Shamir says AI is already vital in lens design because the process would be too demanding without tools to process the amount of data and design parameters collected by the company.
The company is collecting more and more data on how patients use their lenses to create solutions and offer the best vision.
‘Shamir uses our Visual AI engine in the design of our AI range of lenses, which include the second generation of the award-winning Autograph Intelligence progressive and the newly launched Driver Intelligence group of lenses.
‘Mimicking human intelligence, the engine runs a multi-dimensional optimisation process, based on real measurements of real people and sophisticated optimisation algorithms with elements of artificial intelligence.’
For its Driver Intelligence lens, Shamir analysed data gathered from more than 40 drivers in 80 different vehicles, including race drivers from its partnership with Alpine.
‘During development of Driver Intelligence, the research and development (R&D) team used the Shamir AI-tooled engine for analysing this data. An understanding was gained about the most important areas of the lens for optimal driving vision, which allowed the R&D team to develop the most suitable lens design that was carefully adapted to the special visual needs of the driver.
‘Its unique design ensures minimal head movement and greater freedom of eye movement. This contributes to greater comfort and safety while driving, improved reaction times and an improved ability to scan the surroundings, allowing better overall driving control.’
AI is also used in Shamir’s Spark 4 and Spark 4W digital measuring devices to ensure the system continuously learns and upgrades to make measurements more accurate and faster over time.
‘This allows the eye care professional to then provide their customer with a more accurate and smoother journey during the dispensing process.’
Shamir adds that AI will be used across new R&D projects to enable faster and better learning and quicker adaption of newer ideas that come from the R&D team.
Studying visual behaviour
At EssilorLuxottica, Alan Pitcher, commercial director for wholesale lenses, tells Optician that AI is being used to further lens personalisation to meet the differing visual needs of each wearer.
‘Our dream is to make a lens that is optimised for the visual needs of every single wearer and AI is a tool we’re beginning to use to help us achieve just that. Making the design process much more accurate and efficient.
‘As a leader in the ophthalmic lens industry, EssilorLuxottica is able to collect a vast amount of anonymised data from lens orders, test data and internal studies. AI enables us to extract knowledge out of this data and use this to improve our lens designs,’ Pitcher says.
Behavioural AI was used in the design of EssilorLuxottica’s latest generation of varifocal lens, the Varilux XR series (pictured right), which launched in May this year.
Pitcher explains the company asked 4,000 wearers about difficulties they experienced when wearing progressive lenses. Respondents bemoaned the problems of living a hyper-connected life and this became the starting point for the lens design.
He says: ‘Varilux XR series is our first eye-responsive varifocal lens created using digital twin technology, which replicates a patient performing multiple realistic visual tasks in a 3D environment.
‘Using the AI digital twin technology, we were able to study a patient’s visual behaviour, which is the combination of head and eye movements the patient makes to view his environment. Taking into account the visual behaviour is key to designing progressive lenses, to adapt to patients’ needs and visual demands.
‘The power of AI lies in the quantity, quality and variety of data analysed. We analysed more than one million pieces of data, which allowed us to establish a visual behaviour profile using this virtual modelling for every single prescription.’
Pitcher says XR-motion technology was built into the lens design to address the challenges outlined in patient feedback, which reduces the optical binocular disparities between the two lenses and ensures precise positioning of the focus zones adapted to each wearer.
‘Both lenses are optimised binocularly according to the predicted visual behaviour profile to provide simultaneously better positioning of the near vision zone and optimised sharp binocular vision,’ Pitcher adds.
Biometric data
Rodenstock was founded by Joseph Rodenstock in 1877 upon a simple principle: ‘To realise 100% of every individual’s visual potential.’ Andrew Copley, head of professional services at Rodenstock UK, tells Optician this principle continues to guide and drive the company into the future.
‘It should, therefore, come as no surprise that Rodenstock embraced the power of predictive analysis many years ago when they released their MyLife and FreeSign lens designs.
‘By using machine learning to identify the behavioural patterns of different skill sets and/or visual demands, Rodenstock was able to create lens designs that had the optimal lens powers in the optimal place to make both adaptation and utilisation quick and comfortable,’ he says.
Copley adds that there is an abundance of ophthalmic lenses on the market that incorporate AI, which he believes is better for patients, and this includes Rodenstock’s Biometric Intelligent Glasses (Big) Norm products.
‘Coming off the back of the Big Exact lenses, with which Rodenstock stands as the only lens manufacturer capable of directly incorporating biometric parameters of the individual eye into the lens production process, Rodenstock applied AI towards the huge pool of biometric data collected by our DNEye scanners.
‘When Big Norm was launched, we had a pool of over half a million precise biometric scans of real eyes and this pool of data is constantly growing. AI was used to find correlations between the standard prescription values found in a lens order and the measurements of the eye’s anatomy, such as axial length, corneal power, chamber depth, etc,’ Copley explains.
Rodenstock says when calculating the final lens design and power for an individual, it does not believe it should use a simplified static model of the eye, where the assumption is made that everybody’s eyes are the same size and shape.
Copley says: ‘Rodenstock created a new norm for calculating the lens with the use of AI, making Biometric Intelligent Glasses available to everyone. I, for one, cannot wait to introduce this technology to more and more people.’
The beginning of the future
Sean Donnachie, technical director at Norville, says AI is one of the most quickly developing fields in the ophthalmic industry and has the potential to improve production processes over the next few years.
He tells Optician: ‘It is well known that producing an ophthalmic lens involves a high degree of waste, much of which cannot be recycled or reused. A finished spectacle lens begins life as a large chunk of plastic, which has approximately 80% of its material removed by the time it reaches a patient. Aside from recycling the water used during the process, the swarf (tiny particles of plastic removed during grinding) goes to landfill. One of the many potentials for AI in the future is to address the issue.’
Donnachie hypothesises that, in years to come, laboratory machines could communicate directly with lens casters to create slimmer semi-finished blanks, depending on the requirements of a particular job, which would lead to far less swarf being sent to landfill. Or there could be a system that monitors each line of production and can identify the ‘lulls’ in production, shutting down the machines to conserve energy when appropriate without
affecting delivery times.
‘A great deal of time in the production process is spent on quality control, often through visual inspection, which is a timely process. We already have systems that monitor production lines to identify when a process is failing, alerting the operator to take corrective action.
AI could be integrated to these systems to create an environment where the machine identifies an error and makes decisions to take corrective action without requiring human intervention. For example, if the AR coating chamber becomes compromised by dust, the machine could identify this and implement the necessary steps to ensure no other batches are affected, reducing waste and avoiding customer complaints.
‘Machine learning algorithms to create progressive lenses with an almost 100% acceptance rate have been in the market since 2019. The future for AI in lens design is already written; for production, it is only beginning,’ Donnachie concludes.