An Australian company is taking on the giants of ophthalmology with its own OCT device. LEWIS WILLIAMS looks at what the Cylite HP-OCT is doing different to set itself apart.
When visiting trade displays at conferences, it is always wise to pay particular attention to the lower-key exhibits. That was certainly the case at the recent O=MEGA19 conference and trade exhibition held at Melbourne’s Convention and Exhibition Centre in July.
There, on the Medmont stand, was an understated inclusion – an Australian designed and made optical coherence tomography (OCT) instrument.
It is difficult to imagine a more unlikely or more unexpected development. The USA, Germany and Japan come to mind as OCT device developers, although there are also products from South Korea, Poland, and China already in the market.
Against that line-up of established sources, it would be pointless for an Australian company to develop a ‘me-too’ product. Fortunately, that is exactly what Cylite, a Melbourne-based start-up, is avoiding.
Founded in 2013, the company is backed by four of Australia’s most experienced scientists drawn from the fields of photonics, optics, instrumentation, and software development. Add a dose of funds derived from venture capitalists, and you have an OCT developer – in Australia.
In the beginning
The instrument’s concept came from the mind of Cylite’s CEO Dr Steve Frisken, a 25-year veteran of the photonics industry and the holder of more than 30 patents. The company’s CTO is his brother Mr Grant Frisken, the brains behind the software powering the Medmont E300 topographer; a well-established and Australian-made world-wide success story. Most of the others involved in the project have backgrounds in telecommunications and photonics.
Dr Simon Poole AO is Cylite’s co-founder and the project’s customer-facing front man. He has more than 37 years of experience in academia and photonics, and in 2018 was awarded the Order of Australia for his contribution to the latter. Steve Frisken and Poole worked together for 30 years in a previous company. Vice President of R&D Mr Trevor Anderson is also connected to Poole through a previous company.
The seeds of developing an OCT device were sown when some of the founders worked for the Australian division of Finisar Corporation, a multinational optical communications business. However, the project was eventually cancelled because it was thought to be too far removed from the company’s core activities.
The newish OCT initiative is a furtherance of that project. Accepting that there was no room for a ‘me too’ project, an analysis by the Friskens suggested that eye motion during OCT scanning was the Achilles heel of current devices, which established the catalyst for the current project.
Removing the artefacts associated with eye motion also allowed the development of greatly improved metrology and biometry, with a further goal to eliminate the need for a technician to use the device and greater ease-of-use. To that end, the device is largely autonomous. It aligns and focuses itself, thereby automating data acquisition.
The new technology was revealed for the first time in a paper presented at the 2015 SPIE BIOS conference. Since then the company has exhibited at several conferences, including recent ARVO and ESCRS events. In a local context, two of the company founders were honoured with a Prime Ministers’ Prize for Innovation in 2018.
The Cylite HP-OCT
The company is targeting both eyecare practitioners as well as eye researchers with their device. Clinician-aligned applications to be offered include: IOL and refractive surgery planning (including epithelial thickness), pre-surgical screening (retina and cornea), corneal, miniscleral, and scleral CL fitting, Ortho-K fitting and monitoring, axial length monitoring in a myopia/myopia progression context, and anterior angle screening (virtual gonioscopy) in a glaucoma context.
Researchers are being induced with offers of: crystalline lens accommodation studies, longitudinal studies of axial length, retinal shape, myopia progression, volumetric analysis of retina and choroid, corneal biomechanics and elastography, epithelial thickness and tear film dynamics. The release of the second version of the device’s evolving software suite is now imminent.
The device’s key differentiating feature is its so-called HyperParallel (HP-OCT) optical basis (see schematic) that allows the instrument to take a ‘snapshot’ of the subject’s eye, negating most of the downside of their eye’s movements. In essence, it is parallel 3D spectral-domain OCT.
Negating the effects of eye movement is achieved by creating 1,000 beamlets of light simultaneously, rather than scanning a single spot across the eye as is done by all current OCT products. The 1,000 beamlets are created using a 24 x 42 lenslet array giving, theoretically, a maximum of 1,008 beamlets.
In effect, the 1,000 beamlets created act as 1,000 simultaneous optical A-scans of the eye. Up to 300 of those simultaneous scans are imaged and logged every second, meaning a 300,000 data point per second processing load for the computing chain that starts with in-sensor pre-processing followed by image processing on an attached computer.
Dithering around each of the 1,000 ‘spots’ allows the instrument to ‘fill-in’ the regions around the initial 1,000 spots to provide a unique volume image of the eye. Eye motion is countered further by significant registration of overlapping areas in serial images (300 per second).
Conceived originally as an anterior eye device, the team’s ambition has since expanded to include retinal and high-resolution modes. Each of those three capabilities is dependent on the installation of specific objectives to the front of the instrument. Each objective contains specialised electronic ‘smarts’ that engage/interface automatically and the whole assembly attaches precisely, magnetically.
The light source is a pulsed superluminescent diode (SLD) with a centre wavelength of 840 nm and a bandwidth of around 30 nm. The power of the SLD is modest to comply with regulatory controls that limit the eye’s, and especially the retina’s, exposure to excessive light levels. Pulsing allows greater peak-power output without exceeding averaged radiation limits.
The SLD’s output is relayed telecentrically to a beam splitter that provides light to the subject’s eye, an internal reference, and also mixes returning light from the eye and the reference. That returning light is passed to an internal spectrometer which then images the processed light onto a fast, 2D, digital sensor array, which also undertakes the early stages of image processing.
Data processing into images, maps, and numerical information is outsourced to an attached computer. Claimed axial resolution is 9.4 microns with lateral resolutions of 11 microns (high-resolution objective), 19 microns (retinal objective), and 33 microns (anterior objective).
In a lengthy conversation with Poole, it was explained that while resolution can be increased, any such improvement is achieved at a cost to the depth of penetration in the images. The area scanned depends on the objective used and ranges from 16 x 9.5 to 14 mm (anterior objective) to 5.5 x 3.2 mm (high-resolution objective). The depth imaged ranges from 11 mm (17 mm in extended mode) (anterior objective) to 8.3 mm (retinal objective). A 16 x 16 mm scan is possible by generating two overlapping images with image registration provided by the overlap.
All data processing not done internally is handled by an external computer connected via a USB 3 interface. An early challenge was making the processing and transfer fast enough for the amount of data created, but the use of parallel processing now means that excessive power is not required and a good GPU in a laptop is adequate.
Currently, the on-screen imaging, image rotation, image sectioning and slicing, and so on are done using standard medical industry software, much of which is open source. This means that knowledge of the software is more readily available and customisation on the user end is feasible; something not possible with closed source, proprietary software solutions.
Applications for US patents (9,913,579, 9,861,277, and 9,955,863) covering the technology were submitted early in the development process, and were granted in 2018.
The crystal ball
Production is scheduled to be done in Australia but manufacture and supply of several of the specialised optical and electronic sub-assemblies will be outsourced, possibly overseas.
Surprisingly, some early instruments are already in the hands of ophthalmic researchers at various institutions around the world. Before general release and use on humans, various regulatory regimes will have to be successfully navigated. Pricing will be competitive and, in keeping with expectations based on extensive commercial experience, the company expects some price erosion as the OCT market matures following release.
When asked about the possibility of a hand-held version of an HP-OCT, Poole stated it is an impossibility due to the ‘incompressibility’ of the optical engine at the heart of the device.
The company does not plan on being a one-instrument company in the longer term, and R&D is already underway on further extensions of the technology into new areas including holoscopy and hyperspectral imaging.
Historically, Australia had a significant ophthalmic instrument manufacturing presence, albeit when instrumentation was much simpler, with Australian Optical, British Optical (now BOC), SOLA and others manufacturing devices.
The manufacture of WWII optical munitions in Australia by the first two companies mentioned above, as well as others, makes an interesting story in itself. However, the capabilities demonstrated and developed during the war were central to post-war optical endeavours, especially to instrumentation and spectacle lens manufacture.
An expansion of the current local ophthalmic instrument industry would be welcomed by most Australians, provided competitive products are offered.