Melbourne Rapid Fields

Melbourne Rapid Fields: Expanding Your Practice with Modern Vision Testing Technology

BY SELWYN PREA, PROFESSOR ALGIS VINGRYS, AND DR GEORGE KONG

LEARNING OBJECTIVES

Understand the benefits and limitations of modern technologies applied to vision testing,
Recognise how acuity optotypes can be used to identify vision disorders at different stages of the visual pathway,
Understand key differences between the MRF tangent perimeter and conventional bowl perimeters,
Appreciate the limitations and correct patient set up when undertaking perimetry with MRF, and
Detail the clinical implementation of a tablet perimeter in patient management.

Medicine and the ophthalmic industries have embraced new technologies such as smart phones, tablets, and virtual reality (VR) headsets. In doing so, clever innovation now allows cheap and reliable tests of vision compared with more expensive and older forms of clinical tests, such as TV acuity charts and bowl perimeters.

More importantly, these newer devices also offer the advantages of being portable, requiring a small footprint in a practice, and using novel testing paradigms to test vision and provide efficient and accurate outcomes.

They can be used to test many aspects of vision, such as spatial vision, contrast sensitivity, colour thresholds and with some clever innovation, increment thresholds across the visual field. They can also be applied for eye movement measurement as used with many modern video games, although for ophthalmic application this may have limited accuracy.

The MRF was initially developed over 2013 to 2014 as an iPad-only application. This was due to the quality of iPad screens, also known as ‘retina’ screens. Our reasons for not choosing head mounted display (HMD) technology will be detailed later.

In 2016, having fine-tuned and undertaken early development trials, our group published on the clinical application of the iPad MRF for the first time. This was followed in 2018 by an independent study that reported similar and very positive findings.

Since then, there have been many clinical trials confirming that the technology does well when compared to routine clinical devices. The MRF has now been registered as a vision test and perimeter in six jurisdictions (TGA (Australia), Medsafe NZ (New Zealand), CDSCO (India), CE-mark (European Union), MHRA (United Kingdom) and FDA (United States)) with some 30,000 tests having been performed worldwide.

COVID brought about many changes to ophthalmic practice and two of these were that bowl perimetry was rendered impractical due to the potential for cross-infection and the complex disinfection procedures required for the bowls. COVID also affected iPad availability, as sources of key components from China were in lockdown. These issues made iPad testing desirable but unattainable.

Given this situation, coupled with rapid improvements in modern PC hardware and screens, we decided to develop software for browser platforms (Chrome, Safari, Firefox, etc.) that could run on any PC, Android, or Apple device. We successfully achieved this outcome by mid 2020 and released our first browser based MRF called MRF-web.

Due to hardware limitations of PC and Android devices, we have released software for visual acuity, contrast sensitivity, and visual field testing, but have left colour vision thresholds in abeyance until screen and hardware quality improve across the majority of devices.

Other Devices and Applications

Having outlined MRF technology, we will discuss the various devices and applications that can be used by optometrists, detail their limitations, and give specific examples of clinical cases tested with the MRF.

HMD Technology

HMD technology was considered during our early development but was shelved at the time due to the limited spatial and chromatic capacity of HMDs, especially with off-axis viewing where the presence of higher order aberrations is problematic, as is the likelihood of cybersickness with low or full immersion.8 Our American partner (M&S Technologies) has developed an HMD vision test (Figure 1) for their specific market.

Figure 1. Smart System virtual reality headset developed by M&S Technologies (Illinois, U.S.A.) for vision testing.

To identify the limitations given in Table 1, Lynn et al. measured two stand-alone virtual reality (VR) systems and four smart-phone VR headsets. The authors noted that in many cases, their measured values were less than those reported by manufacturers and recommended that practitioners need to measure these for themselves. Of note, although the field of view is smaller than claimed by the makers, this dimension is adequate for testing the central visual field (to 30º) and, for low vision, it is like telescopic aids.

Though the authors discuss factors that can impact acuity and visual field thresholds, they were not measured. These are important considerations as the magnitude and effect that the higher order aberrations of high plus lenses in the oculars can have are not well understood in these applications.

Finally, one issue of importance for perimetry is that the screens used with HMDs are tangent to the visual axis, so perimetric spots need to be scaled to allow for this effect. Despite the limitations identified in Table 1, Lynn et al. concluded that contemporary equipment should be adequate for central visual field testing, vergence testing and for use as low vision aids.

Table 1. Limitations of contemporary HMD vision tests from Lynn et al.

Table 2. Things that optometrists need to achieve before running MRF-online using a browser interface on a tablet, laptop, or PC.

Table 3. Key differences of MRF-online compared with other commercial devices.

One benefit of HMD technology is that both eyes can be tested concurrently, with the participant being unaware of which eye is being examined. Nevertheless, cross contamination from HMD screens is possible and practitioners should understand the methods for disinfection.

One high quality HMD perimeter, called the IMO, comes from Japan. The makers (CREWT Medical Systems) have incorporated a field stop to limit the field of view to about 30º, which allows a 30-2 test grid, and limits the effects of off-axis aberrations on vision. This is an excellent device that uses modern Bayes test algorithms for thresholding and eye tracking to produce retinal stabilised image presentations up to +/-5º. However, it is not clear whether spot size is adjusted for the flat screen of the IMO. Publications comparing this device to the Humphrey, report outcomes with high accuracy and repeatability that require about 10 minutes to test both eyes on a 30-2 grid compared with 15 minutes for Humphrey.

Not all patients can cope with a HMD or the VR environment: many feel disoriented or even nauseous on immersion, called cyber sickness.

Although the HMD perimetry task is more akin to semi-immersion, as it does not involve flow or motion in the images, Matirosov et al. reported that 17% of their young participants felt nauseous after 10 minutes of semi-immersion. How this will impact vision testing and perimetry is not well understood.

Recently, the makers of IMO have released a stand-alone, non-head mounted version of IMO, called the IMOvifa which should reduce immersion-induced cyber sickness. As the IMOvifa uses IMO software, it would be expected to have similar and excellent performance comparable to the HMD version.

An interesting observation from a recent clinical trial involving 273 patients is that, given the choice of a head-mounted test (IMO) or the free standing IMOvifa test, 126 patients (46%) chose to be tested by the free standing (IMOvifa) device and not the HMDs. The effect that HMD VR perimeters have for nausea and disorientation needs to be defined.

Current virtual reality headsets have been created for gaming, and the virtual reality community has driven the development of this technology. For optometry, the limited spatial resolution of HMDs will not support visual acuity testing or high spatial frequency image content for amblyopia therapy. These limitations also restrict hyperacuity threshold measurement.

Apart from these applications HMDs are well suited to perimetry as; they have an enclosed test environment that removes the impact of ambient light and other distractions, they can track eye movements and, the size of the spot can be adjusted for the flat screen of the HMD.

Tablets, Laptops, or PCs

Tablet technology does not cause cyber sickness and in most cases the spatial resolution of many modern tablets, and especially the Apple retina screens, allows for 6/6 or better acuity at near (33cm). Furthermore, 93% of elderly Australians have internet access, with 59% using tablet devices and 42% laptops or PCs (or multiple devices) so adopting such technology for vision testing will provide a familiar environment for most elderly Australians. In fact, most people will be able to do the testing at home using their own equipment (telehealth). It is for these reasons that the MRF was designed to operate first on an iPad tablet and later as MRF-web.

MRF-web: Visual Acuity Testing

Tahir et al found that two tablets, iPad (Gen 3 or newer) and Nexus 10, have adequate pixel density to produce optotypes that can achieve acuity testing down to 6/6 at 33 cm. It is likely that the higher quality screens of modern devices will allow acuity testing on many tablets or laptops beyond the 6/6 level, and for this to occur the device needs to have a pixel density greater than 100 pixels per cm (254 pixels per inch).

The MRF tests visual acuity using a Landolt C target shown in a box, with sides that act as interaction bars (Figure 2). The box has been added to give the single letter optotype a similar level of interaction and acuity, as if it were shown in a line of letters (eg. Bailey-Lovie or ETDRS chart). The patient undertakes a four-choice, matching task to identify the orientation of the gap in the boxed optotype then selects one of five options at the bottom of the screen that matches the target.

Figure 2. (A) The MRF acuity test showing the boxed target at the top of the screen and the four orientations that patients can choose for a match at the bottom. Patients are required to guess, but if they cannot guess they should select the question mark at the bottom (far left). (B) High contrast acuity (HCA) target. (C) Low luminance, low contrast acuity (LCA) target. (D) Acuity-in-noise (AiN) target.

Figure 3. (A) Visual acuity outcomes measured multiple times over three years on 223 contact lens-wearing United States aircrew tested with a high contrast chart (black bars) and a low contrast, low luminance chart (grey bars) from Lattimore (2017). Note how few contact lens wearing pilots reach the criterion (6/7.5) for adequate operational vision under these obfuscating conditions. (B) Visual acuity measured with optotypes shown immersed in luminance noise in 50 cases of acute ischemic stroke (black bars) and 30 controls (white bars) from Wijesundera et al.12 The dashed vertical line is the 95% limit for controls. Note how acute stroke reduces acuity-in-noise in most cases.

The MRF acuity test uses one of three optotypes, namely a high contrast acuity (HCA) optotype shown on a bright background (135 cd/m-2, Figure 2B). This is used to ensure the patient understands the testing logic and they are wearing their reading glasses (if needed). It can be used to detect refractive error when a +3 lens is applied. The second optotype is a low contrast (18%) optotype (LCA) shown on a low luminance background (5 cd/m-2, Figure 2C) designed to pick up retinal or corneal abnormality and cataract (see Figure 3A). This was included as a test of contrast sensitivity. And finally, an acuity-in-noise (AiN) target is available to detect brain abnormality, such as amblyopia or acute stroke (Figure 2D).

Lattimore reported on the outcomes for 223 contact lens-wearing United States Army Apache pilots who had their vision tested multiple times over three years. He used a standard HCA projection chart at 6m (106 lux) and a low contrast (8%) Bailey-Lovie chart at the same viewing distance but at a dimmer 31 lux background. Figure 3A is redrawn from his report.

This shows that while all contact lens wearing aircrew had good HCA, very few achieved operationally acceptable vision under conditions of low luminance and low contrast, as would be encountered during flight operations. He suggested aircrew need 6/7.5 (logMAR 0.1) acuity for operation under low luminance and low contrast conditions.

The purpose of the acuity-in-noise optotype is to identify brain dysfunction. Figure 3B shows outcomes for 50 cases of acute ischemic stroke measured with AiN targets. All patients, except for four stroke cases, had normal HCA. Controls give up to a one line loss for AiN targets, but Figure 3B shows an average three-to-four-line loss (0.35 logMAR) experienced by recent stroke cases. Similar losses are observed in cases of amblyopia indicating that acuity-in-noise optotypes are selectively affected in the presence of acquired or congenital brain disorders.

MRF-web: Visual Field Testing

The MRF-web allows visual field thresholding on tablets, laptops or PCs with software that can be accessed with any browser (Safari, Chrome, Firefox). Table 2 lists considerations to ensure accurate outcomes are achieved.

While a full and detailed description of the MRF visual field test can be found at Vingrys et al, Table 3 summarises the key operational aspects. In an early version of the iPad application, the test grid was optimised for specific disease entities, such as glaucoma, maculopathy, diabetes and neural. This was a similar strategy to the Octopus perimeter, however MRF also allowed the traditional 24-2 test grid. Having found minor benefit in using optimised grids in clinical application, we recently adopted a modified 24-2 (and 30-2) grid configuration to facilitate comparison with other devices. Our modification adds four foveal test points (0.75o) to expose foveal losses..

The MRF uses a Bayes predictor for threshold, as do many modern perimeters. The approach adopted by the MRF is similar to SITA (Carl Zeiss Meditec AG), although the presentation levels are not independently set by a staircase. Instead, they are optimised for neighbourhood clusters to better reflect the nature of vision loss. Three presentations are always used at each location to reduce the impact of false errors. Neighbourhood logic identifies and rechecks unexpected responses in any cluster and adds extra test points to screening test clusters to better define the spatial extent of a scotoma (Figure 4).

Figure 4. MRF screening outcome for the left eye of a 68-year-old male with a central scotoma and reduced acuity (6/24). This took two minutes, two seconds with neighbourhood logic adding seven extra points at 3º eccentricity to the usual grid pattern to define the central lesion. The coloured bar to the right (called risk of abnormality) shows the probability that this test is normal, derived from normative data. In this case it is not normal, as indicated by the yellow spot lying at the bottom of the bar (red zone).

Figure 5. (A) The 24-2 test took three minutes, 11 seconds for full thresholding. Note how the four extra foveal test points (at 0.75º) in our modified grid identify the scotoma, whereas the usual 24-2 test grid locations did not, except for one location down by 7dB. The large PD confirms the presence of a deep, localised, scotoma. The coloured flag (red) identifies departure from age matched expected values in one or both global indices (MD and PD). (B) The 10-2 result for the same patient is shown in the upper panel. The 10-2 test required four minutes, 10 seconds for testing, being longer than the 24-2 test because more locations were found to be abnormal with the 10-2 grid (23, 10-2 vs 5, 24-2). This is also reflected by the higher MD and PD values of the 10-2 test.

An adaptive response time, derived from past performance, speeds up (or slows down) test presentations.

MRF-web completes a screening test in about one to 1.5 minutes. (Figure 4) and a 24-2 threshold test in about three to four minutes (Figure 5) in a normal eye. Our clinical trials find that the test times lie in between the SITA fast and SITA faster algorithms, even with the four extra foveal points, false positive, false negative and fixation loss checks.

Test spots are scaled in size across the screen. This allows for the tangent screen and has been further scaled to return near constant thresholds at each normal location (30 dB). Such scaling increases the dynamic range of the test to monitor scotoma and reduces test variability in peripheral locations.

In terms of finding this macula abnormality, the screening test gives the fastest outcome. Moreover, if this patient were to develop glaucoma at a future date, the screening grid is well placed to define such an occurrence. Monitoring change in this patient needs the 10-2 grid, although the 24-2 grid would be best if glaucoma were ever to occur in the future.

Conclusion

MRF vision testing software is presently available via a web browser on Mac, PC, and tablet devices. Clinical trials show that it performs accurate and reliable assays of visual acuity and visual field thresholds in a short duration. A survey of patients with chronic eye disease who undertook telemedicine over 12 months with MRF found it easy to use and indicated a preference for tablet testing over conventional clinical methods. Optometrists can adopt this novel technology within their practices as a cost-effective means of vision testing that can be undertaken by patients at home or during their consultation.

This article is sponsored by Glance Optical Pty. Ltd., the makers of the MRF vision software.

Credits: https://www.mieducation.com/pages/melbourne-rapid-fields-expanding-your-practice-with-modern-vision-testing-technology