Visual Field Testing: The Long View

My interest in vision assessment stems from an undergraduate visual perception laboratory class that introduced me to unresolved problems in visual neuroscience. I was captivated. There was so much to do: we needed a better understanding not just of the neurophysiology of vision, but also of the visual effects of systemic diseases, such as diabetes. Back then, remember, retinal examination was the only way of directly observing living blood vessels within intact tissue, and therefore had applications well beyond vision assessment. So that triggered my fascination with the evaluation of eye diseases, and resulted in a career focused on investigating modalities that can help us assess visual disorders. 

Early days

One of my earliest projects – undertaken with a colleague, John Keltner, a neuroophthalmologist at the University of California, Davis – was the assessment of automated perimeter technology. We started working on the Fieldmaster Perimeter, and then moved on to the Humphrey Field Analyzer and the Octopus Perimeter. In the early days, we were skeptical regarding the real-world utility of this technology – but the more we looked at it, the more optimistic we became. By the mid-seventies, we had concluded that automated perimetry would probably replace manual perimetry in clinical practice. Some ophthalmologists resisted this development, taking the view that VF testing would always require a skilled perimetrist. But we persisted, and I think we have been proved right over time – few rely on manual VF tests today.

One benefit of our first project was that it allowed us to develop techniques to objectively assess and improve the clinical performance of perimeters. That has been an ongoing field of endeavor over the years: our refinements have included short wavelength automated perimetry (SWAP) – a method that can detect glaucomatous visual field deficits significantly (up to 10 years) earlier than standard white-on-white perimetry. More recently, we developed frequency doubling technology (FDT) perimetry; because glaucoma is associated with reduced contrast sensitivity to frequency doubling stimuli, FDT can assist glaucoma evaluation.

In the real world, however, new technology often works best for the inventors – they are familiar with the system and are careful to ensure it works well. Consequently, the true test of an invention is whether it can be broadly adopted – so I am very happy to see that others are successfully using the perimetry techniques and devices that we helped to develop, at both software and hardware levels. In fact, it has been extraordinarily gratifying to see our technology so widely disseminated.

Current affairs

Our present research interests include the application of tablet devices and virtual reality (VR) headsets in VF testing. One objective is to develop low-cost tablet-VR systems applicable to countries that do not have the resources to justify acquisition of expensive automated perimeters. Similarly, we anticipate the portability of tablet-VR systems to benefit patients living in remote areas in these and other countries. Accordingly, we tested prototypes of our tablet-based VF test (Figure 1) in Nepal, where we screened about 400 eyes (1), often in people living at elevations of 15-16,000 feet who had never before had any kind of medical examination, still less an eye test. The results were very encouraging; our system revealed cases of undiagnosed glaucoma and diabetic retinopathy; I am optimistic that this approach, with refinement, could transform the detection and diagnosis of eye diseases in developing countries.

Another objective is to launch tablet-based devices in the developed world, where healthcare systems are under pressure to reduce waiting times and improve efficiency. I envisage employing tablet-VR headset systems to test patients while they are in the clinic waiting room, so that they enter the doctor’s office with their test results in hand. That would eliminate delays associated with first seeing the doctor and then ordering the tests. Similarly, home-testing becomes feasible with these systems. For all these reasons, I am convinced that tablet and VR headset VF testing is the wave of the future.

That said, this technology is still in its early stages and, if it is to be used in the ways outlined above, we must ensure that it is very robust and resistant to sources of error. For example, when using a tablet alone it is difficult to ensure that the patient remains a constant distance from the screen. That issue can be eliminated by headset use; adjusting for eye movement, however, is trickier, and is one of the key challenges to overcome before tablet-VR systems can be reliably and reproducibly used in a home environment. Another point is that headsets were originally developed for computer games – they aren’t made with the same attention to quality, precision and calibration that one would see in the medical device industry. Therefore, to make a device suitable for broad application in VF testing, we need to develop headsets specifically designed for medical use; to that end, we are working with M&S Technologies to modify aspects of headset software and hardware. We are also collaborating closely with Algis Vingrys’ and George Kong’s group in Melbourne, Australia, on the development of quantitative tests applicable to the tablet-headset systems.

Success in this field will significantly benefit particular groups of patients, notably. those who cannot optimally communicate or follow instructions – for example, very young children. In fact, we looked at adapting FDT to this group some years ago (2). With modification, FDT gave nice results in patients between 5 and 14 years, and even provided reliable data from a three-and-a-half-year-old. More recent approaches to develop VF tests for this group of patients include “Caspar’s Castle” (3), which is the result of some very careful work. I believe it is a very promising system for VF testing in children (please see part 3 of this feature: “Building castles for kids”). My only concern about this type of game-based approach is that, while it has the advantage of holding children’s attention, it may be too entertaining – kids might get trigger-happy and start pressing the button at the wrong times! I’d also like to see some independent testing of Caspar’s Castle, to show that it works for people other than the inventors.

Another technique for VF testing in young children, Saccadic Vector Optokinetic Perimetry (4), relies on automatic monitoring of eye movements to establish stimulus detection. It’s a nice concept but, in practice, it requires significant calibration and therefore is somewhat time-consuming; also, I’d need to see more data from a range of different users to be convinced that SVOP would be useful in real-world settings. Furthermore, SVOP may not be quite as objective as it seems – it still requires cooperation from the patient, and it still needs somebody to interpret the results. Nevertheless, I like the idea of using eye movements as a way of testing children, particularly infants.

Tomorrow’s world

In the future, I expect our focus will shift away from refinement of data capture methods towards data analysis methods. At present, we cannot always distinguish genuine systemic changes – improvement or progression – from background variability; inter-test variation due to patient-specific factors remains a major challenge. Maybe two hundred investigators, including myself, have been working on this problem, but there is still no consensus on the best way to monitor changes in visual fields over time. Nevertheless, I now expect to see rapid progress in VF testing, driven in particular by expertise in deep learning and artificial intelligence. Remember, we already have an FDA-approved deep learning/artificial intelligence tool for the diagnosis of proliferative versus non-proliferative diabetic retinopathy by image analysis; using this kind of approach to interrogate visual field data could be very productive.

I also anticipate advances in systems that simultaneously provide both structural and functional information. Automated micro-perimetry is a first step in this direction in that it not only generates perimetry data but also prompts scanning laser ophthalmoscopy. Future development of more sophisticated systems, which can simultaneously measure structure and function, will greatly improve our ability to monitor glaucoma, retinal degeneration and other ocular conditions. But we’re in the very early stages of that endeavor.

Looking much further ahead, I hope to eventually see the advent of techniques that can sensitively and specifically measure the electrical activity of the visual system so as to identify specific points in the visual pathway that are damaged or declining in some way. This technology would facilitate assessment of difficult patients – for example, very young children, or patients with cognitive impairment or neurological disease – and would be more objective than current methods. Furthermore, techniques able to identify areas that are compromised but not yet irreversibly damaged would be far more useful than our current tools, which only identify areas that have already failed. It is far better to identify at-risk areas by searching for discrete regions of dysfunction, maybe with regard to oxygen consumption, mitochondrial activity or other measures of metabolic function. Indeed, I can envisage localized metabolic function being monitored in this way on a continuous basis, perhaps via an implanted sensor. Such an approach might give us the opportunity to prevent significant damage, rather than attempting to fix it after the event.

You may be thinking, “this all seems pie-in-the-sky!” But having seen how the visual testing field – not to mention information and communications technology – has changed over the last forty years, I am very optimistic about the future clinical management of patients with vision problems. From here on, it’s all change!