Purpose: Keratoconus (KC) represents one of the leading causes of corneal transplantation worldwide. Detecting subclinical KC would lead to better management to avoid the need for corneal grafts, but the condition is clinically challenging to diagnose. We wished to compare eight commonly used machine learning algorithms using a range of parameter combinations by applying them to our KC dataset and build models to better differentiate subclinical KC from non-KC eyes. Methods: Oculus Pentacam was used to obtain corneal parameters on 49 subclinical KC and 39 control eyes, along with clinical and demographic parameters. Eight machine learning methods were applied to build models to differentiate subclinical KC from control eyes. Dominant algorithms were trained with all combinations of the considered parameters to select important parameter combinations. The performance of each model was evaluated and compared. Results: Using a total of eleven parameters, random forest, support vector machine and k-nearest neighbors had better performance in detecting subclinical KC. The highest area under the curve of 0.97 for detecting subclinical KC was achieved using five parameters by the random forest method. The highest sensitivity (0.94) and specificity (0.90) were obtained by the support vector machine and the k-nearest neighbor model, respectively. Conclusions: This study showed machine learning algorithms can be applied to identify subclinical KC using a minimal parameter set that are routinely collected during clinical eye examination. Translational Relevance: Machine learning algorithms can be built using routinely collected clinical parameters that will assist in the objective detection of subclinical KC.

Purpose: To examine 2-year progression rate and associated biometric changes in highly myopic eyes. Methods: This is a longitudinal, observational cohort study that included 657 participants aged 7 to 70 years with bilateral high myopia (≤-6.00 diopters [D]) and followed for 2 years. All participants underwent ocular biometry and cycloplegic refraction examinations. Main outcome measures were changes in spherical equivalent refraction (SE) and ocular biometry in the right eyes. Results: Mean age of participants was 21.6 ± 12.2 years. At baseline, mean SE was -9.82 ± 3.28 D and ocular biometric measurements were 27.40 ± 1.56 mm for axial length, 3.16 ± 0.27 mm for anterior chamber depth, 3.60 ± 0.35 mm for lens thickness, and 20.09 ± 1.50 mm for vitreous chamber depth. After 2 years of follow-up, there was a trend toward more myopia and greater axial elongation in all age groups. Younger participants (≤20 years) had significantly (P < 0.001) greater rates of myopic shift and axial elongation compared with older participants (>20 years). However, highly myopic adults aged 40 to 70 years continued to demonstrate refractive progression, particularly if they had extremely high myopia (≤-10.00 D). In the multiple regression analysis, each additional diopter of myopia at baseline was associated with a 11% higher risk of a >1.00-D/y myopic shift (odds ratio, 1.11; 95% confidence interval, 1.04-1.18; P = 0.002). Conclusions: Longitudinal data from this large Chinese cohort suggest that highly myopic eyes continue to progress in SE throughout life, with the greatest rates of progression observed in younger participants. Axial elongation rates appeared to stabilize after 20 years of age and were predominantly due to an increase in the vitreous chamber depth. Other risk factors for a myopic shift included a higher degree of myopic refraction at baseline.

BACKGROUND: In the assessment of a pituitary mass, objective visual field testing represents a valuable means of evaluating mass effect, and thus in deciding whether surgical management is warranted. CASE PRESENTATION: In this vignette, we describe a 73 year-old lady who presented with a three-week history of frontal headache, and 'blurriness' in the left side of her vision, due to a WHO grade III anaplastic haemangiopericytoma compressing the optic chiasm. We report how timely investigations, including an iPad-based visual field test (Melbourne Rapid Field, (MRF)) conducted at the bedside aided swift and appropriate management of the patient. CONCLUSIONS: We envisage such a test having a role in assessing bed-bound patients in hospital where access to formal visual field testing is difficult, or indeed in rapid testing of visual fields at the bedside to screen for post-operative complications, such as haematoma.

PURPOSE: To investigate the safety and efficacy of subretinal injection of human Wharton's Jelly-derived mesenchymal stem cells (hWJ-MSCs) on retinal structure and function in Royal College of Surgeons (RCS) rats. METHODS: RCS rats were divided into 2 groups: hWJ-MSCs treated group (n = 8) and placebo control group (n = 8). In the treatment group, hWJ-MSCs from healthy donors were injected into the subretinal space in one eye of each rat at day 21. Control group received saline injection of the same volume. Additional 3 animals were injected with nanogold-labelled stem cells for in vivo tracking of cells localisation using a micro-computed tomography (microCT). Retinal function was assessed by electroretinography (ERG) 3 days before the injection and repeated at days 15, 30 and 70 after the injection. Eyes were collected at day 70 for histology, cellular and molecular studies. RESULTS: No retinal tumor formation was detected by histology during the study period. MicroCT scans showed that hWJ-MSCs stayed localised in the eye with no systemic migration. Transmission electron microscopy showed that nanogold-labelled cells were located within the subretinal space. Histology showed preservation of the outer nuclear layer (ONL) in the treated group but not in the control group. However, there were no significant differences in the ERG responses between the groups. Confocal microscopy showed evidence of hWJ-MSCs expressing markers for photoreceptor, Müller cells and bipolar cells. CONCLUSIONS: Subretinal injection of hWJ-MSCs delay the loss of the ONL in RCS rats. hWJ-MSCs appears to be safe and has potential to differentiate into retinal-like cells. The potential of this cell-based therapy for the treatment of retinal dystrophies warrants further studies.

PURPOSE: To assess the impact of VA loss on patient reported utilities taking both eyes into account compared to taking only the better or the worse eye into account. METHODS: In this cross-sectional study 1085 patients and 254 controls rated preferences with the generic health-related (EQ-5D; n = 868) and vision-specific (Vision and Quality of Life Index (VisQoL); n = 837) multi-attribute utility instruments (MAUIs). Utilities were calculated for three levels of VA in the better and worse eyes, as well as for 6 different vision states based on combinations of the better and worse eye VA. RESULTS: Using the VisQoL, utility scores decreased significantly with deteriorating vision in both the better and worse eyes when analysed separately. When stratified by the 6 vision states, VisQoL utilities decreased as VA declined in the worse eye despite stable VA in the better eye. Differences in VisQoL scores were statistically significant for cases where the better eye had no vision impairment and the worse seeing fellow eye had mild, moderate or severe vision impairment. In contrast, the EQ-5D failed to capture changes in better or worse eye VA, or any of the six vision states. CONCLUSIONS: Calculating utilities based only on better eye VA or using a generic MAUI is likely to underestimate the impact of vision impairment, particularly when the better eye has no or little VA loss and the worse eye is moderately to severely visually impaired. These findings have considerable implications for the assessment of overall visual impairment as well as economic evaluations within eye health.

Purpose: The photopic negative response (PhNR) of the light-adapted electroretinogram (ERG) holds promise as an objective marker of retinal ganglion cell function. We compared baseline detrending methods to improve PhNR repeatability without compromising its diagnostic ability in glaucoma. Methods: Photopic ERGs were recorded in 20 glaucoma and 18 age-matched control participants. A total of 50 brief, red-flashes (1.6 cd.s/m2) on a blue background (10 photopic cd/m2) were delivered using the RETeval device. Detrending methods compared were: (1) increasing the high-pass filter from 1 to 10 Hz and (2) estimating and removing the trend with an increasing polynomial (order from 1-10) applied to the prestimulus interval, prestimulus and postsignal interval, or the whole ERG signal. Coefficient of repeatability (COR%), unpaired Student's t-test, and area under the receiver operating characteristic curve (AUC) were used to compare the detrending methods. Results: Most detrending methods improved PhNR test-retest repeatability compared to the International Society for Clinical Electrophysiology of Vision (ISCEV) recommended 0.3 to 300 Hz band-pass filter (COR% ± 200%). In particular, detrending with a polynomial (order 3) applied to the whole signal performed the best (COR% ± 44%) while achieving similar diagnostic ability as ISCEV band-pass (AUC 0.74 vs. 0.75, respectively). However, over-correcting with higher orders of processing can cause waveform distortion and reduce diagnostic ability. Conclusions: Baseline detrending can improve the PhNR repeatability without compromising its clinical use in glaucoma. Further studies exploring more complex processing methods are encouraged. Translational Relevance: Baseline detrending can significantly improve the quality of the PhNR.

Purpose: To compare the RETeval sensor strip and Dawson-Trick-Litzkow (DTL) electrodes for recording the photopic negative response (PhNR) using a portable electroretinogram (ERG) device in eyes with and without glaucoma. Methods: Twenty-six control and 31 glaucoma or glaucoma-suspect participants were recruited. Photopic ERGs were recorded with sensor strip and DTL electrodes in random order using the LKC RETeval device. Stimuli consisted of brief, red flashes (1.7 cd.s/m2) on a blue background (photopic 10 cd/m2). The PhNR amplitude was measured from baseline to trough and also expressed as a ratio over the b-wave amplitude. Results: The sensor strip-recorded PhNR amplitude was significantly attenuated (mean ± standard deviation [SD], 4.8 ± 2.1 vs. 12.7 ± 4.8 μV, P < 0.0001), with lower signal-to-noise ratio (SNR; 5.5 ± 2.1 vs. 8.1 ± 3.9, P < 0.0001), and a trend toward a larger PhNR/b-wave ratio compared with DTL electrodes. The PhNR amplitude, implicit time and PhNR/b-wave ratio correlated with visual field mean light sensitivity, although this fell short of significance for the sensor strip recorded PhNR amplitude. The electrodes demonstrated similar intersession repeatability with a coefficient of repeatability of ±27% and ±28% for the DTL and sensor strip, respectively. Conclusions: Sensor strip electrodes are a viable alternative for recording reproducible PhNRs, especially when values are normalized to the b-wave. However, DTL electrodes should be considered in cases of attenuated PhNR, or in elevated noise levels, due to its better signal-to-noise quality. Translational Relevance: Sensor strip electrodes can simplify PhNR recordings in the clinic, potentially eliminating the need for an experienced operator.

Purpose: The purpose of this study was to investigate the effect of interstimulus frequency on the photopic negative response (PhNR) in the clinical electroretinogram (ERG) in glaucoma and healthy eyes. Methods: Participants with open angle glaucoma (n = 15) and age-matched controls (n = 20) were recruited. Photopic ERGs were recorded in one eye using five frequencies (1-5 Hz) delivered in random order. ERGs were analyzed for changes to amplitude and timing between groups and interstimulus frequency. Coefficient of variation (CoV) was used to examine variability within recordings for each frequency. Results: While the a-wave and b-wave showed minimal alteration, the PhNR was highly sensitive to changes in interstimulus frequency. The PhNR signal was largest at 1 Hz and steadily diminished with higher frequencies in both control and glaucoma groups. Significant differences in PhNR amplitude were found between controls and glaucoma groups at 2 and 3 Hz. While 1 Hz delivered the largest PhNR, it also showed a significantly greater CoV compared to other frequencies. Conclusions: An interstimulus frequency of 2 Hz was optimal for recording the PhNR, creating a good balance between testing time and signal quality. A higher frequency could be used to further shorten clinical testing times; however, this may compromise its clinical utility by dampening the PhNR. Translational Relevance: Here we show the importance of considering flash interstimulus frequency when designing ERG protocols for recording the PhNR as while higher frequencies can shorten test times, they also have considerable effects on the PhNR.

PURPOSE: To determine the measure of the photopic negative response (PhNR) of the full-field electroretinogram (ERG) that exhibits the optimal level of test-retest repeatability, and examine its repeatability under different conditions using a handheld, nonmydriatic ERG system and self-adhering skin electrodes. METHODS: Multiple ERG recordings (using 200 sweeps each) were performed in both eyes of 20 normal participants at two different sessions to compare its coefficient of repeatability (CoR; where 95% of the test-retest difference is expected to lie) between different PhNR measures and under different testing conditions (within and between examiners, and between sessions). RESULTS: The ratio between the PhNR trough to b-wave peak and b-wave peak to a-wave trough amplitude (PhNR/B ratio) exhibited the lowest CoR relative to its effective dynamic range (30 ± 4%) when including three recordings. There were no significant changes in the PhNR/B ratio over seven measurements (4 right and 3 left eyes) at either session (P ≥ 0.100), or significant difference in its CoR between different testing conditions (P = 0.314). CONCLUSION: The PhNR/B ratio was the measure that minimized variability, and its measurements using a novel handheld ERG system with self-adhering skin electrodes and the protocols described in this study were comparable under different testing conditions and over multiple recordings. TRANSLATIONAL RELEVANCE: The PhNR can be measured for clinical and research purposes using a simple-to-implement technique that is consistent within and between visits, and also between examiners.

Studies of rodent models of Alzheimer's disease (AD) and of human tissues suggest that the retinal changes that occur in AD, including the accumulation of amyloid beta (Aβ), may serve as surrogate markers of brain Aβ levels. As Aβ has a wavelength-dependent effect on light scatter, we investigate the potential for in vivo retinal hyperspectral imaging to serve as a biomarker of brain Aβ. Significant differences in the retinal reflectance spectra are found between individuals with high Aβ burden on brain PET imaging and mild cognitive impairment (n = 15), and age-matched PET-negative controls (n = 20). Retinal imaging scores are correlated with brain Aβ loads. The findings are validated in an independent cohort, using a second hyperspectral camera. A similar spectral difference is found between control and 5xFAD transgenic mice that accumulate Aβ in the brain and retina. These findings indicate that retinal hyperspectral imaging may predict brain Aβ load.

Retinal ganglion cell degeneration underlies several conditions which give rise to significant visual compromise, including glaucoma, hereditary optic neuropathies, ischaemic optic neuropathies, and demyelinating disease. In this review, we discuss the emerging strategies for neuroprotection specifically in the context of glaucoma, including pharmacological neuroprotection, mesenchymal stem cells, and gene therapy approaches. We highlight potential pitfalls that need to be considered when developing these strategies and outline future directions, including the prospects for clinical trials.