Clinicopathological studies have demonstrated that the cornerstones of clinical vision testing, acuity and perimetry, may fail to identify substantial losses of retinal receptive fields.1–3 This tolerance to damage may, at least in part, be attributable to the use of test targets that are many times larger than the receptive fields. Test targets that match receptive field sizes may provide better sensitivity. Several studies have attested to the effectiveness of this so-called rarebit approach, which can be applied in central and peripheral vision alike and offers a unified metric of the state of the neuro-visual system.4 The rarebit term refers to the minuscule information content of the test targets.
The first rarebit test had the format of rarebit perimetry (RBP), with an inbuilt rarebit fovea test.4 The rarebit fovea test may have the potential to meet the well-documented need for sensitive tests for macular disorders,5–8 but the original computer screen set-up was somewhat impractical outside laboratory settings. A compact, self-contained implementation, provisionally named the MacuBit test (MBT), has recently been developed for practicable screening of the central-most visual field, particularly for conditions directly involving the macular receptive fields.9 The MBT has several potential advantages over conventional acuity tests, including rigorously fixed test conditions, a larger test area, and a presentation time that is brief enough to eliminate potentially misleading gains from scanning eye movements. The MBT also provides an intuitive grading of the severity of any impairment.
To fully illuminate the MBT's diagnostic potential, it is necessary to test other types of neural matrix disorders, including disorders characterized by upstream disconnections of receptive fields. The present study examined MBT's performance with non-glaucomatous optic neuropathies and chiasmal lesions of light to moderate severities, and compared results with those of two clinical mainstay tests, acuity and automated perimetry.
Twenty-two consecutive patients (mean age 45 years ± 14 (SD), range: 21 to 67 years) were recruited from the author's neuro-ophthalmology consultation service at Sahlgrenska university hospital, a tertiary-care center, on the basis of symptoms, signs, and neuro-imaging results indicating compression of the optic nerve (seven instances) or the chiasm (seven instances), resolved optic neuritis (seven instances), or anterior ischemic optic neuropathy (1 instance). Subjects with bilateral disease contributed results from the least involved eye only. Exclusion criteria were a decimal visual acuity <0.5 and the presence of additional, non-neurological causes of visual loss. All were naïve to rarebit testing. All the participants gave informed consent. The examinations adhered to the tenets of the Declaration of Helsinki. The study was approved by the institutional review board.
The MBT device has been described in detail.9 In brief, it used an 800 × 600 Digital Light Projection matrix with a 16 μm mirror pitch, illuminated by light-emitting diodes (Mitsubishi PK10, Irvine, CA). The display was viewed at 120 mm distance through an ocular fitted with appropriate light-attenuating filters. A faintly luminous peripheral ring aided orientation: there were no specific fixation demands.
The test display was controlled by a laptop computer running at the same resolution, using purpose-written software to present identical outputs on the computer screen and in the presentation unit. The latter was provided with an internal mask that limited the subject's field of view to a circular area with a diameter of 500 pixels, subtending 5.3°. The mask prevented the subject from seeing the operator's controls and the results statistics.
The test targets subtended 0.6′ at the eye. Background and target luminances were set to 5 and 200 cd/m2, respectively. Test targets were presented for 200 ms, in ever-new, randomly selected locations. Each presentation was cued by sound. Subjects were informed that each presentation involved one or two bright dots, or sometimes none at all, and they were asked to indicate the number of dots seen by clicking the computer mouse, once, twice, or not at all. Visual and auditory feedback was provided after each presentation. To make the subject feel to be in command of the procedure, the interval between presentations was automatically adapted to the current reaction time. In the interest of saving on test time, most presentations comprised pairs of test dots, with a horizontal or vertical separation of 30′. Some 10% of the presentations (selected at random in space and time) comprised one test dot only, or none at all, to allow checking for false responses.
After a brief initial training period, responses were collected for a total of 50 dot pair presentations. Results were expressed as hit rates, i.e., the percentage of targets seen relative to the number of targets shown. Responses to control presentations were recorded separately and were not included in the hit rate index. In line with its screening profile, the MBT dispenses with topographic maps.
Because normal eyes hold seamless and non-overlapping arrays of receptive fields, the expected result is a hit rate near 100% (see further). Eyes that have lost receptive fields or upstream connections should score poorer in proportion to the loss of neural elements.4 Here, previously reported MBT (and acuity) results from 49 normal subjects served as reference.9 These subjects were recruited from among outpatient referrals to the eye clinic on the basis of age ≥50 years and normal results in at least one eye on a comprehensive ophthalmological examination. The latter included refraction, biomicroscopy of the anterior segment, direct ophthalmoscopy of the posterior pole, and applanation tonometry. Each subject contributed results from one eye only. Unless found abnormal in the clinical examination, the right eye was selected. Mean age was 60 years ± 8, and range was 50 to 78 years.
Visual acuity was assessed, as previously described, in a study of normal thresholds over a wide age-range.10 In brief, the test used a non-translucent chart with 10 letters per line and a line step factor of 0.1 log units. Test distance was 4 m, and background luminance was 200 cd/m2. The percentage of correctly read letters was recorded for each line. Probit analysis provided 50% correct thresholds in the decimal format.
High-Pass Resolution Perimetry
High-pass resolution perimetry (HRP) (ver. 3; HighTech Vision, Gothenburg, Sweden;) is a 5 min computer-graphic test using so-called vanishing resolution targets. Vanishing targets derive their name from the near coincidence of detection and resolution thresholds. HRP uses ring-shaped vanishing targets of varying sizes; contrast is fixed. HRP thresholds are directly proportional to grating and Landolt C thresholds (but not to thresholds of differential light sensitivity), and the HRP thresholds are directly proportional to ganglion cell densities.11,12
HRP tests 50 locations between 4° and 28° of eccentricity. Clinically, HRP results are known to be closely comparable with conventional, differential light sensitivity perimetry in optic neuropathies and chiasmal syndromes.13,14
Among the several automatically provided statistical indices, the age-adjusted neural capacity summary index was used here. It is held to reflect the number of functional neuro-visual channels relative to normal. The normal average within HRP's internal database is 100%. Here, calculations of sensitivity and specificity required access to individual normal results. These were drawn from a previous report.15
Statistical analyses used two-sided Wilcoxon signed rank tests and Spearman rank correlation coefficients. P values <0.05 were considered statistically significant.
The 22 patients presented a wide variety of visual field defects, representative of the various underlying conditions. Their HRP neural capacity summary indices are shown along the horizontal axis of Fig. 1. The distribution was significantly different from that of the normal control subjects (p = 0.0007). The threshold acuity tests showed similar outcomes (Fig. 3, p = 0.006).
All subjects completed the MBT without difficulty. Their hit rates are shown along the vertical axes of Figs. 1 and 2. The hit rates ranged from 1 to 100%, with a median value of 69%. The distribution was significantly different from that of the normal control subjects (p = 0.007). The control subjects were somewhat older [mean 60 years ± 8 vs. 45 years ± 14]. In normal subjects, hit rates appear to begin to decline around the age of 40 years, at a rate of about 0.4% units per year.9 Corrections for age were not made here because they simply would have increased the statistical significance.
MBT hit rates showed no meaningful correlations with either HRP or acuity results (rank correlation coefficients, 0.25 and 0.43, respectively). The same was true for HRP vs. acuity (0.12).
A broader view of the diagnostic performance of the three tests can be obtained by plotting their true positive rates (sensitivity) against false positive rates (1-specificity), using ranked observations. The resulting receiver-operating characteristic curves16 indicate closely similar performances (Fig. 2). The relatively small number of subjects argues against a detailed analysis.
MBT test times averaged 102 ± 21 s. The modal number of errors was 0, indicating reliable performance. HRP and acuity thresholding required approximately 5 min each.
The most remarkable outcome of the present study may be the poor correlations between the results of the three tests (Figs. 1 and 3). Even the two resolution tests, acuity and HRP, presented a rank correlation coefficient no larger than 0.12. For these two tests, the explanation presumably relates to the spatial distribution of functional deficits across the visual field: central vision may be spared in the presence of non-central visual field defects and vice versa.
The explanation for the poor correlations between the MBT and the resolution tests presumably relates to several factors. Resolution tests essentially reflect the spatial density of functional receptive fields and their upstream cortical connections.1,11 Rarebit tests, on the other hand, essentially probe the receptive field matrix (and its upstream connections) for functional locations and return a proportion, i.e., a non-dimensional number. Another important difference concerns the information content of the test targets. Laboratory studies using optotype letters broken down into elements have shown that substantial proportions of elements can be removed without affecting resolution.17,18 By design, the rarebits hold a minimum of information, and they should in principle lack redundancy. On the other hand, their small size carries a risk of interception by retinal vessels, with the production of angioscotomata. Vascular anatomy indicates that the risk should be small when testing is restricted to the macular area.
Irrespective of the different test principles and the poor inter-test correlations, the receiver-operating characteristic curves (Fig. 3) indicate closely similar powers of discrimination between normal and abnormal subjects.
Previous experience with the perimetry implementation of rarebit testing, RBP, indicates that the latter generally presents more extensive defects than does HRP.4,15 In this light, the MBT would have been expected to provide the best power of discrimination. Its failure to do so may relate to spatial distributions of neural matrix defects. The MBT distributes the test targets at random across the 5.3° diameter test area. The random allocation results in a fairly uniform distribution across the test area, which appears suitable for macular disorders. The situation may be different with non-glaucomatous optic neuropathies and chiasmal lesions, where the thinnest axons, which subserve the central-most receptive fields, classically are held to be the most vulnerable. Hence, the disconnections of receptive fields that occur with these disorders may be more pronounced in the central-most test area and may taper off toward its edge.
The possible advantage of estimating the central-most hit rate was not recognized until near the completion of the present study. At that time, the software was modified to allow separate calculations of hit rates within the central 1.2° of the test field and the surrounding ring-shaped area. Results from seven subjects could be analyzed in this way, and they revealed a significantly lower mean hit rate within the central area compared with the surrounding area, 51 ± 18 vs. 65 ± 15 (p = 0.043). This result indicates that concentration on central-area hit rates should raise MBT's discriminative power for disorders characterized by disconnection of receptive fields. For disorders directly involving receptive fields, for example, macular edema or degeneration, combined inner and outer hit rates may constitute the better index. Indeed, in an on-going independent study of early-stage age-related macula degeneration, using the MBT, the combined indices appear to provide excellent sensitivity (FD Verbraak, MD, personal communication, 2012). These observations indicate that the use of both indices may aid differential diagnosis.
Previous research using RBP in normal subjects has indicated a good reproducibility.4 The MBT has been shown to have excellent short-term reproducibility in subjects with advanced age-related macula degeneration.9 Long-term reproducibility remains to be explored.
Blue Street 7:5, SU/S
SE-413 45 Gothenburg
Supported by the University of Gothenburg. The author thanks Ms. Rose-Marie Rang for providing expert technical assistance. The author holds patents on the MBT device.
1. Frisén L, Quigley HA. Visual acuity in optic atrophy: a quantitative clinicopathological analysis. Graefes Arch Clin Exp Ophthalmol 1984;222:71–4.
2. Quigley HA, Dunkelberger GR, Green WR. Retinal ganglion cell atrophy correlated with automated perimetry in human eyes with glaucoma. Am J Ophthalmol 1989;107:453–64.
3. Kerrigan-Baumrind LA, Quigley HA, Pease ME, Kerrigan DF, Mitchell RS. Number of ganglion cells in glaucoma eyes compared with threshold visual field tests in the same persons. Invest Ophthalmol Vis Sci 2000;41:741–8.
4. Frisén L. New, sensitive window on abnormal spatial vision: rarebit probing. Vision Res 2002;42:1931–9.
5. Nilsson M, von Wendt G, Wanger P, Martin L. Early detection of macular changes in patients with diabetes using rarebit fovea test and optical coherence tomography. Br J Ophthalmol 2007;91:1596–8.
6. Nowomiejska K, Oleszczuk A, Zubilewicz A, Krukowski J, Mankowska A, Rejdak R, Zagorski Z. Assessment of the macula function by static perimetry, microperimetry and rarebit perimetry in patients suffering from dry age related macular degeneration [in Polish]. Klin Oczna 2007;109:131–4.
7. Nilsson M, von Wendt G, Brautaset R, Wanger P, Martin L. Macular structure and function and the development of retinopathy in diabetes. Clin Exp Optom 2011;95:306–10.
8. van Dijk HW, Verbraak FD, Stehouwer M, Kok PH, Garvin MK, Sonka M, DeVries JH, Schlingemann RO, Abramoff MD. Association of visual function and ganglion cell layer thickness in patients with diabetes mellitus type 1 and no or minimal diabetic retinopathy. Vision Res 2011;51:224–8.
9. Winther C, Frisén L. A compact rarebit test for macular diseases. Br J Ophthalmol 2010;94:324–7.
10. Frisén L, Frisén M. How good is normal visual acuity? A study of letter acuity thresholds as a function of age. Albrecht Von Graefes Arch Klin Exp Ophthalmol 1981;215:149–57.
11. Frisén L, Nikolajeff F. Properties of high-pass resolution perimetry targets. Acta Ophthalmol (Copenh) 1993;71:320–6.
12. Popovic Z, Sjöstrand J. The relation between resolution measurements and numbers of retinal ganglion cells in the same human subjects. Vision Res 2005;45:2331–8.
13. Wall M. High-pass resolution perimetry in optic neuritis. Invest Ophthalmol Vis Sci 1991;32:2525–9.
14. Dannheim F, Roggenbuck C. Comparison of automated conventional and spatial resolution perimetry in chiasmal lesions. In: Heijl A, ed. Proceedings of the VIIIth International Perimetric Society Meeting, Vancouver (Canada), May 9–12, 1988. Amsterdam, The Netherland: Kugler & Ghedini; 1989:377–82. Perimetry update, 1988–9.
15. Frisén L. Spatial vision in visually asymptomatic subjects at high risk for multiple sclerosis. J Neurol Neurosurg Psychiatry 2003;74:1145–7.
16. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993;39:561–77.
17. Seiple W, Holopigian K, Szlyk JP, Greenstein VC. The effects of random element loss on letter identification: implications for visual acuity loss in patients with retinitis pigmentosa. Vision Res 1995;35:2057–66.
18. Frisén L. Scope of super-resolution in central vision. Br J Ophthalmol 2010;94:97–100.
vision test; rarebit; visual acuity; optic nerve; chiasm