Standard deviations for sets of 10 repeats were pooled and are plotted in Fig. 9. They are calculated from the logarithmic spatial frequencies and are given in log units. Standard deviations decrease with increasing contrast and average between 0.11 and 0.05 log units. Moreover, except for a 10% contrast, the standard deviations for rotating stars were lower than the values for stationary stars in the ring adjustment mode.
The means of the first 5 presentations of each trial were compared with the means of all 10 presentations for this mode as well. In 87% of all trials, the mean of 5 presentations deviated by <0.05 log units from the mean of 10 presentations. As before, no clear trend in either direction was noted.
The average duration of the trial for 1 contrast in the ring adjustment mode was 46.5 s for both stationary and rotating patterns (10 presentations).
In this experiment, the validity of the Star-Ring Test was investigated by comparison with the Vistech chart. Fig. 10 shows the averages of the logCS estimated by the Star-Ring Test (contrast adjustment mode) and by the Vistech chart. For both test methods, the peak spatial frequency of the CSF is 3 cpd. With higher spatial frequencies, the CSF shows the typical decrease. In the range between 1.5 cpd and 3 cpd, the CSF estimated by the Star-Ring Test is flatter than that estimated by the Vistech chart (Fig. 10).
Mean logarithmized contrast sensitivities of both tests and the significances (p values) calculated by the Wilcoxon signed-ranks test are listed in Table 4. For spatial frequencies of 3, 6, and 12 cpd, contrast sensitivities estimated by the Star-Ring Test are significantly lower (difference averages 0.19 to 0.40 logCS) than those estimated by the Vistech chart. For 1.5 and 18 cpd, there is no significant difference.
The RCs for the Star-Ring Test ranged from RC = 0.22 log units to RC = 0.38 log units, depending on the tested spatial frequency. For the Vistech chart it ranged from RC = 0.34 log units to RC = 0.67 log units (Table 5). The average duration of one trial in the contrast adjustment mode of the Star-Ring Test was 67 s. For the Vistech chart, the measurement (all lines) took 101 s on average, which means 20.2 s for each spatial frequency.
In the ring adjustment mode of the Star-Ring Test, the geometric mean of the adjusted spatial frequency at 10% contrast was 24.87 cpd. The RC for rotating Siemens stars was RC = 0.13 log units. The average duration of one trial in the ring adjustment mode was 81 s, whereas the fastest trial took 29 s and the slowest 172 s.
In this experiment, three subjects with abnormal contrast sensitivity were tested with the Star-Ring Test (contrast adjustment mode) and the Vistech chart. Estimated contrast sensitivities are plotted in Fig. 11. It is shown that contrast sensitivities are reduced for all spatial frequencies in both test methods, compared with averaged contrast sensitivities shown in experiment 2. The reduction ranged from 0.28 to 0.94 log units (Star-Ring Test) and from 0.48 to 1.32 log units (Vistech chart). Subject 3 could not perceive any pattern on the Vistech chart at spatial frequencies of 12 cpd and 18 cpd.
Measurements of contrast sensitivity by means of chart-based tests usually require the subject to read aloud the optotypes seen. This may be undesirable in certain circumstances, for example, when the head of the subject is fixed for experimental purposes by means of a forehead and chin rest to look at the test pattern through an optical system in an exactly centered position. No verbal response is required in the determination of contrast sensitivity using the Siemens star. The responses are made by pressing buttons, which means head motions are eliminated during the examination.
The Star-Ring Test is performed according to the method of adjustment. This makes the test rapid but also renders the measuring result susceptible to the influence of the subjective criterion. A motivated subject may tend to adjust the contrast threshold more likely too low, whereas a cautious subject may tend to set to a position where he/she is absolutely sure to recognize the pattern. Accordingly, the test is best suited for measurements of progression or side-by-side comparison of two viewing situations (e.g., contrast sensitivity in the presence/absence of contact lenses) presuming the subjective judgment to be unchanged. Results concerning the RC of the Star-Ring Test (see experiment 2) support this statement.
Defining the threshold of recognition using the Star-Ring Test is a challenging visual task. Rotating stars are experienced to be more pleasant. Subjectively, this makes the adjustment of contrast or ring size easier. Influences of motion perception due to the rotation of the test pattern vs. the stationary pattern do not appear to play a significant role (Figs. 6 and 8). No significant differences were detected in any comparison in experiment 1. The measured values show little deviation for rotating patterns (Figs. 7 and 9). Another advantage is that local adaptation is prevented and interfering after-images are reduced.
In experiment 1 in the ring adjustment mode, the standard deviations of the respective presentations were considerably larger at low contrast (contrast <5%, Fig. 9). This may be related to the smaller slope of the CSF in this range. It therefore makes sense to use the contrast adjustment mode in which the ring sizes; hence, the spatial frequencies are given, and the contrast threshold is adjusted. Conversely, it is conceivable that the deviation increases markedly at high spatial frequencies in contrast adjustment mode because the CSF steepens increasingly in this range. (This, however, will have to be addressed in a subsequent study.) Presumably, a combination of the two modes would be useful, in which contrast adjustment mode is used for low and moderate spatial frequencies while ring adjustment mode produces more accurate results for high spatial frequencies.
Experiment 2 contained a comparison of the Star-Ring Test with a common test method, the Vistech chart. With all spatial frequencies except 1.5 cpd, the Vistech chart showed higher mean contrast sensitivities than the Star-Ring Test (Fig. 10). For 3, 6, and 12 cpd, the differences are statistically significant (Table 4). The mean CSF of the Star-Ring Test consequently is flatter. A possible reason for this is the relatively high chance level for the Vistech chart. It is a three alternative forced choice method, and so the chance level is 33% for one contrast step. This means that a test subject could give the correct answer by chance for two consecutive contrast steps with an 11% probability. The Star-Ring Test uses five decisions per contrast step and is not as susceptible to chance. Therefore, contrast sensitivity measured by the Vistech chart could be overestimated. It is also conceivable that the print quality of the Vistech chart affects the results. The contrasts predetermined by the manufacturer were approximately complied with, as detected by the LMK 98. However, the contrast across one test pattern is often not stable and can be slightly higher than intended in several areas. This principally can lead to an overestimation of contrast sensitivity in all spatial frequencies.
Another reason for differences of contrast sensitivity values could be the task of the Star-Ring Test. The contrast ought to be reduced, until there is no pattern visible inside the ring. However, subjects may tend to stop contrast reduction a bit before having reached the threshold. Thus, contrast sensitivity would be underestimated. However, at spatial frequencies 1.5 and 18 cpd, no significant differences were found comparing the Vistech chart to the Star-Ring Test. For 1.5 cpd, this is explainable with the flatter form of the CSF of the Star-Ring Test between 1.5 and 3 cpd (Fig. 10). The reason for this might again be the task. The pattern inside the ring (i.e., higher spatial frequencies) ought to be invisible. As the CSF decreases to the left, spatial frequencies lower than 1.5 cpd outside the ring are not visible either. Theoretically, a zonular part of the test pattern inside the ring comprising the maximum of the CSF ought to be visible. In practice for most subjects it is not. In doing so, the typical decrease from the maximum of the CSF to the lower spatial frequencies—called low spatial frequency rolloff—does not exist or at least is much flatter. Thus, contrast sensitivities of both tests converge.
The results for 18 cpd also show almost identical values in both tests. Using the Star-Ring Test at 3 m viewing distance for measuring 18 cpd, one sinusoidal period is displayed by 10 pixels. This means that the luminance range of one period is resolved into 10 increments, which appears to be adequate for a proper sinusoidal design. Compared with the Star-Ring Test, Vistech chart results are higher in the midrange spatial frequencies, and there is a stronger decay at the higher spatial frequency end of the CSF. This might indicate that the aforementioned typographic deficiencies influence results especially in high spatial frequencies.
The peak of the mean CSF was determined at 3 cpd (both by the Star-Ring Test and by the Vistech chart) in experiment 2. In experiment 1, it was identified at 6 cpd by the Star-Ring Test. The peak of the standard CSF is located in between these two spatial frequencies, as demonstrated by Souza et al.30
There are only marginal differences between the final results of 5 vs. 10 presentations, as shown in experiment 1. The test routine therefore was down sized to 5 presentations per spatial frequency in experiments 2 and 3. This allows for a reduction of measurement periods and subjective strain, especially if a large number of trials is needed.
In experiment 2, comparison of the Star-Ring Test and Vistech chart shows a better repeatability of the Star-Ring Test. RC of Star-Ring Test ranges from 0.22 to 0.38 log units, whereas RC of the Vistech chart ranges from 0.34 to 0.67 log units (Table 5). These values are consistent with previous studies. Pesudovs et al.,31 for instance, determined RC values from 0.30 to 0.85 log units for the Vistech chart. A possible reason for the poor reliability of the Vistech chart is the use of large contrast increments (averaging ∼0.25 log units), single trials at each contrast level,32 and a high level of chance, as discussed above. For the FACT, RC values from 0.30 to 0.75 log units have been reported.31 Letter tests such as the Pelli-Robson chart (RC value from 0.18 to 0.20 log units)6,33,34 and the Mars Letter Contrast Sensitivity Test (RC value from 0.121 to 0.20 log units)6,33,34 show better repeatability.
As mentioned above, there is an influence of the subjective criterion on the results of the Star-Ring Test. Performing repeated measurements or measurements of progression, it can be assumed that subject's criterion remains unchanged. It appears to have been so in this experiment; otherwise RC values would have been negatively impacted. For the repeatability of the Star-Ring Test, it is beneficial to adjust the contrast several times, minimizing deviations from incorrect measurements. Moreover, the result is not subject to the influence of level of chance. The ring adjustment mode also was found to be highly repeatable (RC = 0.13 log units).
Comparing the repeatabilities of both modes of the Star-Ring Test, RC in the ring adjustment mode is considerably smaller (0.13 log units) than in the contrast adjustment mode (0.22 to 0.38 log units). As discussed above, it must be noted that different ranges of the CSF were examined. On the one hand, the ring adjustment mode was used to determine resolvable spatial frequency in the steep range of the CSF (for a 10% contrast). On the other hand, contrast adjustment mode was applied to examine contrast sensitivities in the middle range of the CSF (3 cpd to 24 cpd). If the ring adjustment mode is used with smaller contrasts (1.25% and 2.5% in experiment 1) larger standard deviations for 10 repeats will be detected than with higher contrasts (Fig. 9). Thus, for the subjects it appears to be more difficult to mark the limit of resolution with a low contrast of the test pattern. Accordingly, the RC values are higher in this case. Therefore, the ring adjustment mode may not always be the better choice; however, in the range of high spatial frequencies, it appears to be more advantageous compared with the contrast adjustment mode.
A comparison of measurement periods in the 2 modes in experiment 1 shows that ring adjustment requires approximately half as much time (46.5 s) as contrast adjustment (85 s for rotating and 94 s for stationary stars). Adjustment of the ring size appears to be the task more easily carried out by the subjects, each of whom had experience in psychophysical testing. As expected, the naïve subjects in experiment 2 needed more time for a trial. In the ring adjustment mode, they needed even considerably longer (81 s in average), although, in contrast to experiment 1, only five presentations per trial were presented. Some subjects performed this test with extreme care (this considered, a time limit might be useful). However, a positive effect of that is the very good repeatability.
In experiment 3, contrast vision of three elderly subjects with incipient cataract was investigated. A reduction of contrast sensitivity was detectable with both test methods, the Star-Ring Test and the Vistech chart. The reduction was caused by the opacity of the lens on one hand, but by normal age-related changes of contrast sensitivity on the other hand. With the Vistech chart, the measured loss of contrast sensitivity tends to be larger than with Star-Ring Test, compared with the CSFs in experiment 2. A possible reason might be that—although the Vistech chart requires a forced choice mode—the elderly subjects did not risk guessing and chose to abort a line when no clear pattern was visible. This may lead to an underestimation of contrast sensitivity. However, with the young healthy subjects in experiments 1 and 2, the forced-choice procedure worked well. Subject 3 did not recognize even the highest contrasts of the Vistech chart at spatial frequencies 12 and 18 cpd. With the Star-Ring Test, contrast thresholds for all five spatial frequencies were determined, as contrast can be adjusted up to ∼90%, depending on the monitor.
Comparability of the Star-Ring Test and the Vistech chart is limited to some degree because of the different test design. Vistech uses discrete steps, whereas the Star-Ring Test uses a continuously variable measure. Thus, direct comparison of contrast sensitivity and measurement error between the 2 tests must be regarded with caution. In a further study, it would be reasonable to test the performance of the Star-Ring Test against a monitor grating contrast sensitivity test without discrete steps, which is commonly used in vision-research laboratories.
It is a general advantage of the Star-Ring Test that contrast is variable in a wide range, unlike Vistech or FACT, which have a limited contrast spread. Another advantage is the variability of the test distance. It is customizable in the settings of the software and thus feasible for different test setups. Printed charts are subject to wear and tear, which might lead to a change of contrast. It is not so with a computer monitor. Initial contrast calibration and occasional recalibration maintains constant quality.
Letter-based contrast tests such as Pelli-Robson-Chart or MARS-Letters are very easy for subjects to understand because reading letters is a very common task. The results provide a notion about contrast sensitivity in everyday situations. Nevertheless, letters do not have a defined spatial frequency. Those tests are not feasible for measuring the entire CSF. The Vistech chart evaluates contrast vision with five defined spatial frequencies. A best fit CSF can be derived. On the other hand, repeatability is limited and the level of chance is high. The Star-Ring Test is appropriate for measuring different spatial frequencies with a very low level of chance and significantly higher repeatability.
In the Star-Ring Test, a black, sharply delimited ring is used to mark the resolution threshold. However, the high contrast of the ring might lead to some interference with the underlying spokes of the star and thus have an impact on the measuring result. It is conceivable that this is partly responsible for the smaller contrast sensitivity values in middle spatial frequency range compared with the Vistech chart. In the contrast adjustment mode, ring diameter in the visual field varies between 0.52° and 1.05° at a test distance of 6 m and between 1.05° and 2.1° at 3 m, respectively. Although the position on the monitor changes, subjects use about the same area on the retina to perceive it. Thus, an effect of ring size on the adjusted contrast threshold is improbable.
An interesting phenomenon called spurious resolution occurs when a subject with incorrect correction views the Siemens star. It causes light stripes to be perceived as dark stripes and vice versa in certain ranges of spatial frequencies. The pattern appears phase-inverted (Fig. 12). This is caused by defocusing which refracts the light to the extent that neighboring brightness peaks, superimposing on an originally dark area, and vice versa.35 This phenomenon occurs with all periodical patterns. Therefore, the test is suitable for contrast testing with good correction, because contrast sensitivity may otherwise be over-assessed.
Because of the given gray levels of the monitor (256 gray levels), the display of contrast is limited. Michelson contrasts of <0.4% cannot be displayed (depending on the monitor). In some cases, contrast threshold may be lower and detract from exact determination. The Star-Ring Test does not use discrete contrast value graduations, and the repeatability was demonstrated to be high. As a result, the test is capable of detecting even subtle changes of contrast sensitivity.
The Star-Ring Test is a sensitive contrast vision test producing reliable results in little time, which renders the test convenient for laboratory investigations. It is easy to understand, and the interactive mode ensures that attention level of the subjects is high. This makes it feasible even for eye care practitioners. The CSF derived from a Star-Ring Test procedure is very similar to that of the Vistech chart. The tests allow for a non-verbal examination. Initial results and experiences obtained with the Star-Ring Test are promising. The new test method appears to be also useful for elderly subjects with eye diseases.
We thank Halle Eye Laser Centre for support in recruitment of subjects with eye diseases. This study was supported by German Federal Ministry of Education and Research (BMBF), grant 01EZ0608.
Course of Optometry
University of Applied Sciences Luebeck
Moenkhofer Weg 239, 23562 Luebeck
1. Elliott DB. Contrast sensitivity and glare testing. In: Benjamin WJ, Borish IM, eds. Borish's Clinical Refraction. Philadelphia: W.B. Saunders; 1998:203–41.
2. Leguire LE. Do letter charts measure contrast sensitivity? Clin Vis Sci 1991;6:391–400.
3. Pelli DG, Robson JG, Wilkins AJ. The design of a new letter chart for measuring contrast sensitivity. Clin Vis Sci 1988;2:187–99.
4. Lovie-Kitchin JE. Validity and reliability of visual acuity measurements. Ophthalmic Physiol Opt 1988;8:363–70.
5. Arditi A. Improving the design of the letter contrast sensitivity test. Invest Ophthalmol Vis Sci 2005;46:2225–9.
6. Dougherty BE, Flom RE, Bullimore MA. An evaluation of the Mars Letter Contrast Sensitivity Test. Optom Vis Sci 2005;82:970–5.
7. Ginsburg AP. Next generation contrast sensitivity testing. In: Rosenthal B, Cole R, eds. Functional Assessment of Low Vision. St. Louis: Mosby; 1996:77–88.
8. Pelli DG, Robson JG. Are letters better than gratings? Clin Vis Sci 1991;6:409–11.
9. deLange H. Research into the dynamic nature of the human fovea-cortex systems with intermittent and modulated light. 1. Attenuation characteristics with white and colored lights. J Opt Soc Am 1958;48:777–84.
10. Campbell FW, Green DG. Optical and retinal factors affecting visual resolution. J Physiol 1965;181:576–93.
11. Bodis-Wollner I. Visual acuity and contrast sensitivity in patients with cerebral lesions. Science 1972;178:769–71.
12. Arden GB, Jacobson JJ. A simple grating test for contrast sensitivity: preliminary results indicate value in screening for glaucoma. Invest Ophthalmol Vis Sci 1978;17:23–32.
13. Ginsburg AP. A new contrast sensitivity vision test chart. Am J Optom Physiol Opt 1984;61:403–7.
14. Simas ML, Santos NA. Contrast sensitivity to radial frequencies modulated by Jn and jn Bessel profiles. Braz J Med Biol Res 2002;35:1357–67.
15. Simas ML, Nogueira RM, Santos NA. Radial frequency stimuli and sine-wave gratings seem to be processed by distinct contrast brain mechanisms. Braz J Med Biol Res 2005;38:419–30.
16. Santos NA, Oliveira AB, Nogueira RM, Simas ML. Mesopic radial frequency contrast sensitivity function for young and older adults. Braz J Med Biol Res 2006;39:791–4.
17. Baehr HP, Voegtle T. Digitale Bildverarbeitung: Anwendung in Photogrammetrie, Kartographie und Fernerkundung, 3rd ed. Heidelberg: Wichmann; 1998.
18. Koest G. Apparatus and method for ocular refractometry. European Patent EP1437085. July 14, 2004.
19. Simas ML, Santos NA, Thiers FA. Contrast sensitivity to angular frequency stimuli is higher than that for sinewave gratings in the respective middle range. Braz J Med Biol Res 1997;30:633–6.
20. Santos NA, Simas ML, Nogueira RM. Comparison of angular frequency contrast sensitivity in young and older adults. Braz J Med Biol Res 2004;37:375–8.
21. Jungnickel H, Strohm K, Gebhardt M, Grein HJ. Evaluation of a modified Siemens star for accurate measuring of contrast sensitivity. Invest Ophthalmol Vis Sci 2008;49:E-Abstract 2534.
22. Brainard DH, Pelli DG, Robson T. Display characterization. In: Hornak J, ed. Encyclopedia of Imaging Science and Technology. New York: John Wiley; 2002:172–88.
24. Larsson E, Rydberg A, Holmstrom G. Contrast sensitivity in 10 year old preterm and full term children: a population based study. Br J Ophthalmol 2006;90:87–90.
25. Rajagopalan AS, Bennett ES, Lakshminarayanan V. Visual performance of subjects wearing presbyopic contact lenses. Optom Vis Sci 2006;83:611–5.
26. Chung SH, Lee SJ, Lee HK, Seo KY, Kim EK. Changes in higher order aberrations and contrast sensitivity after implantation of a phakic artisan intraocular lens. Ophthalmologica 2007;221:167–72.
27. Chew HF, Markowitz SN, Flanagan J, Buys YM. The effect of pupil dilation on driving vision in Canada. Can J Ophthalmol 2007;42:585–91.
28. Igarashi A, Kamiya K, Shimizu K, Komatsu M. Visual performance after implantable collamer lens implantation and wavefront-guided laser in situ keratomileusis for high myopia. Am J Ophthalmol 2009;148:164–70e1.
29. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.
30. Souza GS, Gomes BD, Saito CA, da Silva Filho M, Silveira LC. Spatial luminance contrast sensitivity measured with transient VEP: comparison with psychophysics and evidence of multiple mechanisms. Invest Ophthalmol Vis Sci 2007;48:3396–404.
31. Pesudovs K, Hazel CA, Doran RM, Elliott DB. The usefulness of Vistech and FACT contrast sensitivity charts for cataract and refractive surgery outcomes research. Br J Ophthalmol 2004;88:11–6.
32. Elliott DB, Bullimore MA. Assessing the reliability, discriminative ability, and validity of disability glare tests. Invest Ophthalmol Vis Sci 1993;34:108–19.
33. Haymes SA, Roberts KF, Cruess AF, Nicolela MT, LeBlanc RP, Ramsey MS, Chauhan BC, Artes PH. The letter contrast sensitivity test: clinical evaluation of a new design. Invest Ophthalmol Vis Sci 2006;47:2739–45.
34. Thayaparan K, Crossland MD, Rubin GS. Clinical assessment of two new contrast sensitivity charts. Br J Ophthalmol 2007;91:749–52.
35. Smith G. Ocular defocus, spurious resolution and contrast reversal. Ophthal Physiol Opt 1982;2:5–23.
Keywords:© 2010 American Academy of Optometry
contrast sensitivity; contrast threshold; Siemens star; sinusoidal test pattern; psychophysical test