Since the idea of evaluating visual function using OKN first developed in the 1920s, several researchers have demonstrated a correlation between subjective VA and objective VA measured using the OKN response.
2,9,10,16,20,21 However, none of the studies have presented a method of estimating a subjective VA using the objective VA test results obtained with OKN responses. Recently, by dividing the OKN responses into various steps, we demonstrated that each step of the OKN response corresponded to a certain level Snellen VA and that there was a significant difference in the extent of Snellen VA among several steps.
1,3 As a follow-up study, by recruiting more patients and combining the data of induction and suppression methods, we created a table for estimation of Snellen VA using the objective test results (
Table 1). As Snellen VAs are ordinal data rather than interval data and are not adequate for quantification and statistical analyses, we converted the Snellen VAs to logMAR VAs for statistical analyses.
22,23 Then, we converted the logMAR units to Snellen VAs to express the estimated VA in each objective VA step. Moreover, by analyzing the results, we established a cutoff value for detection of VA of >20/200. Validation of the cutoff revealed high reliability in both groups of patients. In particular, the false-positive rate of the cutoff was low (11.7% in phase I and 3.3% in phase II), and the positive predictive value was high (93.4% in phase I and 97.7% in phase II) in both groups.
In the present study, we developed a cutoff of objective VA for a subjective VA of 20/200. A VA of 20/200 is one of the criteria for legal blindness in the United States and Europe as well as in Korea, where it is an important criterion for diagnosing visual disability. Patients who are diagnosed with legal blindness or visual disability can receive government disability benefits; therefore, malingerers usually feign a low vision of ≤20/200, although their real VAs are substantially higher. Detecting a VA of >20/200 would also be an effective way to uncover malingering; thus, our objective VA test device could be very helpful in the diagnosis of legal blindness, particularly in the detection of FVL, which is defined as any vision decrease with an origin that cannot be attributed to a pathologic or structural abnormality.
24
There is a debate about the roles of the macula and the peripheral retina in the generation of OKN responses. Miyoshi
25 demonstrated the increased difficulty in eliciting an OKN response with narrowing of the central visual field (VF). By contrast, several researchers have revealed that patients with a central VF defect show an OKN gain similar to that in those who do not have a central VF defect.
26 –29 Valmaggia and Gottlob
29 observed that, in patients with a central VF defect, OKN was suppressed when they fixated on the central scotoma, whereas OKN was elicited simultaneously with filling in of the scotoma. They postulated that the OKN response was triggered by activation of the visual cortical area corresponding to the scotoma through filling in—that is, expansion of receptive fields accompanied by feedback pathways.
29 In addition, the OKN response was shown to be suppressed through fixation to the central scotoma with central artificial scotomas where filling in is impossible, suggesting that the central retina is also important in the generation of the OKN response.
11,14,30,31 Therefore, our suppression method may be a useful tool that can specifically reflect the function of the central retina, as it can induce the patients to fixate on the center of the screen, whereas the induction method conceivably reflects both central and peripheral retinal function.
3 Fukai et al.
6 reported that the objective VA test result was worse than the subjective VA in patients with central scotoma due to the lesion of retina or optic nerve, supporting the assumption that the suppression method may be a good option for determining central visual function. We also believe that the suppression method is suitable for measurement of central visual function, and the combination of induction and suppression methods makes the estimation of objective VA more accurate, thus enabling clinical application of the objective VA test.
3
This study had the following limitations: (1) The sensitivity of our cutoff was 91.7% in phase I and 86% in phase II, suggesting that approximately 10% of patients with a VA of >20/200 can be diagnosed as having VA of ≤20/200, which can lead to the misdiagnosis of legal blindness. For more sensitive measurement of objective VA, further studies with adjusted velocity of visual stimuli and size of target are needed. Moreover, technical development, such as, the use of projectors with a higher refresh rate, can enable the use of faster velocity of stripe movement and thus can improve the sensitivity and reduce the false-positive rate of our objective VA test device. (2) The number of patients at each step of the objective VA scale is still not large enough to present a more specific range of corresponding Snellen VA. Therefore, further studies with a larger sample group to accumulate both objective and subjective VA data are needed for more precise and accurate predictions of subjective VA. (3) In its current state, our test device cannot be used in patients with spontaneous nystagmus. Although patients with nystagmus were excluded from this study, application of an objective VA test device in patients with spontaneous nystagmus is still necessary, considering that a substantial number of patients with low vision have spontaneous nystagmus. Upgrade of both hardware and software of the objective VA test device (i.e., development of a new algorithm that can analyze and offset the effect of spontaneous nystagmus—for instance, by quantifying the slow-phase eye velocity of the spontaneous nystagmus and subtracting that from the slow-phase velocity during optokinetic stimulation) can offer a solution. (4) There are some technical restrictions displaying moving stripes with liquid crystal display (LCD) beam projector or LCD/light-emitting diode (LED) flat panel display. The first one is the stroboscopic effect. Conventional LCD beam projector and LCD/LED flat panel display supports a 60-Hz refresh rate signal from the video card of the computer. This restrains the
v max of the stripe movement up to 10 deg/s in the current experimental setting. If the moving speed were increased, then the narrow stripe would be shown as reversed, slow motion instead of running fast. We can experience this phenomenon in a rotating ceiling fan or a rotating wheel of a car. The fast-rotating fan or wheel may sometimes be perceived as still, and this phenomenon can occur in our experimental setting if we increase the velocity of stripe movement beyond 10 deg/s. With the development of an LCD/LED panel with a higher refresh rate (i.e., 120 Hz), we expect that the efficacy of our objective VA test device will be enhanced by overcoming the stroboscopic effect and enabling the use of the velocity of stripe movement beyond 10 deg/s. The second one is the moving direction of the stripes. All the television systems and the standard computer monitor make their image by assembling scanning lines that run from the right to the left, from the top to the bottom of the screen. This determines the direction of the smooth movement of the object on the screen. If the moving object is shown on the screen from left to right, then the horizontal image misalignment can be seen more easily than it can in with right to left movement of the target. We can observe this phenomenon when we watch a soccer game on digital television. If the player moves from the left to the right very fast, then the “block noise” usually occurs, which can be identified as blocked or mosaic patterned border of the player's body. This is why we always used a right-to-left moving target. However, considering that monocular OKN is nearly symmetrical for stimulus in the temporal-to-nasal and nasal-to-temporal directions in adults without any history of binocular disruption in early life, we believe that the fact that we used only one direction for the OKN stripes significantly limits the clinical importance of the study.
1,3 Third is the level of contrast. If we increase the contrast up to 100%, then the rainbow effect can be seen more easily in a moving target. That is, the moving white stripes will cause a rainbow-like color dispersion on the tail-side border, because the white signal is composed of red+green+blue cells, and each cell signal will be delivered sequentially, not simultaneously. This is inevitable in controlling liquid crystal in an LCD display and more severe in a digital light processing (DLP) projector, because the DLP projector uses a rotating red/green/blue filter. The 85% contrast is rather arbitrary; however, it appeared to be the optimal condition for our experiments.
Despite these limitations, however, our results, especially the low false-positive rates, suggest that our objective VA test device could be an effective tool in detection of VA of >20/200 and has a possibility of clinical application (i.e., detection of FVL or diagnosis of legal blindness), even in its current state.
In conclusion, this study showed that an objective VA test using OKN responses can be effective in the detection of VA of >20/200. Widespread clinical use could be possible through development of more precise steps for measurement of objective VA. Development of new technologies is also expected to enhance the efficacy of our objective test device.
Supported by Grant A080299 from the Korea Health 21 R&D Project, Korea Health Industry Development Institute, Republic of Korea.