Doc Ophthalmol (2014) 129:105–114 DOI 10.1007/s10633-014-9457-7
ORIGINAL RESEARCH ARTICLE
Comparing enfant and PowerDiva sweep visual evoked potential (sVEP) acuity estimates William H. Ridder III • Bradley S. Waite Timothy F. Melton
•
Received: 20 March 2014 / Accepted: 15 August 2014 / Published online: 24 August 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose Many studies have examined different variables that affect the outcome of sVEP estimated acuity. However, no studies have compared the estimated sVEP acuity between different instruments. The primary purpose of this study was to compare sVEP acuity estimates obtained with two different sVEP systems: the Enfant and the PowerDiva. Methods Twenty-five normal adults with monocular acuities of 0.10 logMAR or better took part in this study. The sVEP acuities were determined with the two instruments in a single visit with the same electrode placement. For both systems, the stimuli were horizontal sine wave gratings of 80 % contrast, counterphased at 7.5 Hz, with a screen mean luminance of 100 cd/m2. The sweep presented spatial frequencies from 3 to 36 cpd with each spatial frequency presented for 1 s. Ten presentations of the stimuli were averaged together for one acuity measurement. The acuity estimate was made with the specific instruments standard software. Two acuity measurements were made for each system and averaged together for further comparison. The acuity estimates were compared using an ANOVA, paired t tests, and Bland–Altman plots.
W. H. Ridder III (&) B. S. Waite T. F. Melton Southern California College of Optometry, Marshall B. Ketchum University, 2575 Yorba Linda Blvd., Fullerton, CA 92831, USA e-mail: [email protected]
Results The average estimated logMAR acuities with the Enfant (0.064 ± 0.069 logMAR) and PowerDiva (0.065 ± 0.115 logMAR) were not significantly different (t = 0.04, p = 0.97). Consistent with previous studies, the logMAR chart acuity (0.086 ± 0.089 logMAR) was significantly different from the Enfant (t = 8.10, p \ 0.001) and PowerDiva (t = 5.77, p \ 0.001) acuity estimates. The Bland– Altman analysis for the two instruments did not indicate a bias (-0.001), and the limit of agreement was 0.227 logMAR. Conclusions Acuity estimates with the Enfant and PowerDiva are not significantly different for patients with normal acuity. Thus, direct comparisons between the two instruments can be made for patients with normal acuity. Keywords Visual acuity Sweep visual evoked potential LogMAR acuity
Introduction Visual acuity is typically measured with the Snellen chart in clinic. Even though this is the most common chart used in clinic, it has some inherent design flaws. These flaws were addressed, and specific design principles were proposed by Bailey and Lovie [1]. Based on these design principles, the logMAR acuity chart was developed. The logMAR chart has become
123
106
the ‘gold standard’ for measuring acuity. There are several types of logMAR charts that have been developed [2–4]. However, there are some patients that are unable to read a chart with letters or optotypes, thus making it difficult or impossible to determine acuity in these individuals. The sweep visual evoked potential (sVEP) can be used to obtain acuity estimates in children and non-responsive patients [5–14] that are unable to read a letter chart. The sVEP was initially used to determine acuity development in normal infants [6]. This technique demonstrated a gradual increase in acuity until a plateau was reached at about 8 months of age. At this age, the acuity estimate was not significantly different from adults. Other studies indicated that these acuity estimates were reliable from trial to trial in children and adults [5, 15]. In a review of the literature, two principal instruments [i.e., the Enfant (Neuroscientific Corp.) and the PowerDiva (Digital Instrumentation for Visual Assessment)] have been used to estimate the sVEP acuity. These two instruments use different methods to determine the visual acuity. One difference between the two instruments is the method of noise determination. The Enfant system uses the 95 % confidence interval of the signal magnitude to determine the noise level for estimating acuity. At a specific spatial frequency, if the 95 % confidence interval overlapped zero, then the signal at this frequency was not considered different from noise. The PowerDiva uses the magnitude of a Fourier frequency adjacent to 2X the signal frequency as noise [5, 6, 11, 16, 17]. Employing different noise level determinations may alter the acuity extrapolation. A previous study used a single data set obtained on the Enfant and applied the two different noise estimation techniques to determine the sVEP acuity [18]. The sVEP acuity estimates were not found to be different with the two techniques. However, because of other differences between the Enfant and the PowerDiva (e.g., software and hardware differences such as the size of the stimulus screen, data acquisition, or the input amplifier), different sVEP acuities could still be derived from the same subject. The stimulus monitor with the Enfant subtends a larger angle than the PowerDiva (see Methods below). The data acquisition with the Enfant has 12 bit accuracy with a sampling rate of 300 Hz compared to the PowerDiva, which has 16 bit accuracy with a sampling rate of 600 Hz. Additionally, the input filters and the signal
123
Doc Ophthalmol (2014) 129:105–114
gains are not the same with the two instruments (see Methods below). The primary purpose of this study was to compare sVEP acuity estimates obtained with two different sVEP systems: the Enfant and the PowerDiva. The sVEP acuity estimates are also compared to the logMAR chart acuities for individual subjects.
Methods Subjects Twenty-five adult (average age: 26.8 ± 3.81 (SD), range 21–35) subjects having monocular best corrected visual acuity of at least 0.10 logMAR were selected for this study (Table 1). The dominant eye with its best optical correction was always tested. A logMAR chart was used to determine visual acuity (Smart System II, M&S Technologies, Inc.). The ETDRS mode was used, and the letter-by-letter scoring method was employed. A power test based on a sample of 10 subjects indicated that a difference of 3 cpd (less than a 0.05 log unit change) could be detected for a sample size of 18 (alpha = 0.05, power = 0.80, assumed standard deviation for the estimate was 4). A 0.05 log unit change is less than the 95 % confidence interval for test–retest repeatability (i.e., ±0.10 logMAR) of psychophysically measured logMAR acuity chart data [19]. Subjects exhibiting ocular and/or systemic pathologies were excluded. Subjects taking medications that could affect vision were also excluded. There were no other exclusion criteria. The subject pool was limited to University employees and students. All subjects gave informed consent to participate in the study, and the research followed the tenets of the Declaration of Helsinki. The procedures were approved by the Institutional Review Board of the Marshall B. Ketchum University. Enfant stimulus A horizontal-oriented, sine wave grating that swept up the spatial frequency spectrum was used. The Enfant 4010 presented eleven spatial frequencies ranging from 3 to 36 cpd (3, 6, 9, 12, 15, 18, 20.1, 22.5, 25.8, 30, 36 cpd). Ten presentations of the stimuli were averaged together for a single acuity estimate. Two acuity estimates were made. The contrast was set at 80 % and the temporal reversal rate (square wave)
Doc Ophthalmol (2014) 129:105–114
107
Table 1 Subject data displaying the logMAR chart acuities, the Enfant and PowerDiva sVEP acuity estimates in cycles per degree (cpd), and the average Enfant and PowerDiva sVEP acuity estimates converted to logMAR values SubjectGender
Age
logMAR VA
(years) 1-M
31
-0.06
Enfant (cpd)
Enfant logMAR
Trial 1
Trial 2
Ave
28.2
29.3
28.8
0.018
PowerDiva (cpd)
PowerDiva logMAR
Trial 1
Trial 2
Ave
32.6
26.2
29.4
0.009
2-M
28
-0.06
28.8
29.1
29.0
0.015
31.0
30.4
30.7
-0.010
3-M
33
-0.10
22.0
20.3
21.2
0.152
17.1
15.3
16.2
0.267
4-M
28
-0.02
20.0
18.8
19.4
0.189
12.1
25.7
18.9
0.201
5-M
31
-0.10
23.3
21.3
22.3
0.129
27.4
18.9
23.2
0.112
6-M 7-F
31 26
-0.08 -0.10
28.4 38.3
22.1 31.3
25.3 34.8
0.075 -0.064
16.0 34.3
25.6 25.0
20.9 29.7
0.159 0.004
8-M
29
-0.08
23.0
25.8
24.4
0.090
14.8
18.6
16.7
0.254
9-F
25
-0.06
25.8
22.5
24.2
0.094
17.1
17.1
17.1
0.243
10-F
22
-0.08
32.2
23.5
27.9
0.032
17.9
24.1
21.0
0.155
11-F
25
-0.20
32.6
31.0
31.8
-0.025
30.8
37.1
34.0
-0.054
12-F
26
-0.06
22.5
23.9
23.2
0.112
29.2
26.4
27.8
0.033
13-M
21
-0.14
25.8
28.8
27.3
0.041
30.1
30.8
30.5
-0.007
14-F
25
-0.04
22.5
22.4
22.5
0.126
31.3
25.0
28.2
0.028
15-F
25
-0.22
28.0
21.0
24.5
0.088
26.8
32.9
29.8
0.003
16-M
24
-0.20
30.4
30.2
30.3
-0.004
40.4
48.9
44.7
-0.173
17-F
24
-0.06
23.3
23.2
23.3
0.111
26.9
28.0
27.4
0.039
18-M
24
?0.06
25.8
20.1
23.0
0.116
29.8
27.6
28.7
0.019
19-M
29
?0.04
25.8
18.1
22.0
0.136
29.4
32.9
31.1
-0.016
20-M
22
?0.06
23.0
30.0
26.5
0.054
25.6
23.5
24.6
0.087
21-M
28
-0.20
32.0
36.0
34
-0.054
16.9
15.2
16.1
0.272
22-M 23-M
32 24
-0.18 -0.18
30.0 21.4
36.0 24.4
33 22.9
-0.041 0.117
32.7 29.1
32.0 30.4
32.4 29.8
-0.033 0.004
24-M
22
-0.18
22.9
25.3
24.1
0.095
32.3
31.7
32.0
-0.028
25-F
35
?0.10
30.0
28.5
29.3
0.011
28.6
22.7
25.6
0.070
Mean
26.8
-0.086
26.6
25.7
26.2
0.064
26.4
26.9
26.6
SD
3.81
0.089
4.44
5.04
4.24
0.069
7.29
7.35
6.74
0.065 0.115
The data displayed in bold are estimated acuities based on a single spatial frequency (see Methods section)
7.5 Hz. The mean screen luminance was 100 cd/m2. The stimulus screen was viewed monocularly at 1.5 m [12.7° (H) 9 9.7° (V)]. Each spatial frequency was presented for 1 s, and there was 1 s of adaptation before data collection. PowerDiva stimulus The PowerDiva (software version 3.5) also presented a horizontal sine wave grating stimulus. The spatial frequency sweep contained ten spatial frequencies between 3 and 36 cpd (3, 6.67, 10.33, 14, 17.67, 21.33,
25, 28.67, 32.33, 36 cpd). Each spatial frequency was presented for 1 s with 1 s allowed for adaptation before data collection. The contrast was set at 80 % and the temporal reversal rate (square wave) set to 7.5 Hz. The mean screen luminance was 100 cd/m2. The monitor was viewed monocularly at 3 m [7o (H) 9 6.06o (V)]. Ten presentations of the stimuli were averaged together for a single acuity estimate. Two acuity estimates were made. The magnitude and phase plots, from which the acuity was derived, were determined with the recursive least squares (RLS) method [20].
123
108
Fig. 1 An example of the Enfant and PowerDiva sVEP data for subject 17. The arrows delineate the data included in the linear fits for acuity extrapolation. The extrapolated acuities are given
Recording technique Data were acquired for the Enfant and PowerDiva systems in a single session in random order. The recording electrode (Ag/AgCl) was placed 2 cm above the inion on the midline. The reference and ground electrodes were placed on the earlobes. The average electrode resistance was 9.5 ± 5.01 KOhms (Grass Electrode Impedance Meter, Model EZM3A, Quincy, MA). The electrodes were positioned once, and the sVEPs for both instruments were collected. The signal was amplified (10,000X for the Enfant and 50,000X for the PowerDiva), bandpass filtered (0.5–100 Hz for the Enfant and 0.3–100 Hz for the PowerDiva), and digitized. Acuity extrapolation Figure 1 displays an example of the sVEP magnitude data for both instruments for subject 17. The horizontal axis is the stimulus spatial frequency in cpd, and the vertical axis is the sVEP magnitude in microvolts. The linear fits to determine acuity are displayed for both sets of data. For the Enfant, the magnitude of the signal at each spatial frequency was determined from a discrete Fourier transform at two times the stimulus frequency (i.e., stimulus frequency = 7.5 Hz, so the analysis frequency was 15 Hz). A specific data point was considered to be noise if the 95 % confidence interval
123
Doc Ophthalmol (2014) 129:105–114
for the data at a spatial frequency overlapped with zero. The error bars are not shown in Fig. 1, but all data above 18 cpd had 95 % confidence intervals that overlapped zero. The acuity extrapolation technique for the Enfant has previously been described [13, 15, 16, 18]. Briefly, one of the two possible methods was employed. The first method fits a straight line to the data between the peak of the function (6 cpd in Fig. 1) and the highest spatial frequency above noise (18 cpd in Fig. 1). The data included in the fit are delineated by the arrows at 6 and 18 cpd. The line was extrapolated to the x-axis (i.e., zero magnitude) for acuity. In Fig. 1 for the Enfant data, the acuity is 23.3 cpd. The second method was used when a line could not be fit to the data. This could occur when there were multiple peaks in the sVEP function. In these circumstances, the highest spatial frequency above noise was taken as the acuity estimate. For the Enfant, 12 out of 50 acuity estimates were not determined by a linear fit (data in bold in Table 1). For the PowerDiva, the response signal was determined from a discrete Fourier transform at 15 Hz, similar to the Enfant. The noise estimate was taken as the average magnitude at 14.07 and 15.94 Hz from the discrete Fourier transform. The data were initially examined to determine whether at least one spatial frequency had a signal-to-noise ratio (i.e., SNR) greater than 3.0. All the sVEP functions met these criteria. The Tcirc statistic was also used to help determine noise [21]. The Tcirc statistic is the probability that a specific response is above noise. The lower the Tcirc statistic, the more likely the response is above noise. At least one spatial frequency in the data needed a Tcirc value \0.05. A similar fitting paradigm to that of the Enfant was used for the data from the PowerDiva. A line was fit from the peak of the function (6.67 cpd in Fig. 1) to the highest spatial frequency data above noise (21.33 cpd in Fig. 1). These spatial frequencies are identified with an arrow in Fig. 1. The data at each spatial frequency were considered to be noise if the SNR was one or less. For the PowerDiva data in Fig. 1, the SNR for the data above 21.33 cpd was below 1.0. The linear fit was extrapolated to the x-axis for the acuity estimate. Using this technique, the acuity for the PowerDiva data in Fig. 1 is 26.9 cpd. For the PowerDiva, 4 out of 50 acuity estimates could not be determined with a linear fit (bold in Table 1). For these
Doc Ophthalmol (2014) 129:105–114
109
data, the acuity estimate was taken as the highest spatial frequency above noise (i.e., similar to the Enfant method). The Enfant had more data sets that could not be fit with a line (i.e., 12 vs. 4) because it also had more data sets with multiple peaks. There was no consistent difference between data obtained with a linear fit and that determined by a single data point. The phase plots were also used for both instruments. The phase was constant or gradually changing within the range of spatial frequencies employed to determine acuity. Above the acuity limit, the phase varied randomly. Thus, by examining the phase plots, the approximate location of the acuity estimate could be determined. Statistical analysis Initially, a one-way ANOVA was used to determine whether there was a significant difference in acuity with the different instruments. A post hoc, pairedsample t test was then conducted to compare visual acuity measured with the logMAR acuity chart, the Enfant, or the PowerDiva. The significant p value was 0.05. Comparisons between data sets were also made with the intraclass correlation coefficient (ICC). The ICC determines how strongly two measurements agree by assessing the proportion of the total variance within the data that can be explained by the variance between the two measurements. Simply, the ICC indicates how well the two measurements fit the one to one line for the plotted data. The 95 % confidence intervals (i.e., 1.96 standard deviations) of the acuity differences were determined to indicate the repeatability across tests. The statistical tests were performed with the software package Minitab 16 Statistical Software (version 16.2.4 by Minitab, Inc.). Bland–Altman plots were also produced [22, 23]. The bias and the limits of agreement were determined from these plots.
Results The acuity estimates for the subjects are in Table 1. For the Enfant and the PowerDiva, the acuity estimates for the first trial, second trial, and average of the two trials are given in cycles per degree (cpd). These average acuities, in cpd, are then converted to logMAR units (30 cpd = 20/20 = 0.00 logMAR). The acuities were converted to logMAR units with
Fig. 2 The logMAR visual acuities determined from the chart (average = -0.086 ± 0.089 logMAR) and the Enfant (0.064 ± 0.069 logMAR, t = 8.1, p \ 0.001) for each subject. The diagonal line is the 1:1 line. If the data fell on the 1:1 line, the two acuity estimates would be similar. The graph demonstrates that the acuity estimates with the logMAR chart are significantly lower than those with the Enfant (p \ 0.001)
the following equation: logMAR = log(30/cpd) [24]. The column labeled ‘logMAR VA’ in Table 1 contains the acuities determined with the logMAR chart. The means and standard deviations of the data are given at the bottom of each column. A one-way ANOVA comparing logMAR VA, Enfant VA, and PowerDiva VA indicated that there was a significant difference (p \ 0.001, F = 21.99) between the groups. In the one-way ANOVA, the logMAR acuity threshold was the dependent variable and the different test groups (i.e., Enfant, PowerDiva, logMAR chart) were the independent variables. Post hoc, paired t tests indicated that the logMAR VA acuities [average = -0.086 ± 0.089 (SD) logMAR] and the Enfant acuities (0.064 ± 0.069 logMAR) were significantly different [t = 8.10, p \ 0.001, average difference = 0.15 ± 0.093 (SD)]. Similarly, the logMAR VA and the PowerDiva acuities (0.065 ± 0.115 logMAR) were significantly different [t = 5.77, p \ 0.001, average difference = 0.15 ± 0.131 (SD)]. The Enfant and PowerDiva acuities were not significantly different [t = 0.04, p = 0.966, average difference = 0.001 ± 0.116 (SD)].
123
110
Fig. 3 The logMAR visual acuities determined from the chart (average = -0.086 ± 0.089 logMAR) and the PowerDiva (0.065 ± 0.125 logMAR, t = 5.77, p \ 0.001) for each subject. The diagonal line is the 1:1 line. If the data fell on the 1:1 line, the two acuity estimates would be similar. The graph demonstrates that the acuity estimates with the logMAR chart are significantly lower than those with the PowerDiva (p \ 0.001)
Figure 2 displays the Enfant sVEP acuity estimate data plotted against the logMAR chart acuity data for each subject. The solid line is the 1:1 line. The majority of the data are above this line. Thus, the Enfant sVEP acuities are significantly worse (i.e., larger, positive numbers) than the logMAR chart acuities (p \ 0.001). This is similar to previous publications [15, 18]. The logMAR chart acuity has been reported to be 0.25 ± 0.122 [18] or 0.21 ± 0.089 logMAR [15] units better than the Enfant sVEP acuity estimate. For the normal subjects in this study (N = 25), the logMAR chart acuity is 0.15 ± 0.093 logMAR units lower than the Enfant sVEP acuity estimate. Figure 3 displays the PowerDiva sVEP acuity estimate data plotted against the logMAR chart acuity data for each subject. The solid line is the 1:1 line. Similar to the Enfant, the majority of the data lie above this line. Thus, the PowerDiva sVEP acuities are significantly worse than the logMAR chart acuities (p \ 0.001). The average difference between the PowerDiva sVEP acuity estimate and the logMAR
123
Doc Ophthalmol (2014) 129:105–114
Fig. 4 The logMAR visual acuity estimates determined from the Enfant (0.064 ± 0.069 logMAR) and the PowerDiva (0.065 ± 0.125 logMAR, t = 0.04, p = 0.966) for each subject. The diagonal line is the 1:1 line. The ICC is 0.25. The graph indicates that the two acuity estimates are similar (p = 0.966)
chart acuity is 0.15 ± 0.131 logMAR units. This is similar to the measurements with the Enfant. Figure 4 displays the PowerDiva sVEP acuity estimate data plotted against the Enfant sVEP acuity estimate data for each subject. The solid line is the 1:1 line. Much of the data are clustered around the 1:1 line. The paired t test did not find a significant difference between the two sets of data (p = 0.966). The intraclass correlation coefficient (ICC, r2) for the data was 0.25. Figure 5 displays the Bland–Altman plot for the Enfant sVEP acuity estimate data and the logMAR chart acuity data for each subject. The average of the two acuities for each subject is plotted on the horizontal axis, and the difference is plotted on the vertical axis. The data display a bias of ?0.15 (i.e., the line labeled MEAN), and the limit of agreement (±1.96 standard deviations) is ±0.182. Figure 6 displays the Bland–Altman plot for the PowerDiva sVEP acuity estimate data and the logMAR chart acuity data for each subject. The format is the same as Fig. 5. Similar to Fig. 5, the bias is ?0.15 and the limit of agreement is ±0.257. Figure 7 displays the Bland– Altman plot for the PowerDiva sVEP acuity estimate data and the Enfant sVEP acuity estimate data for each subject. The data do not display a bias (an average
Doc Ophthalmol (2014) 129:105–114
Fig. 5 A Bland–Altman plot comparing the logMAR chart and Enfant sVEP acuity estimates. The bias of 0.150 (i.e., the difference between zero and the MEAN line) indicates that the acuity estimates measured with the Enfant are poorer than with the logMAR chart. The limit of agreement is ±0.182 (i.e., ±1.96 standard deviations from the MEAN)
Fig. 6 A Bland–Altman plot comparing the logMAR chart and PowerDiva sVEP acuity estimates. The bias (0.151) indicates that the acuity estimates measured with the PowerDiva are poorer than with the logMAR chart. The limit of agreement is ±0.257
difference of -0.001), and the limit of agreement is ±0.227. The intra-session repeatability of the Enfant and PowerDiva data from trial 1 to trial 2 was also
111
Fig. 7 A Bland–Altman plot comparing the Enfant and PowerDiva sVEP acuity estimates. The bias (-0.001) indicates that the acuity estimates measured with the two instruments are similar. The limit of agreement is ±0.227
determined (data in Table 1). The intra-session, test– retest 95 % confidence interval and bias for the Enfant were ±0.140 and 0.018 logMAR units. The 95 % confidence interval and bias for the PowerDiva were ±0.210 and -0.011 logMAR units. Figure 8 displays the current data along with data from previous publications employing the Enfant or the PowerDiva. The majority of these data were taken from a previous publication (Fig. 4; [18]). The open symbols in the figure are data taken from normal adults with good acuity (i.e., approximately 0.00 logMAR). The filled symbols are the data for subjects with decreased visual acuity. The data from Ridder et al. [16] are the first second of the sVEP data found in Table 1 and Fig. 3. The data from Katsumi et al. [25] are extrapolated from their Figs. 2 and 3 at the zero diopter level for subjects with good acuity. The data from Arai et al. [26] are extrapolated from Fig. 2 for the subjects with good acuity (20/20 or better acuity) or subjects with decreased acuity (20/30–20/200). The data from Yadav et al. [27] are the average of 6 normal adult subjects that they report in their Fig. 3a (category C2, 100 cd/m2). They used the PowerDiva system. The majority of these data for the subjects with normal acuity lie above the solid line (i.e., the 1:1 line) indicating that the logMAR chart acuity is lower than the sVEP acuity estimates. The two dashed lines in Fig. 8 that parallel the solid line (i.e., the 1:1 line)
123
112
Doc Ophthalmol (2014) 129:105–114
Fig. 8 A comparison of the data from this study with previous published data employing the Enfant or the PowerDiva instrument. See text for details
are one octave above or below the 1:1 line. The dotted lines are two octaves above or below the 1:1 line. The majority of the Enfant sVEP acuity estimate data for normal subjects are clustered about 1 octave above the 1:1 line. The sVEP acuity estimate data from the PowerDiva (i.e., open stars and hexagon symbols) do not appear to be different from the Enfant for the normal subjects (i.e., the other open symbols). It was previously noted that the subjects with decreased visual acuity overlay the 1:1 line better than the subjects with normal acuity [18]. The majority of the subjects with decreased visual acuity (49/56) have acuity estimates within 1 octave of the 1:1 line.
Discussion There are several different instruments that have been used to measure the sVEP acuity. However, there are no published studies which compare the sVEP acuity estimates of one instrument with another. The present study compared the sVEP acuities from two commonly referenced instruments. This allows for direct comparisons of sVEP acuity estimate results from different published papers using these two instruments. The results demonstrate that there is a statistical difference between the acuities measured on the
123
logMAR chart versus those estimated with the Enfant (p \ 0.001) or PowerDiva (p \ 0.001). Both instruments give acuities that are about 0.15 logMAR units worse than the logMAR chart acuity for normal subjects. The sVEP acuities measured with the Enfant, and the PowerDiva are not statistically different (p = 0.966). The sample size test indicated that 25 subjects were sufficient to detect a difference of 3 cpd or about 0.05 logMAR units. This is a smaller difference than has been reported for the test–retest repeatability for the logMAR acuity chart [19]. Comparisons with previous data demonstrate that the sVEP acuities measured with the Enfant are consistent across several laboratories (see Fig. 8). Furthermore, these comparisons demonstrate that the sVEP acuities determined with the PowerDiva are also similar across laboratories, as well as, with the Enfant sVEP acuity measurements. Thus, direct comparisons of sVEP acuity estimates of normal subjects can be made with the Enfant and PowerDiva. An intersession, test–retest repeatability study with the Enfant resulted in a 95 % confidence interval of ±0.11 logMAR units and a bias of -0.003 when two sVEP acuity estimates are averaged together per session [15]. This previous study also reported a 95 % confidence interval with the logMAR acuity chart of ±0.07 logMAR units and a bias of 0.01 for the
Doc Ophthalmol (2014) 129:105–114
normal adults [15]. This indicates that the repeatability for the sVEP acuity estimate with the Enfant is nearly as good as the logMAR acuity chart (i.e., the gold standard for acuity measurements). The current study comparing the Enfant and the PowerDiva resulted in a 95 % confidence interval of ±0.227 logMAR units and a bias of -0.001. Thus, the sVEP repeatability is best when the same instrument is used to repeat the test and decreases when repetitions are made across instruments. The intra-session, test–retest repeatability with the Enfant is reported to have a 95 % confidence interval of about ±0.19 logMAR units (±0.191 and ±0.186 for visit one and visit two from Ridder et al. [15]). The bias for visit one and visit two was 0.010 and -0.019 [15]. In the present study, the intra-session, test–retest 95 % confidence interval and bias for the Enfant were ±0.140 and 0.018 logMAR units. These values agree with the previously published data [15]. The 95 % confidence interval and bias for the PowerDiva were ±0.210 and -0.011 logMAR units. Thus, there was no bias with either instrument, and the repeatability was slightly worse with the PowerDiva. The different stimulus size between the Enfant [12.7° (H) 9 9.7° (V)] and the PowerDiva [7° (H) 9 6.06° (V)] might be expected to result in a difference in acuity estimates between the two instruments. The larger stimulus size for the Enfant would result in a greater number of stimulus cycles present for each spatial frequency. However, the decreased number of stimulus cycles with the PowerDiva instrument may only have an effect at low spatial frequencies [28]. The lower spatial frequencies, with the fewer stimulus cycles, would result in lower visual evoked potential magnitudes. An example of this effect is seen in Fig. 1. The response magnitude for the lower spatial frequencies is reduced for the PowerDiva in comparison with the Enfant. A similar difference was seen for 17 of the 25 subjects. One subject displayed greater magnitudes with the PowerDiva, and the remaining seven subjects had similar magnitudes with the two instruments. However, since the acuity is extrapolated from the higher spatial frequencies, this decreased signal magnitude for the lower spatial frequencies does not affect the results. Similar observations have been made by Tyler et al. [29] and Almoqbel et al. [30]. Different stimulus screen sizes will have an additional affect if sVEP comparisons are made in infants
113
or children. Since the fixation of infants and children may be more variable than behaving adults, the smaller screen size may result in more variable data. Thus, the present results in adults may not be translatable to infants if these screen distances are used. In conclusion, the sVEP acuity estimates for the Enfant and the PowerDiva are not statistically different for normal subjects. Thus, comparisons between the two instruments can be made. The present data and several published papers demonstrate that both instruments result in poorer acuity estimates then the logMAR acuity chart for normal subjects. Comparisons with the two instruments still need to be made with subjects with decreased acuity. Conflict of interest All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or nonfinancial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.
References 1. Bailey IL, Lovie JE (1976) New design principles for visual acuity letter charts. Am J Optom Physiol Opt 53(11): 740–745 2. Taylor HR (1978) Applying new design principles to the construction of an illiterate E chart. Am J Optom Physiol Opt 55(5):348–351 3. Ferris FL 3rd, Kassoff A, Bresnick GH, Bailey I (1982) New visual acuity charts for clinical research. Am J Ophthalmol 94(1):91–96 4. Strong G, Woo GC (1985) A distance visual acuity chart incorporating some new design features. Arch Ophthalmol 103(1):44–46 5. Norcia AM, Tyler CW (1985) Infant VEP acuity measurements: analysis of individual differences and measurement error. Electroencephalogr Clin Neurophysiol 61(5):359–369 6. Norcia AM, Tyler CW (1985) Spatial frequency sweep VEP: visual acuity during the first year of life. Vision Res 25(10):1399–1408 7. Norcia AM, Tyler CW, Piecuch R, Clyman R, Grobstein J (1987) Visual acuity development in normal and abnormal preterm human infants. J Pediatr Ophthalmol Strabismus 24(2):70–74 8. Hamer RD, Norcia AM, Tyler CW, Hsu-Winges C (1989) The development of monocular and binocular VEP acuity. Vision Res 29(4):397–408 9. Gottlob I, Fendick MG, Guo S, Zubcov AA, Odom JV, Reinecke RD (1990) Visual acuity measurements by swept
123
114
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Doc Ophthalmol (2014) 129:105–114 spatial frequency visual-evoked-cortical potentials (VECPs): clinical application in children with various visual disorders. J Pediatr Ophthalmol Strabismus 27(1):40–47 Gottlob I, Wizov SS, Odom JV, Reinecke RD (1993) Predicting optotype visual acuity by swept spatial visualevoked potentials. Clin Vision Sci 8:417–423 Katsumi O, Denno S, Arai M, De Faria JL, Hirose T (1997) Comparison of preferential looking acuity and pattern reversal visual evoked response acuity in pediatric patients. Graefes Arch Clin Exp Ophthalmol 235(11):684–690 Birch EE, Hoffman DR, Uauy R, Birch DG, Prestidge C (1998) Visual acuity and the essentiality of docosahexaenoic acid and arachidonic acid in the diet of term infants. Pediatr Res 44(2):201–209 Ridder WH 3rd, Rouse MW (2007) Predicting potential acuities in amblyopes: predicting post-therapy acuity in amblyopes. Doc Ophthalmol 114(3):135–145. doi:10.1007/ s10633-007-9048-y Ghasia F, Brunstom J, Tychsen L (2009) Visual acuity and visually evoked responses in children with cerebral palsy: gross motor function classification scale. Br J Ophthalmol 93(8):1068–1072. doi:10.1136/bjo.2008.156372 Ridder WH 3rd, Tong A, Floresca T (2012) Reliability of acuities determined with the sweep visual evoked potential (sVEP). Doc Ophthalmol 124(2):99–107. doi:10.1007/ s10633-012-9312-7 Ridder WH 3rd, McCulloch D, Herbert AM (1998) Stimulus duration, neural adaptation, and sweep visual evoked potential acuity estimates. Invest Ophthalmol Vis Sci 39(13):2759–2768 Riddell PM, Ladenheim B, Mast J, Catalano T, Nobile R, Hainline L (1997) Comparison of measures of visual acuity in infants: teller acuity cards and sweep visual evoked potentials. Optom Vis Sci 74(9):702–707 Ridder WH 3rd (2004) Methods of visual acuity determination with the spatial frequency sweep visual evoked potential. Doc Ophthalmol 109(3):239–247 Arditi A, Cagenello R (1993) On the statistical reliability of letter-chart visual acuity measurements. Invest Ophthalmol Vis Sci 34(1):120–129
123
20. Tang Y, Norcia AM (1995) An adaptive filter for steadystate evoked responses. Electroencephalogr Clin Neurophysiol 96(3):268–277 21. Victor JD, Mast J (1991) A new statistic for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 78(5):378–388 22. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307–310 23. Bland JM, Altman DG (1999) Measuring agreement in method comparison studies. Stat Methods Med Res 8(2):135–160 24. Thompson DA, Moller H, Russell-Eggitt I, Kriss A (1996) Visual acuity in unilateral cataract. Br J Ophthalmol 80(9):794–798 25. Katsumi O, Arai M, Wajima R, Denno S, Hirose T (1996) Spatial frequency sweep pattern reversal VER acuity vs Snellen visual acuity: effect of optical defocus. Vision Res 36(6):903–909 26. Arai M, Katsumi O, Paranhos FR, De Faria JML, Hirose T (1997) Comparison of Snellen acuity and objective assessment using the spatial frequency sweep PVER. Graefes Arch Clin Exp Ophthalmol 235(7):442–447 27. Yadav NK, Almoqbel F, Head L, Irving EL, Leat SJ (2009) Threshold determination in sweep VEP and the effects of criterion. Doc Ophthalmol 119(2):109–121. doi:10.1007/ s10633-009-9177-6 28. Savoy RL, McCann JJ (1975) Visibility of low-spatial-frequency sine-wave targets: dependence on number of cycles. J Opt Soc Am 65(3):343–350 29. Tyler CW, Apkarian P, Levi DM, Nakayama K (1979) Rapid assessment of visual function: an electronic sweep technique for the pattern visual evoked potential. Invest Ophthalmol Vis Sci 18(7):703–713 30. Almoqbel FM, Yadav NK, Leat SJ, Head LM, Irving EL (2011) Effects of sweep VEP parameters on visual acuity and contrast thresholds in children and adults. Graefes Arch Clin Exp Ophthalmol 249(4):613–623. doi:10.1007/ s00417-010-1469-8
ORIGINAL RESEARCH ARTICLE
Comparing enfant and PowerDiva sweep visual evoked potential (sVEP) acuity estimates William H. Ridder III • Bradley S. Waite Timothy F. Melton
•
Received: 20 March 2014 / Accepted: 15 August 2014 / Published online: 24 August 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose Many studies have examined different variables that affect the outcome of sVEP estimated acuity. However, no studies have compared the estimated sVEP acuity between different instruments. The primary purpose of this study was to compare sVEP acuity estimates obtained with two different sVEP systems: the Enfant and the PowerDiva. Methods Twenty-five normal adults with monocular acuities of 0.10 logMAR or better took part in this study. The sVEP acuities were determined with the two instruments in a single visit with the same electrode placement. For both systems, the stimuli were horizontal sine wave gratings of 80 % contrast, counterphased at 7.5 Hz, with a screen mean luminance of 100 cd/m2. The sweep presented spatial frequencies from 3 to 36 cpd with each spatial frequency presented for 1 s. Ten presentations of the stimuli were averaged together for one acuity measurement. The acuity estimate was made with the specific instruments standard software. Two acuity measurements were made for each system and averaged together for further comparison. The acuity estimates were compared using an ANOVA, paired t tests, and Bland–Altman plots.
W. H. Ridder III (&) B. S. Waite T. F. Melton Southern California College of Optometry, Marshall B. Ketchum University, 2575 Yorba Linda Blvd., Fullerton, CA 92831, USA e-mail: [email protected]
Results The average estimated logMAR acuities with the Enfant (0.064 ± 0.069 logMAR) and PowerDiva (0.065 ± 0.115 logMAR) were not significantly different (t = 0.04, p = 0.97). Consistent with previous studies, the logMAR chart acuity (0.086 ± 0.089 logMAR) was significantly different from the Enfant (t = 8.10, p \ 0.001) and PowerDiva (t = 5.77, p \ 0.001) acuity estimates. The Bland– Altman analysis for the two instruments did not indicate a bias (-0.001), and the limit of agreement was 0.227 logMAR. Conclusions Acuity estimates with the Enfant and PowerDiva are not significantly different for patients with normal acuity. Thus, direct comparisons between the two instruments can be made for patients with normal acuity. Keywords Visual acuity Sweep visual evoked potential LogMAR acuity
Introduction Visual acuity is typically measured with the Snellen chart in clinic. Even though this is the most common chart used in clinic, it has some inherent design flaws. These flaws were addressed, and specific design principles were proposed by Bailey and Lovie [1]. Based on these design principles, the logMAR acuity chart was developed. The logMAR chart has become
123
106
the ‘gold standard’ for measuring acuity. There are several types of logMAR charts that have been developed [2–4]. However, there are some patients that are unable to read a chart with letters or optotypes, thus making it difficult or impossible to determine acuity in these individuals. The sweep visual evoked potential (sVEP) can be used to obtain acuity estimates in children and non-responsive patients [5–14] that are unable to read a letter chart. The sVEP was initially used to determine acuity development in normal infants [6]. This technique demonstrated a gradual increase in acuity until a plateau was reached at about 8 months of age. At this age, the acuity estimate was not significantly different from adults. Other studies indicated that these acuity estimates were reliable from trial to trial in children and adults [5, 15]. In a review of the literature, two principal instruments [i.e., the Enfant (Neuroscientific Corp.) and the PowerDiva (Digital Instrumentation for Visual Assessment)] have been used to estimate the sVEP acuity. These two instruments use different methods to determine the visual acuity. One difference between the two instruments is the method of noise determination. The Enfant system uses the 95 % confidence interval of the signal magnitude to determine the noise level for estimating acuity. At a specific spatial frequency, if the 95 % confidence interval overlapped zero, then the signal at this frequency was not considered different from noise. The PowerDiva uses the magnitude of a Fourier frequency adjacent to 2X the signal frequency as noise [5, 6, 11, 16, 17]. Employing different noise level determinations may alter the acuity extrapolation. A previous study used a single data set obtained on the Enfant and applied the two different noise estimation techniques to determine the sVEP acuity [18]. The sVEP acuity estimates were not found to be different with the two techniques. However, because of other differences between the Enfant and the PowerDiva (e.g., software and hardware differences such as the size of the stimulus screen, data acquisition, or the input amplifier), different sVEP acuities could still be derived from the same subject. The stimulus monitor with the Enfant subtends a larger angle than the PowerDiva (see Methods below). The data acquisition with the Enfant has 12 bit accuracy with a sampling rate of 300 Hz compared to the PowerDiva, which has 16 bit accuracy with a sampling rate of 600 Hz. Additionally, the input filters and the signal
123
Doc Ophthalmol (2014) 129:105–114
gains are not the same with the two instruments (see Methods below). The primary purpose of this study was to compare sVEP acuity estimates obtained with two different sVEP systems: the Enfant and the PowerDiva. The sVEP acuity estimates are also compared to the logMAR chart acuities for individual subjects.
Methods Subjects Twenty-five adult (average age: 26.8 ± 3.81 (SD), range 21–35) subjects having monocular best corrected visual acuity of at least 0.10 logMAR were selected for this study (Table 1). The dominant eye with its best optical correction was always tested. A logMAR chart was used to determine visual acuity (Smart System II, M&S Technologies, Inc.). The ETDRS mode was used, and the letter-by-letter scoring method was employed. A power test based on a sample of 10 subjects indicated that a difference of 3 cpd (less than a 0.05 log unit change) could be detected for a sample size of 18 (alpha = 0.05, power = 0.80, assumed standard deviation for the estimate was 4). A 0.05 log unit change is less than the 95 % confidence interval for test–retest repeatability (i.e., ±0.10 logMAR) of psychophysically measured logMAR acuity chart data [19]. Subjects exhibiting ocular and/or systemic pathologies were excluded. Subjects taking medications that could affect vision were also excluded. There were no other exclusion criteria. The subject pool was limited to University employees and students. All subjects gave informed consent to participate in the study, and the research followed the tenets of the Declaration of Helsinki. The procedures were approved by the Institutional Review Board of the Marshall B. Ketchum University. Enfant stimulus A horizontal-oriented, sine wave grating that swept up the spatial frequency spectrum was used. The Enfant 4010 presented eleven spatial frequencies ranging from 3 to 36 cpd (3, 6, 9, 12, 15, 18, 20.1, 22.5, 25.8, 30, 36 cpd). Ten presentations of the stimuli were averaged together for a single acuity estimate. Two acuity estimates were made. The contrast was set at 80 % and the temporal reversal rate (square wave)
Doc Ophthalmol (2014) 129:105–114
107
Table 1 Subject data displaying the logMAR chart acuities, the Enfant and PowerDiva sVEP acuity estimates in cycles per degree (cpd), and the average Enfant and PowerDiva sVEP acuity estimates converted to logMAR values SubjectGender
Age
logMAR VA
(years) 1-M
31
-0.06
Enfant (cpd)
Enfant logMAR
Trial 1
Trial 2
Ave
28.2
29.3
28.8
0.018
PowerDiva (cpd)
PowerDiva logMAR
Trial 1
Trial 2
Ave
32.6
26.2
29.4
0.009
2-M
28
-0.06
28.8
29.1
29.0
0.015
31.0
30.4
30.7
-0.010
3-M
33
-0.10
22.0
20.3
21.2
0.152
17.1
15.3
16.2
0.267
4-M
28
-0.02
20.0
18.8
19.4
0.189
12.1
25.7
18.9
0.201
5-M
31
-0.10
23.3
21.3
22.3
0.129
27.4
18.9
23.2
0.112
6-M 7-F
31 26
-0.08 -0.10
28.4 38.3
22.1 31.3
25.3 34.8
0.075 -0.064
16.0 34.3
25.6 25.0
20.9 29.7
0.159 0.004
8-M
29
-0.08
23.0
25.8
24.4
0.090
14.8
18.6
16.7
0.254
9-F
25
-0.06
25.8
22.5
24.2
0.094
17.1
17.1
17.1
0.243
10-F
22
-0.08
32.2
23.5
27.9
0.032
17.9
24.1
21.0
0.155
11-F
25
-0.20
32.6
31.0
31.8
-0.025
30.8
37.1
34.0
-0.054
12-F
26
-0.06
22.5
23.9
23.2
0.112
29.2
26.4
27.8
0.033
13-M
21
-0.14
25.8
28.8
27.3
0.041
30.1
30.8
30.5
-0.007
14-F
25
-0.04
22.5
22.4
22.5
0.126
31.3
25.0
28.2
0.028
15-F
25
-0.22
28.0
21.0
24.5
0.088
26.8
32.9
29.8
0.003
16-M
24
-0.20
30.4
30.2
30.3
-0.004
40.4
48.9
44.7
-0.173
17-F
24
-0.06
23.3
23.2
23.3
0.111
26.9
28.0
27.4
0.039
18-M
24
?0.06
25.8
20.1
23.0
0.116
29.8
27.6
28.7
0.019
19-M
29
?0.04
25.8
18.1
22.0
0.136
29.4
32.9
31.1
-0.016
20-M
22
?0.06
23.0
30.0
26.5
0.054
25.6
23.5
24.6
0.087
21-M
28
-0.20
32.0
36.0
34
-0.054
16.9
15.2
16.1
0.272
22-M 23-M
32 24
-0.18 -0.18
30.0 21.4
36.0 24.4
33 22.9
-0.041 0.117
32.7 29.1
32.0 30.4
32.4 29.8
-0.033 0.004
24-M
22
-0.18
22.9
25.3
24.1
0.095
32.3
31.7
32.0
-0.028
25-F
35
?0.10
30.0
28.5
29.3
0.011
28.6
22.7
25.6
0.070
Mean
26.8
-0.086
26.6
25.7
26.2
0.064
26.4
26.9
26.6
SD
3.81
0.089
4.44
5.04
4.24
0.069
7.29
7.35
6.74
0.065 0.115
The data displayed in bold are estimated acuities based on a single spatial frequency (see Methods section)
7.5 Hz. The mean screen luminance was 100 cd/m2. The stimulus screen was viewed monocularly at 1.5 m [12.7° (H) 9 9.7° (V)]. Each spatial frequency was presented for 1 s, and there was 1 s of adaptation before data collection. PowerDiva stimulus The PowerDiva (software version 3.5) also presented a horizontal sine wave grating stimulus. The spatial frequency sweep contained ten spatial frequencies between 3 and 36 cpd (3, 6.67, 10.33, 14, 17.67, 21.33,
25, 28.67, 32.33, 36 cpd). Each spatial frequency was presented for 1 s with 1 s allowed for adaptation before data collection. The contrast was set at 80 % and the temporal reversal rate (square wave) set to 7.5 Hz. The mean screen luminance was 100 cd/m2. The monitor was viewed monocularly at 3 m [7o (H) 9 6.06o (V)]. Ten presentations of the stimuli were averaged together for a single acuity estimate. Two acuity estimates were made. The magnitude and phase plots, from which the acuity was derived, were determined with the recursive least squares (RLS) method [20].
123
108
Fig. 1 An example of the Enfant and PowerDiva sVEP data for subject 17. The arrows delineate the data included in the linear fits for acuity extrapolation. The extrapolated acuities are given
Recording technique Data were acquired for the Enfant and PowerDiva systems in a single session in random order. The recording electrode (Ag/AgCl) was placed 2 cm above the inion on the midline. The reference and ground electrodes were placed on the earlobes. The average electrode resistance was 9.5 ± 5.01 KOhms (Grass Electrode Impedance Meter, Model EZM3A, Quincy, MA). The electrodes were positioned once, and the sVEPs for both instruments were collected. The signal was amplified (10,000X for the Enfant and 50,000X for the PowerDiva), bandpass filtered (0.5–100 Hz for the Enfant and 0.3–100 Hz for the PowerDiva), and digitized. Acuity extrapolation Figure 1 displays an example of the sVEP magnitude data for both instruments for subject 17. The horizontal axis is the stimulus spatial frequency in cpd, and the vertical axis is the sVEP magnitude in microvolts. The linear fits to determine acuity are displayed for both sets of data. For the Enfant, the magnitude of the signal at each spatial frequency was determined from a discrete Fourier transform at two times the stimulus frequency (i.e., stimulus frequency = 7.5 Hz, so the analysis frequency was 15 Hz). A specific data point was considered to be noise if the 95 % confidence interval
123
Doc Ophthalmol (2014) 129:105–114
for the data at a spatial frequency overlapped with zero. The error bars are not shown in Fig. 1, but all data above 18 cpd had 95 % confidence intervals that overlapped zero. The acuity extrapolation technique for the Enfant has previously been described [13, 15, 16, 18]. Briefly, one of the two possible methods was employed. The first method fits a straight line to the data between the peak of the function (6 cpd in Fig. 1) and the highest spatial frequency above noise (18 cpd in Fig. 1). The data included in the fit are delineated by the arrows at 6 and 18 cpd. The line was extrapolated to the x-axis (i.e., zero magnitude) for acuity. In Fig. 1 for the Enfant data, the acuity is 23.3 cpd. The second method was used when a line could not be fit to the data. This could occur when there were multiple peaks in the sVEP function. In these circumstances, the highest spatial frequency above noise was taken as the acuity estimate. For the Enfant, 12 out of 50 acuity estimates were not determined by a linear fit (data in bold in Table 1). For the PowerDiva, the response signal was determined from a discrete Fourier transform at 15 Hz, similar to the Enfant. The noise estimate was taken as the average magnitude at 14.07 and 15.94 Hz from the discrete Fourier transform. The data were initially examined to determine whether at least one spatial frequency had a signal-to-noise ratio (i.e., SNR) greater than 3.0. All the sVEP functions met these criteria. The Tcirc statistic was also used to help determine noise [21]. The Tcirc statistic is the probability that a specific response is above noise. The lower the Tcirc statistic, the more likely the response is above noise. At least one spatial frequency in the data needed a Tcirc value \0.05. A similar fitting paradigm to that of the Enfant was used for the data from the PowerDiva. A line was fit from the peak of the function (6.67 cpd in Fig. 1) to the highest spatial frequency data above noise (21.33 cpd in Fig. 1). These spatial frequencies are identified with an arrow in Fig. 1. The data at each spatial frequency were considered to be noise if the SNR was one or less. For the PowerDiva data in Fig. 1, the SNR for the data above 21.33 cpd was below 1.0. The linear fit was extrapolated to the x-axis for the acuity estimate. Using this technique, the acuity for the PowerDiva data in Fig. 1 is 26.9 cpd. For the PowerDiva, 4 out of 50 acuity estimates could not be determined with a linear fit (bold in Table 1). For these
Doc Ophthalmol (2014) 129:105–114
109
data, the acuity estimate was taken as the highest spatial frequency above noise (i.e., similar to the Enfant method). The Enfant had more data sets that could not be fit with a line (i.e., 12 vs. 4) because it also had more data sets with multiple peaks. There was no consistent difference between data obtained with a linear fit and that determined by a single data point. The phase plots were also used for both instruments. The phase was constant or gradually changing within the range of spatial frequencies employed to determine acuity. Above the acuity limit, the phase varied randomly. Thus, by examining the phase plots, the approximate location of the acuity estimate could be determined. Statistical analysis Initially, a one-way ANOVA was used to determine whether there was a significant difference in acuity with the different instruments. A post hoc, pairedsample t test was then conducted to compare visual acuity measured with the logMAR acuity chart, the Enfant, or the PowerDiva. The significant p value was 0.05. Comparisons between data sets were also made with the intraclass correlation coefficient (ICC). The ICC determines how strongly two measurements agree by assessing the proportion of the total variance within the data that can be explained by the variance between the two measurements. Simply, the ICC indicates how well the two measurements fit the one to one line for the plotted data. The 95 % confidence intervals (i.e., 1.96 standard deviations) of the acuity differences were determined to indicate the repeatability across tests. The statistical tests were performed with the software package Minitab 16 Statistical Software (version 16.2.4 by Minitab, Inc.). Bland–Altman plots were also produced [22, 23]. The bias and the limits of agreement were determined from these plots.
Results The acuity estimates for the subjects are in Table 1. For the Enfant and the PowerDiva, the acuity estimates for the first trial, second trial, and average of the two trials are given in cycles per degree (cpd). These average acuities, in cpd, are then converted to logMAR units (30 cpd = 20/20 = 0.00 logMAR). The acuities were converted to logMAR units with
Fig. 2 The logMAR visual acuities determined from the chart (average = -0.086 ± 0.089 logMAR) and the Enfant (0.064 ± 0.069 logMAR, t = 8.1, p \ 0.001) for each subject. The diagonal line is the 1:1 line. If the data fell on the 1:1 line, the two acuity estimates would be similar. The graph demonstrates that the acuity estimates with the logMAR chart are significantly lower than those with the Enfant (p \ 0.001)
the following equation: logMAR = log(30/cpd) [24]. The column labeled ‘logMAR VA’ in Table 1 contains the acuities determined with the logMAR chart. The means and standard deviations of the data are given at the bottom of each column. A one-way ANOVA comparing logMAR VA, Enfant VA, and PowerDiva VA indicated that there was a significant difference (p \ 0.001, F = 21.99) between the groups. In the one-way ANOVA, the logMAR acuity threshold was the dependent variable and the different test groups (i.e., Enfant, PowerDiva, logMAR chart) were the independent variables. Post hoc, paired t tests indicated that the logMAR VA acuities [average = -0.086 ± 0.089 (SD) logMAR] and the Enfant acuities (0.064 ± 0.069 logMAR) were significantly different [t = 8.10, p \ 0.001, average difference = 0.15 ± 0.093 (SD)]. Similarly, the logMAR VA and the PowerDiva acuities (0.065 ± 0.115 logMAR) were significantly different [t = 5.77, p \ 0.001, average difference = 0.15 ± 0.131 (SD)]. The Enfant and PowerDiva acuities were not significantly different [t = 0.04, p = 0.966, average difference = 0.001 ± 0.116 (SD)].
123
110
Fig. 3 The logMAR visual acuities determined from the chart (average = -0.086 ± 0.089 logMAR) and the PowerDiva (0.065 ± 0.125 logMAR, t = 5.77, p \ 0.001) for each subject. The diagonal line is the 1:1 line. If the data fell on the 1:1 line, the two acuity estimates would be similar. The graph demonstrates that the acuity estimates with the logMAR chart are significantly lower than those with the PowerDiva (p \ 0.001)
Figure 2 displays the Enfant sVEP acuity estimate data plotted against the logMAR chart acuity data for each subject. The solid line is the 1:1 line. The majority of the data are above this line. Thus, the Enfant sVEP acuities are significantly worse (i.e., larger, positive numbers) than the logMAR chart acuities (p \ 0.001). This is similar to previous publications [15, 18]. The logMAR chart acuity has been reported to be 0.25 ± 0.122 [18] or 0.21 ± 0.089 logMAR [15] units better than the Enfant sVEP acuity estimate. For the normal subjects in this study (N = 25), the logMAR chart acuity is 0.15 ± 0.093 logMAR units lower than the Enfant sVEP acuity estimate. Figure 3 displays the PowerDiva sVEP acuity estimate data plotted against the logMAR chart acuity data for each subject. The solid line is the 1:1 line. Similar to the Enfant, the majority of the data lie above this line. Thus, the PowerDiva sVEP acuities are significantly worse than the logMAR chart acuities (p \ 0.001). The average difference between the PowerDiva sVEP acuity estimate and the logMAR
123
Doc Ophthalmol (2014) 129:105–114
Fig. 4 The logMAR visual acuity estimates determined from the Enfant (0.064 ± 0.069 logMAR) and the PowerDiva (0.065 ± 0.125 logMAR, t = 0.04, p = 0.966) for each subject. The diagonal line is the 1:1 line. The ICC is 0.25. The graph indicates that the two acuity estimates are similar (p = 0.966)
chart acuity is 0.15 ± 0.131 logMAR units. This is similar to the measurements with the Enfant. Figure 4 displays the PowerDiva sVEP acuity estimate data plotted against the Enfant sVEP acuity estimate data for each subject. The solid line is the 1:1 line. Much of the data are clustered around the 1:1 line. The paired t test did not find a significant difference between the two sets of data (p = 0.966). The intraclass correlation coefficient (ICC, r2) for the data was 0.25. Figure 5 displays the Bland–Altman plot for the Enfant sVEP acuity estimate data and the logMAR chart acuity data for each subject. The average of the two acuities for each subject is plotted on the horizontal axis, and the difference is plotted on the vertical axis. The data display a bias of ?0.15 (i.e., the line labeled MEAN), and the limit of agreement (±1.96 standard deviations) is ±0.182. Figure 6 displays the Bland–Altman plot for the PowerDiva sVEP acuity estimate data and the logMAR chart acuity data for each subject. The format is the same as Fig. 5. Similar to Fig. 5, the bias is ?0.15 and the limit of agreement is ±0.257. Figure 7 displays the Bland– Altman plot for the PowerDiva sVEP acuity estimate data and the Enfant sVEP acuity estimate data for each subject. The data do not display a bias (an average
Doc Ophthalmol (2014) 129:105–114
Fig. 5 A Bland–Altman plot comparing the logMAR chart and Enfant sVEP acuity estimates. The bias of 0.150 (i.e., the difference between zero and the MEAN line) indicates that the acuity estimates measured with the Enfant are poorer than with the logMAR chart. The limit of agreement is ±0.182 (i.e., ±1.96 standard deviations from the MEAN)
Fig. 6 A Bland–Altman plot comparing the logMAR chart and PowerDiva sVEP acuity estimates. The bias (0.151) indicates that the acuity estimates measured with the PowerDiva are poorer than with the logMAR chart. The limit of agreement is ±0.257
difference of -0.001), and the limit of agreement is ±0.227. The intra-session repeatability of the Enfant and PowerDiva data from trial 1 to trial 2 was also
111
Fig. 7 A Bland–Altman plot comparing the Enfant and PowerDiva sVEP acuity estimates. The bias (-0.001) indicates that the acuity estimates measured with the two instruments are similar. The limit of agreement is ±0.227
determined (data in Table 1). The intra-session, test– retest 95 % confidence interval and bias for the Enfant were ±0.140 and 0.018 logMAR units. The 95 % confidence interval and bias for the PowerDiva were ±0.210 and -0.011 logMAR units. Figure 8 displays the current data along with data from previous publications employing the Enfant or the PowerDiva. The majority of these data were taken from a previous publication (Fig. 4; [18]). The open symbols in the figure are data taken from normal adults with good acuity (i.e., approximately 0.00 logMAR). The filled symbols are the data for subjects with decreased visual acuity. The data from Ridder et al. [16] are the first second of the sVEP data found in Table 1 and Fig. 3. The data from Katsumi et al. [25] are extrapolated from their Figs. 2 and 3 at the zero diopter level for subjects with good acuity. The data from Arai et al. [26] are extrapolated from Fig. 2 for the subjects with good acuity (20/20 or better acuity) or subjects with decreased acuity (20/30–20/200). The data from Yadav et al. [27] are the average of 6 normal adult subjects that they report in their Fig. 3a (category C2, 100 cd/m2). They used the PowerDiva system. The majority of these data for the subjects with normal acuity lie above the solid line (i.e., the 1:1 line) indicating that the logMAR chart acuity is lower than the sVEP acuity estimates. The two dashed lines in Fig. 8 that parallel the solid line (i.e., the 1:1 line)
123
112
Doc Ophthalmol (2014) 129:105–114
Fig. 8 A comparison of the data from this study with previous published data employing the Enfant or the PowerDiva instrument. See text for details
are one octave above or below the 1:1 line. The dotted lines are two octaves above or below the 1:1 line. The majority of the Enfant sVEP acuity estimate data for normal subjects are clustered about 1 octave above the 1:1 line. The sVEP acuity estimate data from the PowerDiva (i.e., open stars and hexagon symbols) do not appear to be different from the Enfant for the normal subjects (i.e., the other open symbols). It was previously noted that the subjects with decreased visual acuity overlay the 1:1 line better than the subjects with normal acuity [18]. The majority of the subjects with decreased visual acuity (49/56) have acuity estimates within 1 octave of the 1:1 line.
Discussion There are several different instruments that have been used to measure the sVEP acuity. However, there are no published studies which compare the sVEP acuity estimates of one instrument with another. The present study compared the sVEP acuities from two commonly referenced instruments. This allows for direct comparisons of sVEP acuity estimate results from different published papers using these two instruments. The results demonstrate that there is a statistical difference between the acuities measured on the
123
logMAR chart versus those estimated with the Enfant (p \ 0.001) or PowerDiva (p \ 0.001). Both instruments give acuities that are about 0.15 logMAR units worse than the logMAR chart acuity for normal subjects. The sVEP acuities measured with the Enfant, and the PowerDiva are not statistically different (p = 0.966). The sample size test indicated that 25 subjects were sufficient to detect a difference of 3 cpd or about 0.05 logMAR units. This is a smaller difference than has been reported for the test–retest repeatability for the logMAR acuity chart [19]. Comparisons with previous data demonstrate that the sVEP acuities measured with the Enfant are consistent across several laboratories (see Fig. 8). Furthermore, these comparisons demonstrate that the sVEP acuities determined with the PowerDiva are also similar across laboratories, as well as, with the Enfant sVEP acuity measurements. Thus, direct comparisons of sVEP acuity estimates of normal subjects can be made with the Enfant and PowerDiva. An intersession, test–retest repeatability study with the Enfant resulted in a 95 % confidence interval of ±0.11 logMAR units and a bias of -0.003 when two sVEP acuity estimates are averaged together per session [15]. This previous study also reported a 95 % confidence interval with the logMAR acuity chart of ±0.07 logMAR units and a bias of 0.01 for the
Doc Ophthalmol (2014) 129:105–114
normal adults [15]. This indicates that the repeatability for the sVEP acuity estimate with the Enfant is nearly as good as the logMAR acuity chart (i.e., the gold standard for acuity measurements). The current study comparing the Enfant and the PowerDiva resulted in a 95 % confidence interval of ±0.227 logMAR units and a bias of -0.001. Thus, the sVEP repeatability is best when the same instrument is used to repeat the test and decreases when repetitions are made across instruments. The intra-session, test–retest repeatability with the Enfant is reported to have a 95 % confidence interval of about ±0.19 logMAR units (±0.191 and ±0.186 for visit one and visit two from Ridder et al. [15]). The bias for visit one and visit two was 0.010 and -0.019 [15]. In the present study, the intra-session, test–retest 95 % confidence interval and bias for the Enfant were ±0.140 and 0.018 logMAR units. These values agree with the previously published data [15]. The 95 % confidence interval and bias for the PowerDiva were ±0.210 and -0.011 logMAR units. Thus, there was no bias with either instrument, and the repeatability was slightly worse with the PowerDiva. The different stimulus size between the Enfant [12.7° (H) 9 9.7° (V)] and the PowerDiva [7° (H) 9 6.06° (V)] might be expected to result in a difference in acuity estimates between the two instruments. The larger stimulus size for the Enfant would result in a greater number of stimulus cycles present for each spatial frequency. However, the decreased number of stimulus cycles with the PowerDiva instrument may only have an effect at low spatial frequencies [28]. The lower spatial frequencies, with the fewer stimulus cycles, would result in lower visual evoked potential magnitudes. An example of this effect is seen in Fig. 1. The response magnitude for the lower spatial frequencies is reduced for the PowerDiva in comparison with the Enfant. A similar difference was seen for 17 of the 25 subjects. One subject displayed greater magnitudes with the PowerDiva, and the remaining seven subjects had similar magnitudes with the two instruments. However, since the acuity is extrapolated from the higher spatial frequencies, this decreased signal magnitude for the lower spatial frequencies does not affect the results. Similar observations have been made by Tyler et al. [29] and Almoqbel et al. [30]. Different stimulus screen sizes will have an additional affect if sVEP comparisons are made in infants
113
or children. Since the fixation of infants and children may be more variable than behaving adults, the smaller screen size may result in more variable data. Thus, the present results in adults may not be translatable to infants if these screen distances are used. In conclusion, the sVEP acuity estimates for the Enfant and the PowerDiva are not statistically different for normal subjects. Thus, comparisons between the two instruments can be made. The present data and several published papers demonstrate that both instruments result in poorer acuity estimates then the logMAR acuity chart for normal subjects. Comparisons with the two instruments still need to be made with subjects with decreased acuity. Conflict of interest All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or nonfinancial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.
References 1. Bailey IL, Lovie JE (1976) New design principles for visual acuity letter charts. Am J Optom Physiol Opt 53(11): 740–745 2. Taylor HR (1978) Applying new design principles to the construction of an illiterate E chart. Am J Optom Physiol Opt 55(5):348–351 3. Ferris FL 3rd, Kassoff A, Bresnick GH, Bailey I (1982) New visual acuity charts for clinical research. Am J Ophthalmol 94(1):91–96 4. Strong G, Woo GC (1985) A distance visual acuity chart incorporating some new design features. Arch Ophthalmol 103(1):44–46 5. Norcia AM, Tyler CW (1985) Infant VEP acuity measurements: analysis of individual differences and measurement error. Electroencephalogr Clin Neurophysiol 61(5):359–369 6. Norcia AM, Tyler CW (1985) Spatial frequency sweep VEP: visual acuity during the first year of life. Vision Res 25(10):1399–1408 7. Norcia AM, Tyler CW, Piecuch R, Clyman R, Grobstein J (1987) Visual acuity development in normal and abnormal preterm human infants. J Pediatr Ophthalmol Strabismus 24(2):70–74 8. Hamer RD, Norcia AM, Tyler CW, Hsu-Winges C (1989) The development of monocular and binocular VEP acuity. Vision Res 29(4):397–408 9. Gottlob I, Fendick MG, Guo S, Zubcov AA, Odom JV, Reinecke RD (1990) Visual acuity measurements by swept
123
114
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Doc Ophthalmol (2014) 129:105–114 spatial frequency visual-evoked-cortical potentials (VECPs): clinical application in children with various visual disorders. J Pediatr Ophthalmol Strabismus 27(1):40–47 Gottlob I, Wizov SS, Odom JV, Reinecke RD (1993) Predicting optotype visual acuity by swept spatial visualevoked potentials. Clin Vision Sci 8:417–423 Katsumi O, Denno S, Arai M, De Faria JL, Hirose T (1997) Comparison of preferential looking acuity and pattern reversal visual evoked response acuity in pediatric patients. Graefes Arch Clin Exp Ophthalmol 235(11):684–690 Birch EE, Hoffman DR, Uauy R, Birch DG, Prestidge C (1998) Visual acuity and the essentiality of docosahexaenoic acid and arachidonic acid in the diet of term infants. Pediatr Res 44(2):201–209 Ridder WH 3rd, Rouse MW (2007) Predicting potential acuities in amblyopes: predicting post-therapy acuity in amblyopes. Doc Ophthalmol 114(3):135–145. doi:10.1007/ s10633-007-9048-y Ghasia F, Brunstom J, Tychsen L (2009) Visual acuity and visually evoked responses in children with cerebral palsy: gross motor function classification scale. Br J Ophthalmol 93(8):1068–1072. doi:10.1136/bjo.2008.156372 Ridder WH 3rd, Tong A, Floresca T (2012) Reliability of acuities determined with the sweep visual evoked potential (sVEP). Doc Ophthalmol 124(2):99–107. doi:10.1007/ s10633-012-9312-7 Ridder WH 3rd, McCulloch D, Herbert AM (1998) Stimulus duration, neural adaptation, and sweep visual evoked potential acuity estimates. Invest Ophthalmol Vis Sci 39(13):2759–2768 Riddell PM, Ladenheim B, Mast J, Catalano T, Nobile R, Hainline L (1997) Comparison of measures of visual acuity in infants: teller acuity cards and sweep visual evoked potentials. Optom Vis Sci 74(9):702–707 Ridder WH 3rd (2004) Methods of visual acuity determination with the spatial frequency sweep visual evoked potential. Doc Ophthalmol 109(3):239–247 Arditi A, Cagenello R (1993) On the statistical reliability of letter-chart visual acuity measurements. Invest Ophthalmol Vis Sci 34(1):120–129
123
20. Tang Y, Norcia AM (1995) An adaptive filter for steadystate evoked responses. Electroencephalogr Clin Neurophysiol 96(3):268–277 21. Victor JD, Mast J (1991) A new statistic for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 78(5):378–388 22. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307–310 23. Bland JM, Altman DG (1999) Measuring agreement in method comparison studies. Stat Methods Med Res 8(2):135–160 24. Thompson DA, Moller H, Russell-Eggitt I, Kriss A (1996) Visual acuity in unilateral cataract. Br J Ophthalmol 80(9):794–798 25. Katsumi O, Arai M, Wajima R, Denno S, Hirose T (1996) Spatial frequency sweep pattern reversal VER acuity vs Snellen visual acuity: effect of optical defocus. Vision Res 36(6):903–909 26. Arai M, Katsumi O, Paranhos FR, De Faria JML, Hirose T (1997) Comparison of Snellen acuity and objective assessment using the spatial frequency sweep PVER. Graefes Arch Clin Exp Ophthalmol 235(7):442–447 27. Yadav NK, Almoqbel F, Head L, Irving EL, Leat SJ (2009) Threshold determination in sweep VEP and the effects of criterion. Doc Ophthalmol 119(2):109–121. doi:10.1007/ s10633-009-9177-6 28. Savoy RL, McCann JJ (1975) Visibility of low-spatial-frequency sine-wave targets: dependence on number of cycles. J Opt Soc Am 65(3):343–350 29. Tyler CW, Apkarian P, Levi DM, Nakayama K (1979) Rapid assessment of visual function: an electronic sweep technique for the pattern visual evoked potential. Invest Ophthalmol Vis Sci 18(7):703–713 30. Almoqbel FM, Yadav NK, Leat SJ, Head LM, Irving EL (2011) Effects of sweep VEP parameters on visual acuity and contrast thresholds in children and adults. Graefes Arch Clin Exp Ophthalmol 249(4):613–623. doi:10.1007/ s00417-010-1469-8
