Abstract
Purpose.:
Impaired fixation stability is associated with reduced reading speed. In previous research, fixation stability has been assessed using an infrared eye tracker or a confocal scanning laser ophthalmoscope. The new MP-1 microperimeter from Nidek Technologies (Padova, Italy) provides another option for the assessment of fixation. Here the authors compare fixation stability values measured using the MP-1 microperimeter and the Rodenstock scanning laser ophthalmoscope (SLO; Rodenstock GmbH, Munich, Germany) in persons with and without diabetic maculopathy.
Methods.:
Sixteen normally sighted volunteers and 21 patients with diabetic maculopathy were recruited. Fixation stability was recorded monocularly on the SLO and the MP-1 in counterbalanced order while participants fixated a red 1° cross. Fixation data collected from each instrument were used to calculate a bivariate contour ellipse area (BCEA) that encompassed 68% of fixation points.
Results.:
For control subjects, MP-1 BCEA values were larger than SLO by 0.25 log min arc2, though the difference was small (10%) and of borderline significance (MP-1, 2.51 log min arc2; SLO, 2.26 log min arc2; P = 0.06). In patients with diabetic maculopathy there was no significant difference between MP-1 and SLO values (MP-1, 2.94 log min arc2; SLO, 2.90 log min arc2; P = 0.88).
Conclusions.:
No significant difference was found in BCEA values from the SLO and MP-1 in control subjects and patients with diabetic maculopathy. The authors suggest that the similarity between BCEA values, together with the consistent and reliable operation of the MP-1, make it a useful and viable alternative to the SLO in the assessment of fixation.
In normal vision, an eye fixating a static target does not remain stationary
1 ; it constantly makes small involuntary eye movements such as microsaccades, drifts, and tremors.
2 Elimination of such movements would cause our perception of a stationary target to fade completely
3–6 . However, excessive instability degrades visual resolution
7 and may interfere with the performance of everyday tasks such as reading.
8 Eye conditions affecting central vision are known to impair fixation.
9–14 Therefore, awareness of a patient's ability to fixate is important when considering functional vision.
Although it is no longer commercially available, one well-established instrument in the assessment of fixation is the Rodenstock scanning laser ophthalmoscope (SLO; Rodenstock GmbH, Munich, Germany). Since its introduction in the early 1980s it has been used in numerous studies of fixation
9–11,14–18 and has proven to be particularly useful for the examination of fixation in those with eye disease.
9,11,13–17,19 The instrument was not specifically designed to measure fixation, but different methods have been described that allow its quantification.
11,14,15,19 One established method follows that first described by Steinman, whereby the position of each fixation point is plotted on Cartesian axes and the elliptical area encompassing a given percentage of points is calculated.
20 This bivariate contour ellipse area (BCEA) presents a value of fixation stability, with smaller values indicating more stable fixation.
A new instrument, the Nidek MP-1 microperimeter (Nidek Technologies, Padova, Italy), has been designed with fixation stability assessment capability. It allows fixation to be assessed during a microperimetric examination or as an isolated assessment. The MP-1 offers classification of fixation stability based on the system described by Fujii et al.,
21 whereby fixation is termed
stable if >75% of fixation points fall within a 2° diameter circle centered on the gravitational center of all fixation points,
relatively unstable if <75% of fixation points fall within a 2° circle but >75% are located within a 4° diameter circle, and
unstable if <75% of all fixation points fall within a 4° diameter circle.
21 The lack of scientific foundation to this classification has been criticized in the literature.
22 However, it is possible to extract raw fixation data from the MP-1, thus allowing fixation points to be plotted and characterized by a BCEA value, as described. Recently we published data showing a lack of correlation between reading speed and fixation stability as classified by the inbuilt MP-1 strategy but a stronger correlation between reading performance and fixation quantified by calculating a BCEA.
23
Because both the SLO and the MP-1 are used in the evaluation of fixation, an understanding of their comparability is vital. Different methods of assessment are known to produce different BCEA values; for example, in healthy young persons, SLO BCEA values are up to 2.25 times smaller than those from a head-mounted eye tracker system.
24 Given that the factors influencing measurement differ between those with steady fixation and those with poor fixation, we feel it important to compare fixation measurements not only in healthy young subjects but also in patients with poorer fixation.
Diabetic eye disease is the leading cause of blindness in the working-age population of the United Kingdom, and the incidence of diabetes mellitus continues to rise in many developed countries.
25–27 Although fixation behavior has been widely studied in macular disease,
9,11–13,19,22,23,28 the impact of diabetic maculopathy on fixation stability is less well understood.
14–17,29,30 A better understanding of fixation in the presence of diabetic maculopathy and its impact on visual function is necessary for clinicians involved in the visual rehabilitation of patients and for those responsible for the strategic planning of such services.
Here we evaluated fixation stability in persons with and without diabetic maculopathy using two different instruments, the Rodenstock confocal SLO and the Nidek MP-1 microperimeter.
We used a scanning laser ophthalmoscope (SLO-101; Rodenstock GmbH) consisting of a helium-neon laser of wavelength 632.8 nm that produces the stimuli and an infrared laser of 780 nm that simultaneously images the fundus according to a confocal principle.
32 Images were captured on a professional digital video recorder at a resolution of 768 × 576 pixels (model BR-DV600E; JVC, Yokohama, Japan) and a frequency of 12.5 Hz.
Software provided with the SLO (scotometry module) was used to produce a 1° red cross fixation target. Subjects were asked to view the center of the cross until 10 seconds of relatively blink free data were obtained. The digital video recorder simultaneously recorded fundus images throughout.
Video images were digitized using a frame grabber (Orion Frame Grabber; Matrox, Montreal, Canada) and retinal position was retrospectively analyzed using software developed in house. The software automatically tracked fundus features within a delineated square of predetermined location on the retinal image at 12.5 Hz, producing x and y coordinates of its position in pixels. If tracking was lost, the square jumped to the extremity of the image, and the related coordinates had unrealistic values. Any such coordinates were manually deleted; complete trials were discarded if >20% of coordinates were deleted for this reason.
The SLO was calibrated to quantify the amount of retinal movement shown in the captured fundus image. A semi-silvered mirror was placed in front of the SLO, allowing external targets to be viewed on the same visual axis as the SLO. Two cross fixation targets of known horizontal separation were placed on the wall parallel to the observer's line of sight. The distance between the targets and the semi-silvered mirror was recorded, and, therefore, the angular separation of the crosses could be calculated. The observer was instructed to look steadily at one target for 10 seconds and then switch to the second target for another 10 seconds. Fundus images were simultaneously recorded. The position of a retinal landmark was tracked at a frequency of 12.5 Hz, recording eye position in two-dimensional pixel coordinates. The horizontal movement of the retinal image between the two positions in image pixels was determined and used in a simple transformation, with the angular separation of the two crosses to describe the retinal motion seen on SLO recording in terms of visual angle. The resultant conversion factor, 1 pixel:2.6 min arc, was used in all SLO BCEA calculations. Because pixels within the central 5° of the SLO screen have been shown to be square with respect to the retina,
33 this conversion factor is applicable in both the horizontal and the vertical planes.
The MP-1 microperimeter (Nidek Technologies) was used. This instrument comprises an infrared fundus camera and a liquid crystal display (LCD) that presents stimuli to the observer.
A 1° red cross was displayed on the center of the LCD screen, and subjects were asked to view the center of the cross. Standard fixation measurement was performed using the techniques recommended by Nidek. First, a reference image of the fundus was captured, and a reference area of high-contrast retinal features was selected. During the examination, inbuilt software (MP-1 SW 1.7) tracked this reference area, calculating any shift in its position between the reference image and subsequent frames within the image at a frequency of 25 Hz, producing x and y coordinates of retinal position in degrees of visual angle. If tracking of the real-time image failed, coordinates were not generated until tracking was resumed. Nidek calculates the degree/pixel ratio of each individual instrument, found to be 1:15.714 when recently serviced. The MP-1 reports the total time of a fixation trial and the tracked time; therefore, the amount of time during which tracking fails is known. Ten seconds of tracked data were collected and exported for offline analysis.
Thirty-seven participants were recruited. The 16 normally sighted volunteers (6 men, 10 women; age range, 21–41 years) had visual acuity of 0.0 logMAR (6/6, 20/20) or better. The 21 (14 men, 7 women; age range, 24–77 years) patients with diabetic maculopathy had visual acuity between 0.8 and 0.0 logMAR (6/38, 20/125 to 6/6, 20/20). All participants had refractive error of between −6 and +4 DS spherical equivalent.
Complete data were obtained from 28 subjects (14 in each group). The nine instances of incomplete data—four instances of He-Ne laser failure and five instances during which tracking software failed to track recorded images accurately—all arose from technical problems with the SLO. Analysis was conducted on complete data only.
On average, no significant difference in BCEA values between the SLO and the MP-1 was observed in subjects with normal vision and patients with diabetic maculopathy. Somewhat surprisingly, the correlation between BCEA values from the two instruments was weak in both study groups but was slightly stronger over the larger range of BCEA values recorded in patients with diabetic maculopathy.
A previous study reported moderate agreement between SLO and MP-1 fixation assessments in eyes with retinal disease, but the means of quantifying fixation differed significantly from our method. Fixation was classified according to the MP-1 classification system described
21 and was compared for agreement using Cohen's κ coefficient. In the case of the SLO data, fixation was stable if the SD of fixation points around the mean fixation point was less then 0.6°, relatively unstable if the SD was between 0.6° and 1.2°, and unstable if the SD was greater than 1.2°.
35 Additionally, fixation was assessed during microperimetric examination compared with our isolated fixation task. Earlier work by the same author noted steadier fixation during an isolated task than during microperimetry.
15
In contrast to our results, a significant correlation between MP-1 and SLO fixation assessment in patients with macular disorders has been observed; however, the method of quantification differed considerably between each instrument. MP-1 fixation was quantified by mean extent, which is double the square root of the product of the
x and
y degree positions of fixation points, whereas SLO fixation was calculated as the percentage of fixation points within the central 2° on SLO measurement.
28 Because of notable differences in study design, the findings from these studies cannot be directly compared with our results.
To make direct comparisons between our BCEA measurements and other published values, we examined the standard deviations of fixation points along the horizontal and vertical axes. In normal vision, while fixating a stationary target, fixation points tend to spread out horizontally more than vertically.
10,11,22,36,37 The mean horizontal and vertical SDs of our SLO and MP-1 data from subjects with normal vision conformed to this description, with average horizontal and vertical SDs of 8 min arc and 6 min arc, respectively, on SLO and 9 min arc and 7 min arc, respectively, on MP-1. Our SLO standard deviations fall within the narrow range cited by Culham (4–8 min arc horizontally and 3–7 min arc vertically)
10 and toward the lower ranges quoted by Rohrschneider
11 (8–88 min arc horizontally and 6–65 min arc vertically) and Timberlake
37 (4–38 min arc horizontally and 4–18 min arc vertically). Data published in 2008 reported horizontal and vertical ranges of 4 to 9 min arc and 3 to 6 min arc, respectively, on the MP-1 for 10 experienced observers with normal vision.
22 Although our mean values fall slightly above these top limits, we suggest this to be a consequence of the relative inexperience of our observers.
In those with diabetic maculopathy, the mean horizontal and vertical standard deviations were 29 min arc and 22 min arc, respectively, on SLO and 34 min arc and 59 min arc, respectively, on MP-1. We do not know of any previous studies that have quantified fixation in patients with diabetic maculopathy in terms of BCEA or horizontal and vertical SDs. Our findings agree with the value of 45 min arc given for a standard deviation around a mean fixation point in eyes with clinically significant diabetic macular edema.
15 Fixation characteristics in this patient group are not well defined. One recent study looking at fixation in this population using the MP-1 found stable fixation in more than 70% of eyes,
29 whereas another observed found unstable fixation in most (60%) eyes.
30
To discover whether the different sampling rates of the two instruments, 12.5 Hz on SLO and 25 Hz on MP-1, should influence the size of the BCEA, we under sampled three MP-1 data files. Every second frame was removed, thereby simulating a sampling rate of 12.5 Hz, equal to that of the SLO. Because nearly equivalent values were found in each case (full data sets: 212, 80 and 514 min arc2; half data sets: 213, 80 and 528 min arc2), it is unlikely to be a source of error.
We chose fixation durations of 10 seconds because longer durations of blink-free data are difficult to record. Because previous work revealed no systematic variation over time in BCEAs calculated from the first 10 seconds of each of 8 consecutive minutes of fixation, we believe 10-second fixation trials to be of adequate length for BCEA calculation.
38
It has been reported that the SLO raster is distorted in a trapezoidal manner such that the raster is 10% larger at the bottom than the top.
39 Misalignment between the infrared imaging system and the LCD screen of the MP-1 has also been described. Spatial alignment errors of 0.5° have been observed between recorded retinal position and the true retinal location stimulated (Woods RL, et al.
IOVS 2007;48:ARVO E-Abstract 144). Because our subjects viewed a single fixation target in a fixed central position, these distortions are unlikely to meaningfully influence our findings.
In summary, fixation stability values measured using the SLO and the MP-1 did not differ significantly on average. Because fixation stability is of more clinical interest in patients with macular disease, we were encouraged to find such small differences in the values of patients with diabetic maculopathy. As described earlier, the collection of complete data was hampered by persistent technical problems with the SLO. In contrast, the MP-1 was operational throughout. The Rodenstock is no longer commercially available, difficult to maintain, and expensive to service. The MP-1 is backed up by technical and maintenance support from Nidek distributors. We suggest that the similarity found in BCEA values and the consistent and reliable operation of the MP-1 make it a useful and viable alternative to the SLO in the assessment of fixation.
Supported by Fight for Sight Clinical Fellowship Grant 1775/76.
Disclosure:
H.M.P. Dunbar, None;
M.D. Crossland, None;
G.S. Rubin, None
The authors thank William Seiple (Lighthouse International) for his assistance with validation of the BCEA calculation.