Field Analyser (HFA) and the Medmont. M600 Automated Perimeter (M600). Clinical experience has shown that the sensitivity data given by the HFA are gen-.
C L I N I C A L
A N D
E X P E R I M E N T A L
Automated perimeter conversion Pye, Herse, Nguyen, Vuong and Pham
OPTOMETRY
Conversion factor for comparison of data from Humphrey and Medmont automated perimeters David Pye MOptom FCLSA Peter Herse PhD FAAO Ha Nguyen BOptom Lan Vuong BOptom Quoc Pham BOptom School of Optometry The University of New South Wales
Accepted for publication: 3 September 1998
Background: Clinical experience has shown that the sensitivity indices reported by the Humphrey Field Analyser (HFA) are generally higher than those given by the Medmont Automated Perimeter (M600). It is the purpose of this paper to determine a conversion factor for the two perimeters and to confirm this prediction using clinical data. Theory predicted that HFA sensitivity - 5 dB = M600sensitivity. Methods: Sensitivity versus eccentricity profiles were measured over the central visual field on 10 young subjects using both perimeters. Results: Both the HFA and the M600 operate within the realms of the Weber law and measure similar Weber fractions. The sensitivity profiles had similar slopes (about - 0.2 dB/degree) and were separated by about six decibels with the HFA reporting higher sensitivity values. This result confirmed the theoretical prediction. Conclusion: The difference in threshold sensitivities between the two perimeters is a result of differences in scaling factors and instrument luminances. A suggested clinical conversion factor is to subtract 5 dB from the HFA data to approximate those of the M600. (Clin Exp Optom 1999; 82: 1: 11–13)
Key words: Humphrey Field Analyser, Medmont M600 automated perimeter, perimetry, Weber fraction
The two most commonly used automated perimeters in Australia are the Humphrey Field Analyser (HFA) and the Medmont M600 Automated Perimeter (M600). Clinical experience has shown that the sensitivity data given by the HFA are generally higher than those provided by the M600. It would be useful to be able to compare directly the data of one perimeter against those collected on the other. Inter-perimeter conversion factors have been suggested.1,2 However, the conversion factors given in these reports are based on either small sample sizes or outdated perimeters. It is the purpose of this study to com-
pare the sensitivity data of two current perimeters and to suggest a simple clinical conversion factor to allow comparison of data between instruments. Examination of the design characteristics of the two perimeters may enable determination of a theoretical conversion factor. Relevant instrument parameters are discussed.
Stimulus characteristics The stimuli used in the two perimeters are similar in size (Goldmann stimulus III), testing distance (66 cm HFA; 70 cm M600) and duration time (200 msec). While the stimulus delivery methods are different (HFA projected stimulus; M600 diffused
light emitting diode) this difference is not considered likely to produce a major clinical effect.
Threshold procedures Thresholds are calculated in both perimeters using staircase paradigms, where the intensity of the stimulus is increased in a step-wise manner until the subject can just perceive the stimulus. A step reversal is then used to confirm this threshold. The HFA uses 4 dB steps to approach threshold and 2 dB steps to confirm threshold, while the M600 uses 6 dB steps to approach threshold and 3 dB steps to confirm threshold. These slight differences
Clinical and Experimental Optometry 82.1 January–February 1999
11
82-1PYE.P65
11
1/07/99, 15:32
Automated perimeter conversion Pye, Herse, Nguyen, Vuong and Pham
in staircase paradigm are not thought sufficient to produce significant differences in threshold end points, though the larger steps used by the M600 enable faster threshold determination. 1 The faster thresholding advantage of the M600 has since been eliminated in more recent versions of the HFA, which include fast adaptive thresholding software (for example, fast SITA).
ground luminance. The overall effect could be summarised as HFA sensitivity - 5 dB = M600 sensitivity. An experiment was performed to confirm the theoretical prediction. METHOD
The threshold sensitivity reported by the perimeters indicates the final attenuation of the maximal stimulus luminance. Yet, as the perimeters have differing maximal stimulus luminances, the same reported threshold sensitivity corresponds to markedly different stimulus luminances. For example a 1,000 asb stimulus would be reported as a 10 dB sensitivity in the HFA and a zero decibel sensitivity in the M600. If the background luminances of the two perimeters were identical, the M600 sensitivity value could be approximated by subtracting 10 dB from the HFA sensitivity data.
Automated threshold perimetry was performed on 10 optometry students aged 21 to 34 years (mean 25 years). All subjects were free of ocular pathology, were familiar with the test and had corrected visual acuities of 6/6 or better. The subjects performed both the 30–2 threshold procedure on the HFA 620 and the central 100 point threshold procedure on the M600. The order of the testing was randomly chosen to eliminate any procedural bias. Allocation of right or left eye testing was also randomised. All subjects were corrected with the appropriate distance correction. Mean sensitivity values were calculated at three, 10, 15, 20 and 28 degrees for the HFA 620 and three, 10, 15, 22 and 30 degrees for the M600.
Background luminance
RESULTS
Scale effects
However, the background luminances of the perimeters are different, with the background luminance (L) of the HFA being 3.15 times greater than that of the M600. Perimetry is based on increment thresholds or the Weber law. The Weber law (∆L/L = K) operates over a wide range of photopic conditions and states that the increment threshold (∆L) increases in direct proportion to an increase in the background luminance (L). As the HFA background luminance is 5 dB (or 3.15 times) brighter than that of the M600, the Weber law would suggest that the increment threshold (∆L) of the HFA would be 5 dB brighter than that of the M600.
Theoretical conversion factor From the previous discussion we could predict that to approximate the M600 data the HFA sensitivity data would need to be: 1. decreased by 10 dB to account for the scaling difference 2. increased by 5 dB (spot made dimmer) to account for the difference in back-
Hill of vision Sensitivity will be defined as the attenuation of the maximal stimulus intensity recorded by the perimeter. Figure 1 shows the normal variation in sensitivity with eccentricity measured using both the HFA and M600 perimeters. Clinically, we think of these curves as Traquair’s ‘hill of vision’. Both perimeters show sensitivity decreasing with eccentricity. The slopes of the central fields are similar at - 0.25 dB/ degree for the HFA and - 0.18 dB/degree for the M600. These values agree with previous reports of the slope of the normal ‘hill of vision’ over this eccentricity range (-0.20 1 to -0.25 3 dB/degree). The curves are separated by a relatively constant sensitivity difference, with the HFA sensitivity data being about 6 dB greater than the M600 data at each eccentricity. This result agrees with the theoretical prediction and confirms a result similar to that reported for the central visual field in an earlier study (5 dB 1).
Weber fraction The Weber fraction can be used to eliminate the confounding of effect of differing background and stimulus luminances. The Weber fraction is the increment threshold (asb) divided by the background illumination (asb). These data are plotted in Figure 2. Data from an earlier study are also included to provide support for the experimental result.4 It can be seen that the Weber fractions are similar for both perimeters across the visual field. The 0.5 ND difference in background luminance between the instruments was not sufficient to disrupt the Weber law. This interpretation is supported by a previous report in which the Weber law was shown to be maintained after both stimulus and background luminance were reduced using a 0.6 ND filter.4 Yet, when does the Weber law break down and normal perimetry become invalid? The perimetric Weber fraction has been shown to remain stable over a wide range of background luminances from 10 to 45 asb 5,6 with no significant increase in Weber fraction occurring until the background luminance decreases to about 1 asb.4 This result is shown in Figure 2 as the effect of per forming perimetry through a 2.0 ND filter. The usefulness of variation in background luminance level has been a point of debate in perimetry for many years. Some researchers suggest that low luminance perimetry (for example, 4 asb) has diagnostic advantages in detecting rod dysfunction.6,7 However, the adaptation time needed before performing low luminance perimetry makes this technique clinically difficult. 8 Conversely, high luminance perimetry (for example, 315 asb) has been shown to give more uniformity in scotoma characterisation 9 and may enable detection of more subtle field losses in glaucoma or neurological conditions.10 CONCLUSION This paper confirms and extends the previous work of Vingrys and Helfrich.1 The similarity in Weber fractions for both the HFA and the M600 indicates that both perimeters are working within the range
Clinical and Experimental Optometry 82.1 January–February 1999
12
82-1PYE.P65
12
1/07/99, 15:32
Automated perimeter conversion Pye, Herse, Nguyen, Vuong and Pham
of the Weber law. The difference in reported threshold sensitivity between the two perimeters is a result of the differences in scaling and stimulus and background luminances. For clinical purposes, it is possible to subtract 5 dB from the HFA sensitivity data to approximate that reported by the M600.
Figure 1. Relationship between sensitivity and retinal eccentricity for the Humphrey Field Analyser (open circles) and the Medmont M600 Perimeter (closed circles). The error bars represent ± 1 SEM.
Figure 2. Relationship between the Weber fraction and retinal eccentricity for the Humphrey Field Analyser and the Medmont M600 Automated Perimeter. Also plotted are neutral density filter results from Heuer and colleagues.4 The 0.0 ND filter data matched the control Humphrey Field Analyser data of this study and are left off for clarity. The error bars represent ± 1 SEM.
REFERENCES 1. Vingrys AJ, Helfrich KA. The Opticom M-600: a new LED automated perimeter. Clin Exp Optom 1990; 73: 3-17. 2. Anderson DR, Fener WJ, Alward WLM, Skuta GL. Threshold equivalence between perimeters. Ophthalmology 1989; 107: 493505. 3. Goldstick BJ, Weinreb RN. The effect of refractive error on automated global analysis program G-1. Amer J Ophthalmol 1987; 104: 229-232. 4. Heuer DK, Anderson DR, Feuer WJ, Gressel MG. The influence of decreased retinal illumination on automated perimetric threshold measurements. Amer J Ophthalmol 1989; 108: 643-650. 5. Wood JM, Wild JM, Bullimore MA, Gilmartin B. Factors affecting the normal perimetric profile derived by automated static perimetry. I. Pupil size. Ophthal Physiol Opt 1988; 8: 26-31. 6. Vingrys AJ, Demirel S. Temporal modulation thresholds isolate mechanisms with different adaptational and spatial properties. In: Vision Science and its Applications, Technical Digest 98-1. Washington DC: Optical Society of America. 1998: 78-81. 7. Fankhauser F. Problems related to the design of automated perimeters. Doc Ophthalmol 1979; 47: 89-138. 8. Heijl A. The Humphrey Field Analyser, construction and concepts. Doc Ophthalmol Proc Ser 1985; 42: 77-84. 9. Starita RJ, Fellman RL, Lynn JR. Static automated perimetry: background luminance and global visual field indices in the quantification of normal, suspect and glaucomatous visual fields. Invest Ophthalmol Vis Sci 1987; 28 (Suppl): 269. 10. Paige GD. Effect of increased background luminance on static threshold perimetry. Invest Ophthalmol Vis Sci 1985; 26 (Suppl): 226.
Author’s address: David Pye School of Optometry The University of New South Wales Sydney NSW 2052 AUSTRALIA
Clinical and Experimental Optometry 82.1 January–February 1999
13
82-1PYE.P65
13
1/07/99, 15:33
