advertisement
Editors Selection IGR 7-1
Clinical Examination Methods: New developments: Comparing VF methods by S/N ratio
Comment by Douglas Anderson on:
24828 Signal/noise analysis to compare tests for measuring visual field loss and its progression, Artes PH; Chauhan BC, Investigative Ophthalmology and Visual Science, 2009; 50: 4700-4708
In plain English, 'signal' here means the difference between two visual threshold measurements in perimetry. In clinical settings, it means a change over time, but Artes and Chauhan (1470) very cleverly use measurements in two locations (perhaps a normal area compared to an abnormal area) as a surrogate ‐ difference of thresholds at two locations rather than two times.
Repeat perimetry gives a somewhat different threshold value at any given location, and this is 'noise'. When trying to detect a small difference ('signal'), then test-retest variability ('noise') will interfere. The thesis of this article is that the ability of a test to recognize progression can be evaluated by the signal/noise ratio (S/N) ‐ the difference between two threshold measurements compared to the test-retest variability of measurements.
The concept is illustrated by comparing standard automated perimetry (Humphrey SITA) and the 24-2-like pattern of frequency-doubling perimetry (Matrix, or FDT2). Each test was repeated six times to determine variability of threshold measurements (noise), and this was compared to the average difference between selected locations above and below the horizontal (signal).
These perimetric tests are well suited to show this principle, as the Humphrey perimeter defines contrast between a projected white spot and the background, while the Matrix uses contrast between flickering black and white stripes. Especially with different definitions of a decibel, the detectable threshold difference cannot be compared, but when the difference is divided by the test-retest variability, the relative ability to detect equivalent differences, measured on two different scales, is revealed.
Globally, the Matrix seems to compare favorably with the Humphrey Visual Field Analyzer. However, the main point here is the concept, and refinements are needed before reaching conclusions about particular instruments: The range of disease severity covered by the two instruments is not the same. Variability ('noise') is not the same at all threshold levels (noise may differ in the 5 dB change from 32 to 27 dB and a change from 24 to 19 dB). The S/N could be greater for one instrument at the higher threshold levels, but for the other instrument at the lower levels. This new concept needs to be applied to structural measurements in comparison to perimetry. The considerable contribution is to introduce the use of S/N to compare methods that are not equivalent or even similar. As new methods of quantifying the stage of glaucoma come along, the concept will help select the most effective tests to use while monitoring a person with glaucoma.
Comments
The comment section on the IGR website is restricted to WGA#One members only. Please log-in through your WGA#One account to continue.Log-in through WGA#One
