Abstract
Abstract: :
Purpose: To assess the effectiveness of different abnormality measures for use with a rapid new method for assessing glaucomatous damage using the pupillary light reflex (PLR). Methods: We previously described a method for using the PLR to compare the sensitivity of spatially–defined areas of retina in one eye (Chen et al, 2004; Wyatt et al, 2004). Subjects look at a monitor which displays stimuli covering large areas of the central visual field. The stimuli are shaped to correspond to typical patterns of glaucomatous damage. Stimuli symmetric about the horizontal meridian are displayed alternately, for 1 sec each, and relative sensitivity of the PLR in the two areas is determined. Three such stimulus pairs cover a large portion of the central visual field. Previously, measures for damage were developed in terms of asymmetry between corresponding superior/inferior areas. In the present work we developed amplitude measures for use in conjunction with asymmetry measures, and used receiver operating characteristic (ROC) analysis to assess performance. Results: Amplitude measures were as efficient as asymmetry measures for detecting functional abnormalities. Information from amplitude and asymmetry measures was not identical; abnormalities could appear in either type of measurement alone, or in both. (E.g., a diffuse loss might appear as an amplitude abnormality without appearing as an asymmetry.) With this approach, as many as 12 measures of abnormality can be determined for each eye, using the same group of three tests previously described, and without changing test protocol or duration. Logical conjunctions of asymmetry and amplitude measures were found to augment test sensitivity (ability to detect abnormality in an eye) while retaining substantial specificity. (False–positive rates remained relatively low.) Measures based on asymmetry gave an area under the ROC curve of 0.83; measures based on amplitude gave ROC areas of 0.67 to 0.77. Using conjunctions of asymmetry and amplitude measures, areas as high as 0.90 were obtained. Conclusions: Both amplitude measures and asymmetry measures can be used to identify PLR abnormalities in patients with glaucoma. As with perimetry, conjunctions of different measures can improve sensitivity and specificity.
Keywords: pupillary reflex • visual fields • perimetry