A MODIFIED SCORING TECHNIQUE FOR COLOR VISION PANEL TESTS

Title A MODIFIED SCORING TECHNIQUE FOR COLOR VISION PANEL TESTS
Author, Co-Author P. King-Smith, Algis Vingrys
Topic
Year
1992
Day
Sunday
Program Number
2:30 pm
Room
Ireland B
Affiliation
Abstract We have previously described (IOVS, 1988) a method for scoring color vision panel tests such as the D-15 and 100-Hue tests. The analysis yields three parameters; the first is the confusion angle which identifies the type of color defect; the second is the Confusion index (C-index) which quantifies the degree of color loss; the third is the Selectivity index which quantifies the polarity of the cap arrangement. The method is based on calculating 'color difference vectors' between all pairs of neighboring caps and using these to calculate a moment of inertia (assuming unit mass at the tip of each vector). The confusion angle was shown to yield good discrimination between protan and deutan types of congenital color defect, provided that the defect was relatively severe (large C-index); however, if the C-index was small, confusion angles were not only more scattered, but were also biassed towards the arbitrary confusion angle for a perfect arrangement, so that deutans tended to be classified as protans. To avoid this bias, we propose analyzing only 'incorrect' color difference vectors (e.g. if cap 4 is next to cap 6 but not if it is next to cap 3 or 5). In a retrospective analysis of monocular D-15 tests from 63 congenital color defective eyes (17 protan and 46 deutan), this new method was shown to be always superior to the original method at discriminating protans from deutans; for example, excluding the 6 eyes with perfect arrangements, and the 4 eyes with the lowest C-indices, the area under the ROC curve for protan/deutan discrimination was 0.980 for the new method, compared to 0.961 for the original method.
Affiliation of Co-Authors
Outline