AUTOMATED TRITAN DISCRIMINATION SENSITIVITY - A NEW CLINICAL TECHNIQUE FOR THE EFFECTIVE SCREENING OF SEVERE DIABETIC-RETINOPATHY

We have developed and extensively assessed an automated chromatic disc rimination test which can be used to screen effectively a diabetic pop ulation for sight threatening diabetic eye disease. Equiluminant, sinu soidal, low spatial frequency, chromatic gratings are produced along a tritan confusion axis under computer software control on a high resol ution CRT. The correct position of the tritan axis in colour space was calibrated using subjects whilst they were transiently tritanopic. Th e minimum amplitude about a neutral ''grey point'' along the tritan co nfusion axis at which a subject can just distinguish a grating is foun d using a double staircase reversal algorithm. This measure is taken a s the Tritan Discrimination Sensitivity and it is this that we use to flag those diabetics with severe diabetic retinopathy. This computeris ed tritan discrimination test is quick, non-invasive, easy to operate, inexpensive and reliable. The test-retest reliability coefficient (rh o) is 0.92. The tritan discrimination test effectively identifies thos e diabetics who have or are most at risk of developing severe diabetic retinopathy. The sensitivity of the test for the detection of maculop athy, ischaemic retinopathy (pre-proliferative), and proliferative ret inopathy is 97%, 65%, and 93%, respectively. The corrected specificity of the test is 83%. We conclude that the tritan discrimination test h as potential for use as a valuable screening tool for the early detect ion and treatment of severe diabetic retinopathy.