DETECTION AND MEASUREMENT OF SMALL OCULAR MISALIGNMENTS WITH PURKINJEREFLECTION PATTERN EVALUATION

Background Application test for an automatic classification strategy f or ocular alignment data for the detection of ocular misalignment in s trabismic patients. Methods Photographic Purkinje Reflection Pattern E valuation was used a) with a handheld device for the detection and mea surement of ocular misalignments in near fixation (group 1, n = 64 str abismic patients) and b) with a stationary device for the detection an d measurement of ocular misalignments in near fixation (group 2, n = 3 8 patients) and in distance fixation (group 3, n = 36 patients). The o rthoptic diagnoses were mostly primary and secondary microtropia with manifest angles of strabismus from naught or 0.25 degrees to 3-4 degre es, with maximum angles up to 6-9 degrees. The ocular alignment data w ere classified using the computer based strabismus index procedure. Th is strategy relies on thresholds derived from means and standard devia tions in orthotropic control populations. In this way the data sets we re classified automatically as ''no referral'' or ''referral''. In add ition, an automatic diagnosis of the type of:misalignment was given an d the results were compared to the orthoptic gold standard. Results Th e sensitivity for the detection of a manifest ocular misalignment was ca. 80% in group 1 and 2, and 90% in group 3, with specificities from 90% to 100%. All manifest angles of strabismus larger than 1 degrees w ere correctly classified as ''referral''. There was good agreement bet ween the diagnoses of the type of misalignment in most cases. Discrepa ncies were observed with very small ocular misalignments or with incom plete data sets, or they could be explained by a switch of fixation. T he amount of the misalignment varied markedly as compared to the ortho ptic measurement in a number of cases. Conclusions The examination all ows for a detection of small manifest ocular misalignments with a very high sensitivity. The deviated eye and the type of the misalignment i n the primary position are evaluated automatically by a data base comp uter algorithm. The differences between the measured angles of strabis mus indicate that the photographic examination conditions and the orth optic simultaneous prism and cover test conditions are not exactly ali ke. Purkinje Reflection Pattern Evaluation represents a step towards a n examiner-independent measurement of the angle of strabismus.