False negative results in glaucoma detection with Heidelberg Retina Tomograph II
A total of 100 eyes of 50 patients with ocular hypertension and glaucoma of different etiology were examined with the Humphrey-Field-Anazlyer (program 30-2) and the noise-field campimeter according to Aulhorn and Köst [1,2]. The sensitivity of noise-field campimetry vs light-sense perimetry, i.e. the percentage of eyes with an abnormal result in light-sense perimetry that were also identified as abnormal by noise-field campimetry, was 86%. The specificity, i.e. the percentage of eyes with ocular hypertension and thus by definition a normal visual field in light-sense perimetry, that were also identified as normal by noise-field campimetry, was 31%. In 12% of eyes light-sense perimetry indicated defects while the result of noise-field campimetry was normal, and in 11% of eyes noise-field campimetry showed defects though there was a normal field according to light-sense perimetry. The subjective comparison of the two methods with regard to extent and location of the field defects showed good agreement in 22% of eyes for fine noise and in 25% for coarse noise. Moderate agreement was found in 32% fine noise and 34% for coarse noise, and poor agreement, in 46% of eyes for fine noise and 41% for coarse noise. Cooperation was good in 84% of patients and was not dependent on age. Deep localized nerve fiber bundle defects are detected reliably with the noise-field campimeter. The detection of small to moderate relative defects or of generalized changes in terms of a diffuse loss, however, is unreliable. The depth of the field defects is not related to the perception of the noise field.