Abstract
Purpose: :
The aim of the study was the classification of glaucoma patients and severe glaucoma patients. Therefore cortical response signals after selective color channel stimulation were used. The classification was performed with the help of several classification methods.
Methods: :
Visual evoked potentials were recorded for 29 glaucoma patients (17 f, 12 m, 61.8 ± 10.9 years), 6 severe glaucoma patients (4 f, 2 m, 65.0 ± 14.7 years) and 31 age-matched healthy subjects (23 f, 8 m, 59.8 ± 9.5 years). On the basis of silent substitution stimulation (SST), separate responses were obtained for short- (S) wavelength-sensitive cones and medium- and long- (LM) wavelength-sensitive cones. The flash stimulation (±11°) was presented using a 30" liquid crystal display. The cortical signals were recorded (Oz vs. Fz, @ 512 Hz) with two scalp electrodes, pre-processed and parameterized. Linear discriminant analysis, and support vector machine were used for classification.
Results: :
Good results were reached for the classification of severe glaucoma patients. For the S-cone response the classification rate of the several classification methods was between 99.2 % and 99.6 % (sensitivity: 98.4 - 99.3 %, specificity: 100.0 %). The classification, based on LM-cone responses, showed lower values (classification rate: 93.5 - 98.1 %, sensitivity: 88.1 - 96.3 %, specificity: 98.8 - 99.9 %). At the classification of glaucoma patients the classification rates of the several methods was in the range of 76.7 - 76.9 % for S-cone response (sensitivity: 80.5 - 81.5 %, specificity: 71.0 - 72.5 %) and in the range of 74.4 - 76.2 % for LM-cone response (sensitivity: 78.9 - 84.3 %, specificity: 66.6 - 69.1 %).
Conclusions: :
We used the SST for selective cone stimulation. The pre-processed cortical response signals were classified. The classification methods showed very good classification results for the classification of severe glaucoma patients and good results for the classification of glaucoma patients. The several classification methods showed similar results. Better results were reached by the use of S-cone responses than by the use LM-cone responses. Glaucoma can be assessed by objective electrophysiological testing.
Keywords: electrophysiology: non-clinical • visual cortex • photoreceptors