Abstract
Purpose: :
To analyse the diagnostic value of HD-OCT for glaucoma diagnosis using Discrete Fourier Transform analysis of the TSNIT plot and to compare it with the diagnosis based on average thickness values of the TSNIT plot.
Methods: :
A sample of 55 healthy volunteers and 42 age matched glaucoma patients (MD -7.5 ±6.4) underwent complete ophthalmologic examination, including RNFL assessment with Cirrus® HD-OCT (Carl Zeiss Meditec Inc). The TSNIT plots of the subjects were exported and compacted to 64 segments. The 64 segments of the TSNIT plots were analysed by was Discrete Fourier Transform, computed with fast Fourier Transform (FFT) analysis. The Fourier coefficients of patients and healthy volunteers were compared and the diagnostic separation calculated using the area under the receiver operating characteristic (AROC). Additionally we compared the 2 groups using the standard RNFL thickness parameters as provided by the printouts: the average thickness of the circumference and those of the 4 quadrants.
Results: :
There was a significant difference between the groups for all standard RNFL parameters and also a high AROC (average 0.962, temporal 0.837, superior 0.973, nasal 0.778, inferior 0.945) indicating good diagnostic separation. The AROC of Fourier coefficients resulted in highly significant differences between groups for the coefficients C1 (0.967), C3 (0.986), C4 (0.858), C6 (0.803), C60 (0.803), C62 (0.858) and C63 (0.986). There was no statistically significant difference between the AROCs of average, superior and inferior RNFL and those of the Fourier coefficients C1, C3, C4, and C63.
Conclusions: :
We did not find a significant difference between the diagnostic separation of the average parameters and of the Fourier coefficients.
Keywords: imaging/image analysis: clinical • nerve fiber layer