Glaucoma is one of the leading causes of irreversible blindness in the world. Deterioration of the retinal nerve-fiber layer (RNFL) is a pathology associated with the progression of glaucoma. The technique of optical coherence tomography (OCT) is a leading method for non-invasive imaging of the RNFL. It has been demonstrated that summary measures of RNFL thickness derived from OCT significantly correlate with visual deterioration during glaucoma. However, the method used to summarize RNFL thickness, the fast Fourier transform, is not particularly well suited to describing abrupt changes in thickness, as would be found in an OCT scan of an RNFL undergoing heterogeneous deterioration during the progression of glaucoma. A better method for summarizing abruptly varying complex signals is wavelet decomposition. Wavelet decomposition breaks a waveform into two major components describing trend and fluctuation. Because the fluctuation components are designed to capture localized, short-term signal variation it is expected that RNFL deterioration due to glaucoma will appear as significant deviations in these components. Wavelet decomposition will be applied to OCT scans from both normal individuals and patients diagnosed with primary open-angle glaucoma (POAG). OCT scans consisting of 256 data points (representing RNFL thickness in an annulus around the optic nerve head) will be decomposed into trend and fluctuation components over a range of scales of resolution (multiresolution analysis). The hypothesis that OCT fluctuation components vary deterministically due to disease, will be tested by evaluating the utility of subsets of these components for linear discriminant classification of normal versus POAG eyes. Change in wavelet fluctuation components as a function of glaucoma progression will be assessed by the statistical analysis of regression coefficients of fluctuation components against visual function in normal and diseased eyes. ? ? ?
