Purchase this article with an account.
A. A. Khanifar, M. J. Rondeau, R. H. Silverman, H. O. Lloyd, R. V. P. Chan, D. J. Coleman; Characterization of Dry and Wet Age-Related Macular Degeneration Using High-Resolution Ultrasound Wavelet Analysis of the Choroid. Invest. Ophthalmol. Vis. Sci. 2009;50(13):307.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Choroidal alterations occur in both dry age-related macular degeneration (AMD) and wet AMD. Clinically, these choroidal changes are difficult to detect. Recent advances in optical coherence tomography (OCT) allow imaging of retinal structures in great detail.. However, due to the presence of pigmentation, optical imaging of the choroid has been challenging. High resolution ultrasound can image the choroid. While the choroidal microvasculature cannot be directly resolved ultrasonically, M-band Dual Tree Complex wavelets (M-DTCW) allows for modeling of unresolved scattering elements. Our primary objective is to determine if a significant difference exists in wavelet parameters amongst eyes without AMD compared to eyes with AMD.
Patients with AMD and without AMD in an academic retina practice were recruited for fundus photography, OCT, and high-resolution ultrasound. For each eye, M-DTCW analysis was performed. Wavelet coefficients for non-AMD (control), dry AMD, and wet AMD eyes were compared using ANOVA, and wavelet coefficient structures were modeled with Independent Components Analysis (ICA) and classified using a 10X 10-fold cross validation Support Vector Machine (SVM).
In the 62 eyes of 48 patients studied to date, 16 did not have AMD, 20 had dry AMD, and 26 had wet AMD. The ANOVA analysis on coefficients showed statistically significant differences. Using ICA to find the best independent features and a modern classifier, the SVM, we achieved excellent separation of the three classes with a receiver-operator curve-area under curve (ROC-AUC) value of 0.871.
Wavelet analysis allows characterization of dry and wet AMD and supplements clinical data. Wavelet parameters have significant classification power and may serve as biomarkers for assessing the severity of AMD. Longitudinal studies will be needed to evaluate whether these parameters can better predict risk of disease progression compared to current methodology.
This PDF is available to Subscribers Only