Three–dimensional fundus topographic reconstruction from stereoscopic photographs.

Abstract

Abstract: : Purpose: To compare fundus height measurements reconstructed from stereo photographs to those obtained with optical coherence tomography. Methods: Five stereoscopic film fluorescein angiograms from 5 eyes of 5 patients with choroidal neovascularization were retrospectively reviewed in patients who also had simultaneous assessment with optical coherence tomography (OCT–1). Image pairs with good subjective stereoscopic representation by visual inspection were chosen from each angiogram and digitized at 2400 dpi. Custom software was written to calculate retinal height information by determining the distance between two images at numerous points (average of 1086 points), subtracting a best–fit spherical approximation, and smoothing the final data with a Gaussian blur. Cross–sections of the resulting surfaces were compared to cross–sections from OCT data at 22 points along each radial scan line. Results: 660 calculated relative height values correlated with the same number of OCT retinal height points with a Pearson coefficient of 0.717 (p<.01) and with OCT retinal thickness measurements with a Pearson coefficient of 0.696 (p<.01). As expected, calculated fundus height values did not correlate as well with the RPE interface height (r=0.192, p<.01). Linear regression fitting of scalar–corrected OCT retinal height data with calculated fundus heights demonstrated a slope of 0.0276 pixels per micron. This suggests an average axial resolution of 36.2 microns per pixel. Conclusions: These data suggest that accurate retinal height data can be quantified from standard stereoscopic photographs. Higher resolution image data and advanced signal processing algorithms have the potential to improve the axial resolution. Reconstruction of fundus structures may be an effective way of detecting retinal and choroidal elevation – especially in retrospective studies or in other cases where OCT measurements cannot be obtained.

Keywords: imaging/image analysis: clinical • age–related macular degeneration