A Hyperspectral Cube Is Worth More Than a Thousand Scans: A New One Exposure, Two Dimensional Spatial, One Dimensional Spectral Fundus Imager

Abstract

Purpose:: To demonstrate the feasibility of acquiring complete hyperspectral cubes of data in single exposures from the human retina, in vivo, using a novel fundus imager that marries Fourier Transform Spectroscopy with Integral Field Spectroscopy.

Methods:: A Zeiss FF430 fundus imager was fitted with Kestrel’s hyperspectral Fourier Transform Spectrometer and an integral field unit. The spectrometer has been demonstrated by the investigators before. The integral field unit captures the light returning from the retina onto a 9x9 rectangular array of silica optical fibers which is remapped into an 81 fiber-long linear array at the entrance slit of the spectrometer. Three normal subjects, ages 32, 34, and 53 with no known retinal disease were imaged with the new fundus camera. Hyperspectral cubes were acquired from regions of interest containing spatial features such as vessel crossings and the optic disc rim. Proprietary software was used to process the interferographic data into spectra and to remap the spatial information into 2 dimensional images.

Results:: In the spectrum range 500-800nm each hyperspectral cube has dimensions of 9x9x82 pixels, i.e. 9x9 spatial pixels at 82 spectrally resolved wavelengths. Each cube is acquired in one single exposure. Data from the normal subjects show spatial patterns consistent with normal color images. Spectral signatures of known features are consistent with those taken with other spectral imagers.

Conclusions:: The feasibility of using integral field spectroscopy to collect complete cubes of hyperspectral data in vivo from the human retina has been demonstrated. The data collected provides positive evidence of the value added to the incipient area of spectral fundus imaging by providing a more informative spatial image of the fundus. Current efforts are focused on testing clinical significance by collecting simultaneous morphological and spectral data of retinal diseases such as diabetic retinopathy and macular degeneration.

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