Purchase this article with an account.
Wolfgang Wieser, Thomas Klein, Aljoscha Neubauer, Lukas Reznicek, Anselm Kampik, Robert Huber; Feasability of ultrawide-field retinal-shape measurement with MHz-OCT. Invest. Ophthalmol. Vis. Sci. 2013;54(15):1469.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To investigate the feasibility of high-resolution retinal shape measurement with ultrawide-field MHz-OCT and to assess the dominant source causing deviation from the real shape.
A swept-source OCT system based on a 1060nm Fourier-domain mode-locked (FDML) laser was constructed and wide-field retinal imaging with 60° field of view was performed. OCT datasets consisting of 1900x1900 axial (~3.6million) axial scans were acquired in 2.5s. 10 eyes from healthy volunteers were imaged. An analysis of the optical path showed that OCT wide-field datasets are intrinsically distorted due to the scanning geometry, the imaging optics, the anatomy of the anterior segment of the eye to be imaged, and additionally due the exact position of the eye with respect to the conjugate of the beam scanning system. Hence, OCT datasets do not reflect the real shape of the eye. For OCT images of the human eye, the largest deviation from the true shape originates from the scanning geometry itself, since even an aberration-free imaging system would yield a flat rather than a spherical shape. Thus, we implemented a distortion correction algorithm that corrected for scan geometry distortion only.
Figure 1 shows both an uncorrected and corrected 3D view of an acquired dataset. While the uncorrected dataset is flat due to the scanning geometry, the corrected dataset shows a more realistic spherical retinal shape.
With MHz-OCT, densely-sampled wide-field datasets of the human retina can be acquired that are suitable for shape analysis. However, OCT images need to be corrected for several distortions that affect the position of each axial scan. Correcting for scan geometry only already yield results close to a spherical shape, as compared to the anatomically incorrect flat shape of unprocessed datasets.
This PDF is available to Subscribers Only