Abstract
Purpose :
To generate tomographic retinal cross-sections using en-face images acquired from non-interferometric Adaptive Optics (AO) Ophthalmoscopes, including Flood-Illumination (FIO) and Scanning Laser (SLO). We call this method “Optical Incoherence Tomography”(OIT). We explore the focus sensitivity of the OIT to guide focus positioning, improving the imaging protocol in en-face ophthalmoscopes.
Methods :
The OIT method consists of three steps: acquiring en-face images from different focal planes (forming a Z-stack), filtering out the image low-spatial frequency content, and computing the image energy. This method was applied in the PARIS AO-FIO and in the PSI (Physical Sciences, Inc.) AO-SLO, on bright-field, dark-field and motion contrast modalities.
Results :
OIT tomographic retinal cross-sections were obtained for each imaging modality, even those without apparent optical sectioning capability. Results were compared to clinical OCT and showed that OIT is able to reliably identify most of the retinal layers. Moreover, we show the potential of OIT to guide focus positioning enabling capturing en-face images at desired layers, improving imaging protocol. Here, with the OIT, we can establish that the vascular network is organized in 4 layers at 7deg nasal in a healthy subject, and finely position the imaging focus plane at each layer to capture en-face images. Thanks to this simple imaging protocol, it is possible to generate depth color maps where capillaries can be followed and interconnecting vessels can be identified and the 3D vascular organization can be studied.
Conclusions :
We devised a non-interferometric technique to obtain tomographic retinal cross-sections, based on the high-spatial frequency content of en-face retinal images from different focal planes, which we call “Optical Incoherence Tomography” (OIT). This technique can be employed in different imaging techniques (conventional or scanning) for various imaging modalities (bright-field, confocal, non-confocal, dark-field, motion contrast), without extra hardware, in order to view cross-sectional information in non-multimodal systems, and guide axial positioning, improving imaging protocol.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.