Abstract
Purpose::
To probe the characteristics of sub-clinical retinal changes in type 2 diabetics using scanning laser polarimetry. To compare the contrast of these features using images with differing polarization content.
Methods::
We used a GDx scanning laser polarimeter (GDx, LDT/CZM) to acquire 15 degree macular images (256 x 256 pixels) in 12 diabetics (mean age 62.9, std dev 10.8) and 12 age- and gender- matched normals. All subjects received a comprehensive eye examination, including dilation, within 1 year of imaging. No retinopathy was present at examination in any of the diabetic subjects. The GDx produces image series with 40 raw images: 20 input polarizations and 2 detectors, one parallel to the input polarization and one perpendicular (crossed). We computed 18 images with differing polarization content. We used two of the image types to calculate contrast measurements in the same location containing retinal features of interest: the depolarized (scattered) light image from the crossed detector and the maximum amount of light returning to the parallel detector (maximum of the parallel), which emphasizes superficial retinal features. A 3 x 3 pixel area was used to compute mean grayscale of each feature of interest and an adjacent area. Michelson contrast measurements between these two areas were calculated for each image and compared using a paired-t test.
Results::
Widespread, focal, bright features (generally 5 pixels, 85 microns, or less) could be seen in the depolarized images in all 12 diabetics, but were not seen in age-matched normals. 10 of these patches were analyzed in one eye of each diabetic subject. A paired t-test showed that the contrast of these features relative to the neighboring points was significantly higher in the depolarized image when compared to the maximum parallel detector image (p<0.001). These features were not detected during clinical fundus examination.
Conclusions::
Scanning laser polarimetry can be used to detect unique retinal features in diabetics, even in the absence of clinical retinopathy findings. Visualization of these features can be enhanced using depolarized light imaging, which typically enhances the visibility of deeper retinal features. Given that the contrast was significantly greater in the image that emphasizes the deeper features compared with the image that emphasizes more superficial features, we conclude that many of these features are likely to be in the deeper retinal layers.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical • diabetes