Abstract
Purpose :
To demonstrate quantitative distance and area measurements on the retina using widefield fundus images, and to analyze the accuracy and reliability of these measurements.
Methods :
Widefield fundus images (true-color and fundus autofluorescence) were acquired using a prototype widefield slit-scanning ophthalmoscope (ZEISS, Dublin, CA). In widefield imaging, inherent optical distortion means that 2D fundus images cannot reproducibly encode the shapes of 3D retinal features, given that patient fixation orientation naturally varies. Without making any assumptions about the shape of the retina, 2D image pixels were mapped to 3D co-ordinates defining the corresponding rays of light emerging from the pupil. The angle between rays from retinal features, multiplied by either individualized or conventional optical magnification of the eye, gives a reproducible measure of distances. Retinal regions and features were manually traced on the images by a licensed optometrist. Their measured sizes were validated using comparative assessments obtained via optical coherence tomography (CIRRUS™ HD-OCT 5000, ZEISS, Dublin, CA). Reliability was assessed over repeat captures across different modalities and patient fixation orientations.
Results :
Widefield distance and area measurements on retinal images are presented for healthy subjects. Figure 1 illustrates measurement of optic disc area, using five same-eye captures acquired with different fixation orientations. The apparent shape of the disc is visibly distorted per fixation. The widefield area measurement (mean±SD) was 1.94±0.07 mm2. A comparable optic disc area of 1.87±0.10 mm2 was obtained using optical coherence tomography. The coefficient of variation (CV) for the widefield disc area measurements was 3.6%.
Conclusions :
Measurements on the retina can be reliably determined from widefield fundus images, despite the presence of optical distortion, by employing a suitable mapping from 2D pixels to 3D co-ordinates based on the known imaging geometry.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.