Abstract
Purpose:
To show the reproducibility of 3D imaging and segmentation of hard exudates by a polarization sensitive OCT system (PS-OCT) equipped with an integrated retinal tracker
Methods:
A fiber based PS-OCT system combined with a retinal tracker was built. An assist function to enable follow-up measurements at exactly similar retinal positions was implemented. The PS-OCT system acquires images at 70 kHz A-scan rate up to 28° x 21° field of view. A 1024 x 256 scan 3D volume was recorded in 4.6 sec. The tracker operates at 60 Hz, and records fundus SLO video as well as conditions such as tracked fiducial marker position used in the baseline measurement. Based on that data, the system can reproduce the same imaging condition in follow-up measurements. Series of 3D scans were recorded repeatedly in patients with diabetic retinopathy. Two types of repetition conditions were used to simulate (a) short-term repetition and (b) long-term follow-up measurements. For condition (a), acquisitions were conducted while subject kept staying in the same position on a chair and on a chin-rest. For condition (b), subject was asked to change his or her position on the chair and the chin-rest, and realignment of OCT beam and eye was carried out in between the acquisitions. Hard exudates (HEs) were segmented from the 3D data based on their property to depolarize backscattered light. The reproducibility of the segmented HE area and volume was analyzed.
Results:
The mean area and volume of HE within a central macular field of 6° diameter circle for the measured eyes range from 0.50 to 0.92 mm2 and from 0.010 to 0.018 mm3, respectively. The mean reproducibility (standard deviation for 5 measurements) of the HE area (volume) was 6.3 % (13.6 %) for condition (a), and 11.8 % (21.4 %) for condition (b). When 2 low quality datasets out of 5 for each eye are excluded by visual inspection of pseudo-SLO images to avoid shadowing effect of media opacities in the field, it was 4.3 % (8.9 %) for condition (a), and 4.8 % (7.6 %) for condition (b).
Conclusions:
The retinal tracker improves the performance of PS-OCT in eyes with diabetic retinopathy. The reproducibility of ∼ 5 % for area and ∼ 10 % for volume can be used to discriminate between the system error and real change of lesions to support quantitative analyses in future follow-up studies.
Keywords: 499 diabetic retinopathy •
550 imaging/image analysis: clinical •
688 retina