Using polarization-sensitive OCT to detect and quantify fibrotic lesions in the human retina

Abstract

Purpose : To facilitate the diagnosis and quantification of fibrotic tissue in the human retina by analyzing polarization-sensitive (PS) OCT data using an automated process.

Methods : 59 eyes of 59 patients (77±6 years) with neovascular age-related macular degeneration (nAMD) were imaged using a custom-built spectral domain PS-OCT system operating at 860nm with an A-scan rate of 70kHz and an integrated retinal tracker. Raster scans consisting of 250 B-scans × 1024 A-scans were recorded at the macula, covering an area of 8×6 mm². After standard SD-OCT data processing, PS data (retardation, optic axis orientation, degree of polarization uniformity) were extracted. The PS data were compensated for the birefringence of cornea and Henle’s fiber layer. The double-compensated axis orientation was projected onto an enface map, fibrosis was separated from healthy tissue based on axis uniformity and using a region growing algorithm, and finally the area of the segmented fibrosis was obtained. A sub-set of 16 patients were measured three times to assess the repeatability of the method.
In addition, color fundus photography (CFP) was performed on all eyes.

Results : Fig. 1(a–c) shows an example of measurements in a patient with fibrosis. Patches of well-defined axis orientation (Fig. 1a) are an indicator of fibrosis and are segmented (blue areas in Fig. 1b). The CFP in Fig. 1c shows a yellow-whitish discoloration at the same location, indicating the presence of fibrosis. In 31 of the 59 eyes, fibrosis was diagnosed on CFP. Of the 28 eyes diagnosed as non-fibrotic based on CFP, our algorithm confirmed the absence of fibrosis in 23 cases (82%). Manual inspection of the data of the remaining 5 cases suggests the occurrence of false positives in the PS method. Of the 31 eyes diagnosed with fibrosis based on CFP, our algorithm confirmed fibrosis in 21 cases (68%). After manual inspection of the disagreeing cases, 9 out of 10 cases were assessed to likely be non-fibrotic and the remaining case might have a weak fibrotic lesion.
Fig. 1d shows the results of the repeatability measurements (mean and SD (error bars)). The mean repeatability of the fibrotic lesions > 0.5mm² was 17%. Patients who have been diagnosed as fibrotic based on CFP are marked with an orange background.

Conclusions : PS-OCT can be used to detect and quantify fibrotic lesions in nAMD with good repeatability. Some discrepancies to CFP require further analysis.

This is a 2021 ARVO Annual Meeting abstract.