Abstract
Purpose :
Macular Thickness Analysis (MTA) is an important tool for diagnosing and monitoring patients with retinal disease. The robustness of MTA can be limited in low-cost devices due to the quality of the optical coherence tomography (OCT) data. In this study we evaluate the agreement of MTA between repeat scans of the same eye using a low-cost OCT prototype.
Methods :
A low-cost OCT prototype system (ZEISS, Dublin, CA) was used to image 42 subjects with a range of ocular pathologies, including age-related macular degeneration (AMD). One eye of each subject was scanned three times. The inner limiting membrane (ILM) and the retinal pigment epithelium (RPE) of each OCT volume were segmented using a prototype segmentation algorithm. Macular thickness (MT) map with 512x512 pixels over an area of 5.78x5.78 mm was created by measuring the difference between the ILM and RPE. Two MT maps (out of the three) were selected based on the segmentation quality determined with a prototype segmentation quality algorithm for this study.
The two MT maps were registered to each other and the ETDRS grid with 9 sectors was placed at the center of each map. The ETDRS grid consists of three concentric circles with radiuses of 0.5, 1.5 and 2.89 mm (Fig 1). Each ETDRS sector value was calculated by averaging the MT values within the sector. Linear regression and Bland-Altman analysis were used to compare the MT maps.
Results :
Fig 2 shows the statistical comparison between the ETDRS sectors generated using two MT maps per subject. All sectors show an absolute mean difference less than 2 microns. The 95% lower and higher limits of agreement varied between -18 and 19 microns. R-squared values varied between 0.85 and 0.99.
Conclusions :
This study demonstrated the agreement between two MT maps generated from two scans per eye using the low-cost OCT prototype. Our results show good correlation and agreement between two MT maps which is important for diagnosing and monitoring patients using macular thickness analysis.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.