Abstract
Purpose :
To study the choroidal contour map in central serous chorioretinopathy (CSCR) and age related macular degeneration (AMD) and compare with healthy controls.
Methods :
Retrospective analysis on 13 healthy eyes, 12 eyes with CSCR and 8 eyes with AMD was performed. Volumetric scans were captured using a widefield swept source optical coherence tomography and CIS and COS in 3D were obtained using previously validated residual U-net (ResUnet) deep learning model for choroidal segmentation. Qualitative analysis of CIS and COS was performed using custom built MATLAB based 3D visualization tool and the choroidal contour shapes were determined in the order of decreasing steepness. Quantitative analysis was based on maximum principal curvature (Pmax) of the surface which was calculated at the center, superior, inferior, nasal and temporal macula.
Results :
Qualitative evaluation of CIS demonstrated the shape of a bell, hump, crescent, arc and cupid’s bow in 4,3,2,2 and 2 in normal eyes; 2,6,4,0,0 in CSCR eyes and 0,5,2,1,0 in AMD eyes respectively. Qualitative analysis of COS demonstrated the shape of a bell, hump, crescent, arc and cupid’s bow in 2,6,2,1,2 in normal eyes; 2,8,1,0,1 in CSCR eyes and 0,6,1,1,0 in AMD eyes. On quantitative analysis of CIS, Pmax at central macula was -0.23, 0.45 and -0.02 in normal eyes, CSCR and AMD respectively (p=0.015). Similarly, Pmax in superior, inferior and temporal macula was significantly different between the 3 groups (p=0.001,0.029,0.019).The difference between CIS and COS was the highest for AMD group (-6.73±2.2) as compared to normal eyes (-2.5±0.4) and CSCR (-2.8±1.7)(p=0.038). On making a comparison within the group, the central Pmax, superior Pmax, inferior Pmax and temporal Pmax between the CIS and COS of normal eyes (p=0.0002,0.00003,0.004,0.003), CSCR (p=0.006,0.001,0.0003,0.033) and AMD (p=0.009,0.005,0.006,0.005) were significantly different from each other.
Conclusions :
Choroidal contour was steeper in CSCR and flatter in AMD as compared to normal eyes. More data is continuously being added and will be presented at the meeting.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.