Deep Learning-Based Choroidal Volumetric Analysis After Photodynamic Therapy or Focal Laser For Central Serous Chorioretinopathy

Abstract

Purpose : To utilize deep learning-based volumetric analysis to quantify the volumetric changes of choroidal vessels after photodynamic therapy (PDT) and focal laser photocoagulation (PC) for central serous chorioretinopathy (CSCR).

Methods : The three-dimensional (3D) volume of choroidal vessels and stroma were analyzed using a deep learning-based quantitative method at baseline and three months after treatment. The change of each volume in areas of the 6-mm diameter circle of the early treatment diabetic retinopathy study grid was compared between the two treatments.
Main Outcome Measures: The change in the volume of choroidal vessels and stroma three months after each treatment compared with baseline.

Results : The mean age was 49.3 ± 10.5 years. In the central 3-mm circle of the ETDRS grid, the vessel volume significantly reduced in both the PDT group (88% p <0.001) and the PC group (94% p=0.03), and the higher decrease was observed in the PDT group (p=0.0415). In the 6-mm diameter whole circle and the treated area, the vessel volume showed a significant reduction in both groups equally at three months (6-mm diameter circle; 93% p=0.0005 and 95% p=0.0020, treated area; 90% p=0.0008 and 93% p=0.0033 in PDT and PC group respectively). The stroma volume showed a decreasing trend in the treated area (94% p=0.091 and 96% p=0.097 in PDT and PC groups, respectively) but only in the only 6-mm diameter circle was there a significant decrease (95% p=0.0219 and 95% p=0.0062 in PDT and PC group respectively), similar to both groups.

Conclusions : The 3D deep-learning analysis of OCT revealed that not only PDT but also focal PC reduced the vessel volume of the choroid in eyes with CSCR. The algorithm may be useful for future pathological elucidation.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.