PURPOSE: To evaluate the predictive value of indocyanine green angiography (ICGA)-based gray-level co-occurrence matrix (GLCM) texture analysis for anti-vascular endothelial growth factor (anti-VEGF) treatment response in central serous chorioretinopathy (CSC). METHODS: This retrospective study included 135 CSC patients. Pretreatment ICGA images were obtained at four timepoints (30 seconds, 3 minutes, 10 minutes, and 15 minutes post-dye injection), and GLCM texture analysis was performed to quantify contrast, inverse difference moment (IDM), and entropy within a 6-mm macular region. ICGA staining heterogeneity was assessed across timepoints. Choriocapillaris flow deficit (CCFD), small vessel flow signal (SVFS), and Sattler's layer flow were evaluated by optical coherence tomography angiography and correlated with texture features. RESULTS: Among 135 eyes, 68 were good responders and 67 were poor responders. In the 10- and 15-minute phases, poor responders showed significantly higher contrast and entropy and lower IDM than good responders (all P < 0.005). Mid-late phase entropy (15 min) showed the highest predictive value (AUC = 0.72) and correlated with CCFD + SVFS at 31 μm beneath Bruch's membrane (P = 0.04). CONCLUSIONS: ICGA texture heterogeneity in the mid-late phase, particularly increased entropy, is a potential biomarker for poor anti-VEGF response in CSC. These findings suggest that inner choroidal permeability and structural changes are key prognostic indicators.
Hwang et al. (Wed,) studied this question.