Evaluation of Stromal Riboflavin Penetration Using Slit-Lamp-Based Digital Image Processing in an Ex Vivo...

To evaluate the effect of different iontophoresis current intensities on stromal riboflavin penetration for transepithelial corneal cross-linking (CXL) and to develop a simple, accessible, and quantitative image-processing workflow for its objective assessment using digital slit-lamp biomicroscopy.

Ex vivo bovine corneas (n = 3 per group) were allocated to five groups: untreated control, topical riboflavin, and iontophoresis at 1, 3, or 5 mA for 5 min. After treatment, cobalt-blue-excited slit-lamp images were obtained. Stromal fluorescence was quantified using a Python-OpenCV workflow that converted images to hue-saturation-value space and applied segmentation-based masking to generate a fluorescence score (FS). For validation, stromal extracts were analyzed spectrophotometrically at 370 nm to determine riboflavin concentration. Finally, the association between FS and spectrophotometric riboflavin levels was examined to evaluate the utility of slit-lamp fluorescence as a noninvasive surrogate measure.

Mean FS increased in a current-dependent manner. Iontophoresis enhanced stromal loading, producing mean FS values of 1.04 ± 0.17 (1 mA/5 min), 1.82 ± 0.13 (3 mA/5 min), and 2.53 ± 0.25 (5 mA/5 min), with higher currents also promoting a more homogeneous stromal distribution. Based on the linear FS-concentration relationship, these values corresponded to increasing stromal riboflavin concentrations. A strong linear correlation (R² = 0.96) between FS and spectrophotometrically calculated riboflavin concentration demonstrated that slit-lamp-derived fluorescence accurately predicts stromal riboflavin content.

The proposed digital slit-lamp analysis provides a simple, objective, and reproducible method for quantifying stromal riboflavin fluorescence. The strong correlation between FS and spectrophotometric measurements supports its use as a noninvasive surrogate for stromal riboflavin assessment. With further in vivo validation, this approach may enable real-time monitoring and optimization of transepithelial CXL protocols.