Cross-software comparison shows strong agreement for quantitative indocyanine green fluorescence angiography in...

Quantitative indocyanine green fluorescence angiography (Q-ICG-FA) has emerged as a promising tool for objective intraoperative perfusion assessment, yet there remains uncertainty regarding its reproducibility across different software platforms. This study aimed to evaluate the agreement of key Q-ICG-FA parameters across two independent software platforms using identical ICG-FA recordings.

Eighty ICG-FA recordings from reconstructive procedures were analyzed using two software platforms: AMS and EPA. Both programs generated fluorescence-time curves (FTCs) and calculated seven perfusion parameters. The primary outcome was time-to-peak (TTP). Secondary outcomes included T₀, F_max, absolute mean slope inflow, normalized mean slope inflow, normalized maximum slope inflow, and normalized maximum slope outflow. Agreement between platforms was evaluated using intraclass correlation coefficients (ICC), non-parametric testing, and Bland-Altman analysis.

Excellent agreement was observed for TTP (ICC = 0.979, 95% CI: 0.967-0.987) and normalized mean slope inflow (ICC = 0.944, 95% CI: 0.913-0.964). Good to excellent agreement was found for T₀, F_max, and absolute mean slope inflow. In contrast, parameters based on normalized maximum slopes showed poor to moderate agreement, with ICC values of 0.412 for inflow and 0.315 for outflow. The Wilcoxon signed-rank test revealed significant systematic differences for six out of seven parameters, with AMS showing overall higher TTP values compared to EPA (p < 0.001). No significant difference was found for normalized mean slope inflow (p = 0.158). Bland-Altman analysis showed that while TTP values had a mean difference of +4.2 s, the limits of agreement were wide, ranging from -42.2 to +50.5 s. For the secondary outcomes, normalized mean slope inflow demonstrated the least variability and the best agreement across platforms.

These results suggest that TTP and normalized mean slope inflow may be reliable candidates for defining quantitative perfusion thresholds, although further clinical validation is needed.