Purpose:
To describe and evaluate a technique for aligning stereoscopic image pairs in fluorescein angiogram sequences for better visualisation and interpretation of lesion components in FAs.
Methods:
Five stereo image pairs at about 30’’, 1’, 2’, 5’ and 10’ were selected from each FA sequences from patients with advance age-related macular degeneration (AMD) acquired on a digital confocal scanning laser ophthalmoscope (HRA2, Heidelberg Engineering, Heidelberg, Germany). Five images were aligned by using an in-house program based on the commercial program i2k Retina (DualALign, US) whereby the first image (30’’) was used as the reference. Two observers (DP, GM) evaluated independently the stereoscopic effect of all the pre- and post-processing images into four categories (Good, Fair, Poor and Cannot Grade). Statistical analysis was performed using statistical software SPSS 18 (SPSS Inc., Chicago, IL, USA).
Results:
Images from 50 FA sequences were collected and analysed. Fig. illustrates an example of the effect before and after processing where the green and red lines highlight the corresponding locations (Top row: original images; Bottom row: processed images). For both observers, the Mann-Whitney test demonstrates that there is no statistical significant difference in stereoscopic effect between images before and after processing (p=0.723 and p=0.136). The agreements are moderate before and after processing (Κ=0.56 and 0.42, respectively).
Conclusions:
The proposed technique can align the stereo images pairs in an FA image sequence without reduction of stereoscopic effect. It can improve visualisation and interpretation of typical FA features by allowing the user to concentrate on specific areas of interest, which has the potential to ease the decision-making process.
Keywords: imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • retina