Abstract
Purpose :
To improve the visualization of CNV feeder vessels in swept source OCT-angiography data sets. This is needed due to the difficulty a fully automated segmentation method has with complex pathologies as well as subtle structures.
Methods :
OCT-angiography data sets of patients with type 1 and type 2 CNV as well as myopic CNV were collected using a Topcon Triton Swept Source DRI OCT. The patients had been treated with intra-vitreal anti-VEGF between 4 and 23 times. The resulting raw data was exported and enhanced using temporal as well as de-flickering noise reduction algorithms and specialized plugins in Adobe After Effects. A custom built method inside Maxon Cinema 4D was created in order to facilitate a semi manual override of the inherent layer detection. This method enables the user to manually place 10 points along the layer to be segmented. This is repeated every 50 to 100 frames depending on complexity of pathology. The system automatically places an adaptive cubic spline in the x/z axis and interpolates between the splines in the y axis. The resulting data can then be viewed both as summed or maximum intensity z-projects in ImageJ.
Results :
10 eyes of 10 patients were processed. In all 10 cases an effective noise reduction and increase in visibility of CNV feeder vessels through re-segmentation was achieved.
Conclusions :
Improvement of the feeder vessel visualization will increase our understanding of the effect that treatment with anti-VEGF is having on the vascular structures of lesions. The presented method is a step towards increasing the quality and clarity of such a visualization.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.