Purpose: The study used computational methods to analyze vessel length between branch points from patient OCTa images, in order to explore whether these measurements could serve as biomarkers in retinal detachment. Study Design: Retrospective cross-sectional study.
Methods: We used computational analysis to compare distances between vessel branch points in patients who had retinal detachment (RD) in a given eye to their contralateral (control) eye. This was done to determine if the distances between branch points could serve as biomarkers for retinal displacement. In order to set objective thresholds for which vessels to analyze, we used the AngioTool software from Zudaire et al. The AngioTool program selects vessels based on intensity (the brightness of the vessels on the black and white OCTa image), and vessel width. Additionally, it selects branch points by mapping regions of the image that show vessel forking. From these processed images, we identified the first branch point as the branch point closest to the foveal avascular zone (FAZ), and the second and third branch points as the subsequent branch points towards the periphery of the image. The coordinates of these branch points were then recorded. The distance between branch points in the horizontal, vertical, and diagonal distances were calculated. Between control and RD eyes, the differences in these mean distances were also calculated.
Results: OCTa images from 76 eyes (38 patients) were included. 38 eyes had recent RD, and 38 eyes were patients’ contralateral control eyes. Of RD eyes, 25 received pneumatic retinopexy (PnR), 11 pars plana vitrectomy (PPV), and 2 scleral buckling (SB) surgery. 33 had Fundus Autofluorescence (FAF) imaging gradable, and 10/33 eyes showed retinal displacement through the presence of Retinal Vessels Printings (RVP). The comparisons of horizontal (X), vertical (Y), and diagonal (square distance) showed no significant differences in control versus RD eyes from branch points 1 to 2. There were no significant differences in vertical or square distances in control versus RD eyes from branch points 2 to 3. However, the difference in the horizontal distance from branch points 2 to 3 between control and RD eyes was found to be significantly different. The horizontal distance between branch points 2 to 3 in control eyes had a mean of 172.87±64.24 uM, while the same distance in RD eyes was 137.71±74.43 (p=0.031). Between branch points 1 and 2 in PnR and PPV eyes, PnR eyes had a significantly greater (p<0.05) vertical (Y) distance and significantly greater square distance than PPV eyes. No significant differences in distance were seen when comparing lengths from branch point 2 to 3 in these same groups.
Conclusions: The use of computational methods to assess retinal displacement holds promise. Due to the ubiquity and ease of OCTa following surgery, retinal imaging allows clinicians and researchers to visualize the foveal vasculature well. Using computational methods and landmarks such as branch points, retinal displacement can possibly be examined. However, as demonstrated by this exploratory study, measurements of vessel length by using retinal branch points, does not appear to be a useful biomarker for retinal displacement. Future directions include assessing total vascular area and average vessel length as biomarkers.

Computational vessel analysis in retinal detachment patients: Mapping retinal displacement using OCTa vessel branch points

[Poster Abstract Booklet](https://drive.google.com/file/d/1AGitRGyG9W0FiZ2Xjz6RNO2h7C0XglOL/view)

[Surgical Video Abstract Booklet](https://drive.google.com/file/d/1B-IkojrZhXk4AgXzsEQyuE_FwPG1Lvpe/view?usp=sharing)

[To claim Section 2 credits for ePoster review complete this Poster Transcript](https://drive.google.com/open?id=1dSAjpIIqFqD4SUbpcc6-PzX65ZPFhA7O)