Purchase this article with an account.
JIE WANG, Xiaohui Hua, Liang Liu, Thomas S Hwang, David Huang, Yali Jia; Automated registration and montage algorithm for wide-field OCT angiogram. Invest. Ophthalmol. Vis. Sci. 2018;59(9):2844.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In current optical coherence tomography (OCT) systems, the mismatch in the curvature of its focal plane and the retina causes distortion which degrades the quality of montaged images. We propose an automated algorithm that compensates for this mismatch to create a seamless, wide-field OCT en face angiograms from multiple images.
We obtained 6x6-mm OCTA volume scans centered on the disc, macular and temporal regions using a commercial system (AngioVue, OptoVue, Inc.). En face inner retinal angiograms from a slab bound by the inner limiting member and the outer boundary of outer plexiform layer were created using custom toolkit. An automated algorithm registered and montaged these images. First, the algorithm applied global affine deformable transformations on the input angiogram based on the matched invariant features between reference and moving angiograms. Then a local affine transformation adjusted mismatched vasculatures and the flow signals between two angiograms were equalized. Finally, the seams between angiograms were removed by a blending scheme in multi-scales. We evaluated the algorithm by calculating the correlation and the difference of flow signals between single projected B-frame from the opposing sides of seams.
En face OCTA from 10 eyes with diabetic retinopathy and 10 healthy controls were processed and evaluated. The seams between the angiograms can be easily easily identified on unregistered montaged angiogram. In contrast, the montage created from the automated algorithm shows continuous and coherent microvasculature in the large view without discernible seams (Figure 1). The analysis of the difference between the B-frames of the opposing sides of the seams disclosed high correlation and minimal difference (Table 1).
An automated registration and montage algorithm can generate wide-field OCT angiogram with high accuracy and improve the visualization of capillary plexus on OCTA. This method could improve the visualization and the quantification perfusion in wide-field OCTA.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.
Table 1. Evaluation of non-registered and registered algorithms by the correlation and difference of flow signals on seams.
Figure 1. (A) Wide-field OCT angiogram (14 × 15 mm) of a healthy eye registered and montaged by sixteen 6-mm OCTA scans; (B) Wide-field OCT angiogram (6× 12 mm) of an eye with diabetic retinopathy, registered and montaged by three 6-mm OCTA scans.
This PDF is available to Subscribers Only