Abstract
Purpose :
Axial motion correction (AMC) is essential for OCT image analysis such as retinal multilayer segmentation (MLS). Using orthogonal retrace scans is an effective technique for AMC, but sometimes it suffers from axial bulk motion and low image contrast when the retrace scans are crossing the ONH or large vessels. Here we proposed a solution to address these problems by using OCT B-scan alignment.
Methods :
The axial motion correction algorithm starts first by identifying a reference B-scan in the OCT volume data. The reference B-scan is selected based on the B-scan quality (contrast intensity, and SNR) and central retinal position. Starting with this reference B-scan, the neighboring adjacent B-scans were registered using normalized cross-correlation in axial direction. This axial correction is repeated until the entire OCT volume is corrected axially. CIRRUS™ 6000 AngioPlex (ZEISS, Dublin, CA) was used to image 15 subjects, which was a mix of healthy eyes and diseased eyes, such as aged macular degeneration, macula edema, epiretinal membrane, and diabetic retinopathy. 107 scans in total were acquired with 12 scans of 3 x 3 mm and 95 scans of 6 x 6 mm angiography scans.
A grader reviewed the MLS results before and after axial motion correction to evaluate the performance of the axial motion correction.
Results :
MLS malfunctioned for 31 scans due to the axial motion. After applying the proposed AMC, MLS failed only for 1 case out of 107 scans. Figs. 1 and 2 show examples of the MLS performance before and after the axial motion correction. AMC reduced the failure rate of MLS from 29% to 1%.
Conclusions :
We have proposed an OCT B-scan based motion correction technique which is essential for a more robust retinal MLS.
This is a 2021 Imaging in the Eye Conference abstract.