Robust OCT axial motion correction by using multiple pairs of orthogonal scans

Abstract

Purpose : Axial motion correction (AMC) is essential for retinal multilayer segmentation and visualization in optical coherence tomography (OCT) volumes. Utilizing a pair of sparse scans orthogonal to the OCT volume scan is a common technique for AMC, but it can suffer from low OCT image contrast if the orthogonal scans cross the optic nerve head (ONH) or large blood vessels. We propose a solution to address these problems using three pairs of sparse orthogonal scans.

Methods : The axial motion correction algorithm aligns the B-scans in the slow scan direction to the corresponding orthogonal B-scans followed by applying the AMC to the OCT volume scan. One pair of orthogonal B-scans is defined as two B-scans at the edges of the lateral position orthogonal to the volume scan (Fig. 1). Using three pairs of orthogonal B-scans (at different lateral positions) may achieve more robust AMC by aligning each section of the slow B-scan to the corresponding orthogonal B-scan section using a different pair specifically when intra-scan quality varies (Fig. 2). The algorithm was performed on 83 12x12 mm, 70 15x15 mm, and 80 23x23 mm scans with a 6 mm depth of varying quality from healthy eyes (20 subjects) using PLEX^® Elite 9000 (ZEISS, Dublin, CA). A grader reviewed the motion-corrected B-scans in the slow direction for both methods based on one pair (method (I)) and three pairs (method (II)) of orthogonal scans. The performance of the AMC was evaluated subjectively by comparing the position, shape, and curvature of the slow B-scans with corresponding orthogonal B-scans (Fig. 2). The success rates for both methods were reported.

Results : Fig. 2 shows four examples for AMC using method (I) and method (II). The motion correction failure is highlighted in the ONH and periphery. Two challenging scans and orthogonal scans with varying intra-scan quality were shown. The AMC of the method (II) performs better than method (I). The success rates are 92% and 97% using method (I) and method (II) respectively.

Conclusions : The method of OCT AMC using three pairs of sparse orthogonal scans outperforms the standard method of AMC using only one pair of sparse scans.

This abstract was presented at the 2024 ARVO Imaging in the Eye Conference, held in Seattle, WA, May 4, 2024.

 

Fig. 1: Example of an uncorrected OCT volume scan, a pair of orthogonal B-scans and corresponding slow B-scans.

 

Fig. 2: Four examples of AMC using one pair (method (I)) and three pairs (method (II)) of orthogonal scans. The motion correction failure is marked with a yellow dashed box.