The effects of LSO-based tracking resolution on OCTA image quality

Abstract

Purpose : Line Scan Ophthalmoscopes (LSO) are commonly used in Optical Coherence Tomography Angiography (OCTA) systems to image the fundus of the eye in order to track and correct for motion. Errors in such corrections can lead to artifacts in the OCTA enface image, visible as vessel jaggedness. We analyzed the effects of fundus tracking resolution on these common artifacts.

Methods : PLEX^® Elite 9000 (ZEISS, Dublin, CA) with LSO-based tracking was used to acquire 3x3mm OCTA scans, centered at the fovea, on normal eyes and eyes with retinal disease. Diseased subjects were imaged using one of two acquisition settings:
1) Group 1 (n=15): An LSO with 50µm x 40µm tracking resolution
2) Group 2 (n=15): An LSO with 25µm x 25µm tracking resolution

In addition to the higher resolution LSO, Group 2 also used an improved version of the FastTrac^TM algorithm. 14 normal eyes were scanned using both LSO configurations while the subject was asked to repeat a systematic eye motion pattern during acquisition. Given that in dense OCTA cubes neighboring B-scans have similar vessel morphology, a custom algorithm quantified vessel jaggedness by computing the normalized cross-correlation and lateral displacement between consecutive B-scans. The superficial retinal layer slab of each scan was analyzed with this metric, and the average displacement across the image was used to compare the two groups. Statistical significance was assessed by a two sample t-test with an alpha level of 0.05.

Results : 30 diseased eyes were imaged (15 per group). Scans from Group 2 showed a significant reduction of artifacts (p=0.01), compared to Group 1. The average displacement observed in Group 1 was 2.2±0.55 µm/B-scan (Mean ± SD), compared to displacement values of 0.53±0.51 µm/B-scan for Group 2. Figure 1 shows a comparison of two OCTA enface images of a normal eye using both LSO configurations. The lower resolution acquisition (1A) shows increased vessel jaggedness (1.93 µm/B-scan), as highlighted by the arrows. Conversely, the higher resolution LSO acquisition (1B) results in a cleaner image, with fewer artifacts (0.53 µm/B-scan).

Conclusions : Our study shows that higher resolution fundus images used to track eye motion yield more anatomically accurate OCTA images and, thus, improve image quality.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.

 

Figure 1: OCTA comparison of normal eye using lower (A) and higher (B) LSO-tracking resolutions. Red arrows point to artifacts resulting from imperfect tracking corrections.