The Effect of Different Optical Coherence Tomography Angiography Macular Scan Sizes on Foveal Avascular Zone Metrics in Patients with Diabetic Macular Ischaemia

Abstract

Purpose : Change in the foveal avascular zone (FAZ) metrics over time is a key outcome measure for clinical trials in diabetic macular ischaemia (DMI). However, artefacts and automated delineation of FAZ may influence the results. We aimed to compare the artefact frequency and FAZ metrics in the 3x3 and 6x6 mm macular scans obtained on Optovue optical coherence tomography angiography (OCTA) in patients with DMI.

Methods : This prospective observational study evaluated DMI in patients with stable treated proliferative diabetic retinopathy (DR) and severe non-proliferative DR. Baseline DMI was defined as an automated FAZ (aFAZ) ≥0.5 mm² or parafoveal capillary non-perfusion ≥1 quadrant. The types of artefacts and changes in the aFAZ versus corrected FAZ (cFAZ) metrics were assessed over 12 months on both size scans. The eye with better image quality was included if both eyes were eligible.

Results : Forty out of 48 recruited patients (mean age 58) with gradable scans were included in this analysis. The 6x6 scans had significantly more projection artefacts than the 3x3 scans. Both size scans were prone to segmentation artefacts. All eyes required manual delineation of the FAZ margin due to inconsistencies in the aFAZ. The FAZ in the 3x3 scans were mostly larger than the 6x6 scans (0.56 ± 0.29 vs 0.49 ± 0.27 mm², P <0.01 in aFAZ; and 0.72 ± 0.39 vs 0.71 ± 0.38 mm², P <0.01 in cFAZ) (Table 1). Over 12 months, both size scans showed a decreased trend in the aFAZ. However, after manual correction, the 3x3 and 6x6 cFAZ exhibited a significant increase (0.83 ± 0.40 to 0.89 ± 0.43 mm², P = 0.02, and 0.81 ± 0.39 to 0.88 ± 0.43 mm², P <0.01, respectively) (Table 2).

Conclusions : In DMI, both cross-sectional and longitudinal studies on the FAZ metrics are best evaluated on the 3x3 scan due to better resolution. However, manual correction of the FAZ margin is needed. A significant increase in the cFAZ area was observed over 12 months. The frequency of artefacts and delineation errors of the aFAZ margins suggest that further refinement of these automated FAZ metrics is required.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

 

Table 1. A comparison between the automated and corrected FAZ in two-sized scans.

 

Table 2. Changes over 52 weeks in the automated and corrected FAZ in two-sized scans.*Only eyes with good quality in both size scans from all visits were included.