Study: OCTA analysis tools report different macular vessel density values for glaucoma

A recent study published in Optometry and Vision Science compared the results from 2 open-access optical coherence tomography angiography (OCTA) analysis tools, Erlangen‐Angio (EA) and OCTA vascular analyser (OCTAVA). The study, led by first author Kowsigan Magesan, MOptom, found that the tools showed similar reliability, but returned different macular vessel density (VD) and foveal avascular zone (FAZ) values in patients with glaucoma.¹ “By addressing these aims, this study provides evidence on the reproducibility and practical utility of open-access OCTA tools in both clinical and research settings,” the study authors stated.¹ The authors concluded that these tools cannot be used interchangeably to monitor glaucoma progression.¹

Background

OCTA provides noninvasive, depth-resolved imaging of the retinal and choroidal microvasculature.² Examining vascular structures is important for glaucoma management, with reduced retinal perfusion reported in prior studies. Growing evidence suggests vascular factors play a role in glaucoma development, with reduced retinal perfusion reported in affected eyes. ^3,4 However, quantitative OCTA metrics depend on the postprocessing and thresholding methods used, which limits comparability across studies and platforms.¹

The EA-tool and OCTAVA are nonproprietary, MATLAB-based toolboxes for semiautomated macular OCTA analysis. Earlier repeatability studies of each tool were conducted in healthy participants, within a single session, using separate datasets.¹

Study design

The analysis used data from the Individualised Perimetry Progression Observations in Glaucoma study (ACTRN12623000961640), a single-center, prospective longitudinal study at the Lions Eye Institute. Of 137 enrolled participants, 93 met image inclusion criteria (mean age, 67 ± 10 years; mean deviation, −5.84 ± 4.40 dB). Diagnoses included primary open-angle, pseudoexfoliative, and pigment dispersion glaucoma.

Foveal-centered 10° × 10° superficial capillary plexus scans were acquired on the Spectralis HRA + OCT with OCTA module (Heidelberg Engineering) at 2 visits approximately 113 days apart. The same images were processed with EA-tool version 1.0 and OCTAVA version 2.3. The EA-tool uses global (Otsu) thresholding, while OCTAVA uses local (fuzzy mean) thresholding. Investigators measured VD in 4 sectors, FAZ area, and FAZ circularity. The repeatability analysis included a portion of the cohort (n = 76) with qualifying images at both visits. Visual field tests were conducted on the Compass fundus perimeter (CentreVue SpA, iCare OY) using the Zippy Estimation by Spatial Test FAST strategy on a 24-2 grid.

Results

In the linear mixed-effects model, OCTAVA produced higher VD estimates than the EA-tool overall, with a mean difference of 8.42% (P < 0.001). The largest between-tool discrepancy was in the inferior sector (Sector × Tool interaction, −2.73%; P = 0.0009). No visit-related interactions were detected (P > 0.21).¹

OCTAVA also measured larger FAZ area (estimated mean difference, +0.066 mm²; P < 0.001) and lower FAZ circularity (estimated mean difference, −0.0577; P < 0.001) than the EA-tool.¹

Repeatability coefficients for sectoral VD and FAZ metrics ranged from 0.09 to 7.65 for the EA-tool and 0.11 to 9.53 for OCTAVA. The study reported overlapping intraclass correlation coefficient confidence intervals and comparable CV values for both tools. However, FAZ circularity did not meet the authors’ reliability criteria with either tool. Higher average VD was associated with better central visual field sensitivity with both tools (EA-tool: β = 0.38; OCTAVA: β = 0.57; P < .001). FAZ metrics were not associated with central visual field sensitivity.¹

Impact and limitations

The study authors attributed the between-tool differences in macular vascular metrics to thresholding methods. Global thresholding may overlook true flow due to regional variation, while local thresholding may count noise as flow. They added that OCTAVA’s required image upscaling may inflate VD measurements. According to the authors, absolute values should not be compared across platforms, and the same tool should be used consistently when monitoring progression. Based on the repeatability coefficients, the authors reported that an intervisit change in average VD of more than ~5% with the EA-tool, or ~6% with OCTAVA, is unlikely to result from measurement noise alone. They recommended FAZ circularity not be used for progression assessment with either tool.

The authors suggested FAZ variability may reflect subjective manual marking by graders, image quality, and segmentation error. The 3 × 3 mm scan area does not cover all central visual field test locations. In terms of study methods, the months between visits may have captured some genuine change, and FAZ measurements were not corrected for ocular magnification because axial length data were unavailable. The structure-function analysis used the central 4 points of the 24-2 grid rather than the 10-2 test used in prior studies.

The authors concluded by highlighting how their findings can aid clinicians and researchers assess the reliability of OCTA analysis tools. “These insights emphasize the necessity of understanding tool-specific biases and selecting robust, reproducible parameters when interpreting OCTA data in both research and clinical practice,” they stated.¹

References

1. Magesan K, Tang VTS, Salmanzadeh S, Morgan WH, Chan G, McKendrick AM. Macular microvascular assessment in glaucoma: tool output comparison and intersession reliability. Optom Vis Sci. 2026;e70099. doi:10.1002/ovs2.70099

2. Chen CL, Wang RK. Optical coherence tomography based angiography. Biomed Opt Express. 2017;8(2):1056-1082. doi:10.1364/BOE.8.001056

3. Lee JY, Shin JW, Song MK, Hong JW, Kook MS. Glaucoma diagnostic capabilities of macular vessel density on optical coherence tomography angiography: superficial versus deep layers. Br J Ophthalmol. 2022;106(9):1252-1257. doi:10.1136/bjophthalmol-2020-318449

4. Triolo G, Rabiolo A, Shemonski ND, et al. Optical coherence tomography angiography macular and peripapillary vessel perfusion density in healthy subjects, glaucoma suspects, and glaucoma patients. Invest Ophthalmol Vis Sci. 2017;58(13):5713-5722. doi:10.1167/iovs.17-22865