We read with interest the article by Gadde and colleagues
1 on quantifying vessel density and the foveal avascular zone (FAZ) area in optical coherence tomography angiography (OCTA) images using a fully automated local fractal dimension method. The authors concluded that this novel method can calculate retinal vessel density and the FAZ area. They also contended that retinal vessel density was not correlated with age and that the inferior zone of retinal vessel density is largest in the parafoveal region. We congratulate and applaud their interesting and important work on this topic; however, we have one question for the authors regarding intra-and interobserver variation and the reproducibility of the OCTA examination.
It is well acknowledged that a new diagnostic tool should be validated regarding its reliability and reproducibility before its application. The previous methodologies for measuring ocular parameters such as retinal nerve fiber layer thickness,
2 choroidal thickness,
3 and retinal oxygen saturation
4 all have been tested for their reliability and reproducibility. As far as we are concerned, when the subjects underwent imaging with the OCTA system (AngioVue; OptoVue, Fremont, CA, USA), because of the relative long duration of the examination and unstable fixation, some subjects acquired low-quality images. If the subjects had repeated instances of unstable fixation, the image would appear with white ambiguous lines. The analysis software would mistake these white ambiguous lines for blood vessels and would overestimate the retinal vessel density. Therefore, intra- and interobserver variation would, in theory, be relatively large. We wonder why the authors did not perform a reproducibility analysis to prove the stability of the OCTA system examination. If such an analysis had been performed and intra- and interobserver variation exceeded a certain percent, the results of this study might have been shown to be unreliable.