Editors Selection IGR 19-4

Clinical Examination Methods: Retinal vessel density

Harsha Rao

Comment by Harsha Rao on:

79201 Comparison of methods to quantify macular and peripapillary vessel density in optical coherence tomography angiography, Rabiolo A; Gelormini F; Sacconi R et al., PLoS ONE, 2018; 13: e0205773

Find related abstracts

OCTA is a relatively recent, non-invasive, dye-less imaging modality that enables visualization of the vasculature of the retina, optic nerve head and the choroid. Delineation of the vasculature is achieved using algorithms that evaluate variability produced by moving columns of blood in sequentially acquired OCT B-scans. Each OCTA manufacturer has their own proprietary algorithms to delineate vasculature. The most popular vascular parameter quantified on the OCTA scan is the 'vessel density'. Vessel density is the percentage of angio-cube occupied by vessels.

Patients were recruited from all other the world, and thus results are generalizable to the worldwide pediatric population

Previous studies have used multiple thresholding methods to binarize the OCTA image and measure the vessel density from the exported raw image; some of them being fixed cut-off method, dynamic cutoff method, Image J methods, etc. Rabiolo and colleagues examined the agreement in macular (superficial and deep capillary plexus) and peripapillary (radial capillary plexus) vessel density measurements among seven such binarization methods. Their study included 44 eyes of 27 healthy subjects, 44 eyes of 31 subjects with diabetic retinopathy (DR) and 44 eyes of 26 subjects with POAG. They found that the agreement between the different binarization methods was poor with substantial systemic, non-constant biases evident between most tested methods. For example, the mean difference in peripapillary vessel density was as low as 1.4 between two methods to as high as 9.6 between two other methods. The range of 95% LoA between any two methods of peripapillary vessel density measurement was as low as 2.2 to as high as 28.8. The authors concluded that the vessel density measurements calculated with different methods are not interchangeable and longitudinal monitoring of vessel density in diseases such as DR and POAG should be carried out with the same method. The authors also compared the diagnostic ability of the vessel density measurements derived from these methods in DR and POAG separately and found that no one method was better than the others.

Previous studies have shown that the OCTA vessel density measurements are dependent on the device, angio-cube size, image averaging and signal strength. This study by Rabiolo et al. shows that the vessel density values are also dependent on the post-processing method used. In spite of its popularity, OCTA is still an evolving technology and these results highlight the fact that there is a need for consensus on the method that is best suited for quantification.

Issue 19-4

Select Issue