Methods of Quantification for Optical Coherence Tomography Angiography Image Analysis

Abstract

Purpose : Optical coherence tomography angiography (OCTA) image analysis is becoming more common, but methods of image analysis are not consistent between studies. We performed a prospective cross-sectional study to assess whether differences in OCTA image binarization and histogram adjustment methodologies affect quantification metrics. We also assessed whether variations in methods affects the differences in quantitative metrics between unaveraged and averaged en-face choriocapillaris images.

Methods : 21 eyes of 11 healthy individuals were enrolled. All eyes were imaged using a 3 x 3 mm foveal-centered scan protocol on the Carl Zeiss PLEX Elite 9000 swept source-OCTA. Quantitative metrics including vessel area density and vessel length, as well as number of choriocapillaris flow voids, total flow void area, and average flow void size, were calculated. All metrics were calculated each superficial plexus and choriocapillaris image, both unaveraged and averaged, using five different binarization thresholds and five histogram adjustment methodologies. A mixed-effects ANOVA model was used to assess for significant differences.

Results : For all metrics, there was a statistically significant difference between values from different binarization thresholding methods (p < 0.0001), as well metrics from different histogram adjustments (p < 0.0001). The binarization thresholds yielded different metrics when combined with variable histogram adjustments. The method of analysis affected the directionality of changes in imaging metrics between unaveraged and averaged images for all metrics.

Conclusions : The method of OCTA image binarization thresholding and histogram adjustment significantly alters quantitative metrics and can even affect the directionality of trends in comparing unaveraged to averaged choriocapillaris en-face images. Results obtained from different OCTA quantification procedures should be seen as valid only for that given method, and a consensus is needed across the research community on a consistent OCTA image quantification methodology.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.

 

Figure 1: Effect of different binarization thresholds on superficial plexus and choriocapillaris en-face OCTA images.

 

Figure 2: Effect of different brightness and contrast adjustments on superficial plexus and choriocapillaris en face OCTA images.