Inter-Observer Reproducibility of Heidelberg Retinal Flowmetry Utilizing Pixel by Pixel Analysis: Reading Center Analysis

Abstract

Purpose: : The purpose of this investigation was to determine the inter-observer reproducibility of Heidelberg retinal flowmeter (HRF) blood flow measurements performed by two independent observers utilizing a pre-set reading center protocol with independently selected areas for pixel by pixel analysis.

Methods: : Blood flow analysis was performed on 257 scans from 15 patients with primary open angle glaucoma and/or ocular hypertension. Capillary perfusion in a 2560x640 µm area of the superior temporal peripapillary region was recorded and pixel by pixel analysis was performed by two independent observers. Measurements were taken from an area adjacent to the optic disc with a minimum of 1600 pixels collected by each observer. Each observer using a pre-set reading center protocol independently selected the area for their analysis. The percentage of pixels with less than 1 unit of flow (zero flow) and 10, 25, 50, 75, and 90^th percentile flow values of total flow were calculated. Inter-observer variability was assessed by estimating the intra-class correlation coefficient (ICC) and 95% confidence intervals (CI). Bland-Altman plots of the difference between the two observers versus the average of the two observers for each outcome were created.

Results: : ICC was 0.79 (range: 0.74-0.83) for HRF mean retinal blood flow values of the superior temporal peripapillary region. For zero flow, 10, 25, 50, 75 and 90^th percentiles of flow, the ICC and ranges were 0.67 (0.60-0.73), 0.74 (0.68-0.79), 0.82 (0.78-0.86), 0.85 (0.82-0.88), 0.85 (0.81-0.88) and 0.77 (0.72-0.82), respectively.

Conclusions: : This study demonstrates that independent observers utilizing a pre-set reading center protocol can review HRF retinal capillary blood flow scans and obtain statistically similar results. We report acceptable reproducibility in both the number of zero flow pixels as well as within the stratified percentiles of total retinal blood flow utilizing the pixel by pixel method.

Keywords: image processing • imaging/image analysis: clinical • blood supply