Intraobserver reproducibility of a two-dimensional mapping of the optic nerve head perfusion

To evaluate the intraobserver reproducibility of a software designed to assess retinal blood flow with the Heidelberg Retina Flowmeter (HRF).

Ten subjects were consecutively recruited, and one eye of each patient was randomly selected for study. Blood flow measurements were analyzed by using an automatic full field perfusion image analysis (AFFPIA) program, which calculates the Doppler frequency shift and hemodynamic variables (flow, volume, and velocity) for each pixel. The resulting perfusion image is processed with respect to underexposed and overexposed pixels, saccades, and retinal vessel tree. Intraobserver reproducibility was calculated for the AFFPIA program. All the optic nerve heads were horizontally divided into three sections (superior, central, and inferior). The retinal blood flow was calculated in the superior and inferior section, and each section was further divided into three areas (temporal, nasal, and rim). The blood flow was evaluated for each area.

When the same observer analyzed the same image five times (intraobserver intraimage reproducibility), the AFFPIA coefficient of variation ranged from 0.5% to 5% in the temporal area, from 0.1% to 5.3% in the nasal area, and from 0.5 to 28% in the rim area. When the same observer analyzed three different images of the same section once (intraobserver interimage reproducibility), the AFFPIA coefficient of variation of flow measurements ranged from 1% to 7.3% in the temporal area, from 1.5% to 10% in the nasal area, and from 2 to 30% in the rim area.

Retinal blood flow measured by HRF and analyzed by AFFPIA had good intraobserver reproducibility. The reproducibility was significantly better in the temporal and nasal areas than in the rim area.