Colocalization error between the scanning laser ophthalmoscope infrared reflectance and optical coherence tomography images of the heidelberg spectralis
PURPOSE: To examine the colocalization error between the infrared reflectance (IR) scanning laser ophthalmoscope (SLO) and spectral domain optical coherence tomography (SD-OCT) images of the Heidelberg Spectralis. METHODS: The IR and corresponding horizontal raster SD-OCT images were compared in 10 healthy volunteers examined with 3 Heidelberg Spectralis + OCT instruments. The center points of retinal vessels selected by random uniform sampling in scanning laser ophthalmoscope-IR images were compared with colocalizing points in corresponding SD-OCT images by two masked readers. The error of colocalization was measured in the SD-OCT image. The point positions were recorded using Cartesian coordinates measured in microns. The error of colocalization was evaluated using a fixed-effects generalized least squares regression model with location and instrument as predictor variables. RESULTS: A total of 1,617 points in the IR and SD-OCT images were analyzed, and the mean error of colocalization was 42.2 +/- 32.4 mum. The Heidelberg Spectralis instrument used was not a significant predictor of colocalization error (P = 0.15). The colocalization errors were negatively correlated with x-coordinate position (P < 0.001). CONCLUSION: There is a mean colocalization error between the IR and SD-OCT images produced by the Heidelberg Spectralis that is on the scale of many features being evaluated in the fundus. The variability in these measurements means the confidence interval for the exact colocalization is much larger. Because of the magnitude and variability of the error, the colocalization feature of the Heidelberg Spectralis should be used as a rough guide, not an absolute determinant.