Purpose:
Registration artifact can occur when different imaging techniques are applied to the same tissue, or when imaging in living subjects. Our goal was to determine whether image registration applied to two-photon microscopy images of the trabecular meshwork (TM) could improve colocalization results.
Methods:
Two-photon microscopy was used to obtain optical slices of the entire TM from ex vivo human corneo-scleral tissue. This included two-photon excitation fluorescence, or autofluorescence (AF), and second-harmonic generation (SHG) images (Fig 1A, 1B). Intensity-based image registration was performed in Matlab (Mathworks, Natick MA). Specifically, a sum-of-squares similarity score was calculated for each pair of source and target images, while allowing the target image to shift in the x,y (horizontal) and z-axis (vertical). Image registration was first tested on positive controls by applying the algorithm to identical and nearly-identical images that had been randomly shifted (data not shown). Then the algorithm was applied to the test image stacks. Colocalization analysis was performed on the test image stacks using ImageJ (http://imagej.nih.gov/ij/) both before and after image registration.
Results:
Colocalization analysis prior to image registration resulted in Mander’s Correlation Coefficients for pixels above threshold (tM1 and tM2) of 0.1566 and 0.1837 (range 0-1) for AF and SHG, respectively (Fig 1C). The number of colocalized pixels overall was 164,591, representing less than 2% of total volume. After image registration, tM1 and tM2 improved to 0.7637 and 0.3751 for AF and SHG, respectively (Fig 1D). The number of colocalized pixels increased more than five-fold to 1,010,426 (>11% of total volume).
Conclusions:
Automated image registration improved subsequent colocalization results in two-photon microscopy imaging of the TM, a complex, three-dimensional structure. This type of image processing will become especially important when imaging live subjects because of entities such as movement artifact.
Keywords: trabecular meshwork • microscopy: light/fluorescence/immunohistochemistry