Image Deconvolution

Deconvolution is a powerful technique used to mathematically “deblur” microscopic images. ImageTrak uses the XCOSM algorithm developed by José Conchello and colleagues. A point-spread function (PSF) must be generated for the deconvolution to proceed. PSFs can be calculated from optical parameters characteristic of your system, or acquired using sub-resolution fluorescent beads. ImageTrak lets you average and register any number of bead images in 3-dimensional space, then extract a clean empirical PSF from the averaged image volume.

The window above shows an example dialog box that allows the user to enter optical parameters used to calculate theoretical PSFs. The window below is an example of a batch deconvolution set-up. Here the user selects any number of XYZ or XYZT data sets for unattended batch processing, useful because these calculations can take a long time to complete:

Above figure shows triple-labeled spinal axons imaged on a Nikon C1 LSCM. A: raw XY image showing the blue fiber and tiny clusters of Ca channels and RyR. B: After deconvolution. C: “tilted” YZ projection of area in B outlined by the 2 white lines. E: raw XZ projection of area in D; the red and green clusters are indistinguishable along the z-axis [arrowhead].

F: XZ projection of same area after deconvolution now shows the two proteins closely associated, but not completely overlapping [arrowhead] (from Ouardouz et al, Neuron, 2003).