Specimens with very weak fluorescence or those that photobleach easily may be difficult to observe using a confocal microscope. Therefore, standard epifluorescence imaging is often a method of choice for these objects. Deconvolution is a computational image restoration technique that can remove the out-of-focus blur typical for epifluorescence images and improve both lateral and axial resolution. Deconvolution is best suited for relatively thin specimens (microorganisms, single cells, tissue sections) where it can bring significant improvement in resolution and constrast. We have two deconvolution software packages:

XCOSM (link here) is a freely available software and is installed on our dual-processor Linux machine. It offers several algorithms, from the very fast LLS (linearized least squares) to the computationally intensive, iterative EM (Expectation maximization) method, which gives the best quality results (see figure below).

AutoDeblur (Autoquant Imaging, link here) is a commercial software suite installed on a PC running Windows. It offers several processing algorithms, including the iterative blind deconvolution. It has a user friendly interface and allows batch processing for unattended deconvolution of multiple datasets.

Currently we offer the use of both deconvolution packages free of charge. For more information on deconvolution and for help with image acquisition for deconvolution, contact Stan Vitha.

Example: Deconvolution with the xcosm software.

Chlorophyll in the cyanobacterium Synechococcus sp. PCC 7942 is localized in a layer underneath the cell membrane. Chlorophyll autofluorescence was recorded with Plan Apochromat 63x/1.4 oil immersion objective and a Coolsnap cf digital camera. Sixty four optical sections were acquired with a z-step of 0.2 µm. The image stack was processed using xcosm using a calculated point-sperad function. Each data set is shown in the XY view (top panels) and in a size (XZ) view (bottom panels). The original image stack shows severe blur because of the out-of-focus signal. This has been partially corrected by deconvolution.