2-1) and 3-d deconvolution of confocal fluorescence images by maximum likelihood estimation

Deconvolution of confocal fluorescence images by maximum likelihood estimation (MLE) was investigated for its ability to increase the information content in the images. 3-D MLE algorithms, applied to confocal image stacks, increase lateral and axial resolution and result in a finer optical section. Contrast, especially at edges, is enhanced, improving the documentation quality of high magnification images beyond that possible by confocal microscopy alone. Axial smear associated with spherical aberration was not removed by deconvolution and a lateral thinning artifact was introduced. Single confocal images can be rapidly deconvolved by 2-D MLE by applying a two-dimensional point spread function and treating them as images of planar objects. The 2-D algorithm can also deconvolve a maximum projection of a stack. The method works best when there is a minimal overlap of fluorescent structures. The projection is treated as a two dimensional object. Intensity information excluded by the projection operation cannot be recovered. Deconvolution of images acquired with the pinhole opened to increase sensitivity closely matches images acquired through an optimal opening, although in 2-D MLE, colocalization cannot be distinguished from overlap and the integrity of quantitative data cannot be guaranteed. Properly applied, MLE deconvolution increases the useful information content of confocal images.