Stephan Preibisch's multiview deconvolution algorithm and the associated Fiji plugin was published at Nature Methods. The approach is relevant as a fusion strategy for OpenSPIM data, which it improves dramatically due to relative thickness of the lightsheet.
Stephan Preibisch, Fernando Amat, Evangelia Stamataki, Mihail Sarov, Robert H Singer, Eugene Myers & Pavel Tomancak (2014) Efficient Bayesian-based multiview deconvolution Nature Methods AOP doi:10.1038/nmeth.2929 PDF, Supplement
Check out the extensive supplement that provides the derivation of the algorithm (for the mathematically inclined) and extensive evaluation and benchmarking against other approaches. The GPU code was developed by Fernando Amat from Philipp Keller's lab at the Janelia Farm.
The paper comes with an extensive collection of Supplementary Videos available at Nature Methods website.
If you are interested in the paper's genesis under intense but constructive scrutiny of reviewers, you may want to have a look at the various submitted pre-prints on arxiv: