Ultra-fast 4D microscopy and high throughput distributed deconvolution

Ultra-fast tri-dimensional time lapse (4D) imaging can record structures labelled by fluorescence-tagged proteins moving in all directions at speeds reaching 2-3 μm/sec. It has proven potential for describing hitherto hidden aspects of cellular processes and for studying protein function, but the approach is still in its childhood. Here, we detail a novel 4D acquisition system and the adaptation of a constrained iterative deconvolution algorithm, not widely used in light microscopy, which yields greatly improved image quality at much increased speed on a standard PC. Nevertheless, the amount of data generated by ultra-fast 4D-imaging, typically 2 to 3 Gb/hour in many thousands of film requires a lot of time to be processed off line. We present a solution based on network distribution, using available computers connected together in LAN, which we believe to be a first step towards online deconvolution.