Contribution of data pre-processing to deconvolution of 3-D fluorescence microscopy images

3-D optical fluorescence microscopy is an efficient tool for volumic investigation of biological samples. Nevertheless the image acquired by this way is altered by the properties of the microscope, according to its Point Spread Function (PSF). The aim of deconvolution algorithms is the reassignment of defocused information. This method provides an improvement in data quality and the possibility to compare specimens acquired using different systems. But deconvolution requires making a compromise between the precision of the result and the stability of the process, since this stability is directly related to the noise level of the data. This noise can be of different types, mainly electronic noise due to the sensors but we also include in the term "noise" the variation of fluorescence during the acquisition. Numerous deconvolution algorithms exist, giving variable results according to specimen characteristics. For the cases where deconvolution is not enough to obtain usable data, we developed some pre-process treatments. These tools can be used separately or consecutively depending on the application needs and specimen requirements.