Study On Image Pre-processing Of Structured Illumination Microscopy Serial Tomography

Exploration to understand the work mechanism of the brain is a big challenge in the area of the international science and technology today, the neurons in the brain and the neural network structure composed of not only determines the organisms higher functions such as cognition, emotion, memory, also provide important basis for the research of brain diseases. In recent decades, the rapid development of various imaging technology, provides the reliable tool for the research of brain science. Especially rapid and continuous development of high resolution optical microscopic imaging technology to improve, makes the structure information of data sets contain complex neurons growing in massive amounts of way, for the brain research on the structure and function of the neural network provides a strong guarantee. Especially the emergence of a continuous fault based on structure light microscope imaging system has a wide range, high resolution of whole brain datasets for speed increase again. But because the imaging.Based on structure light microscope imaging system to obtain the full cranial nerve yuan data sets and contains cells in mice to construct information data set a variety of image degradation problems, put forward a set of automatic pretreatment solution, implement the containing neurons correction of complex data set, cell structures, etc. In this paper, the pretreatment of the proposed method consists of the following four parts:(1) using frequency domain trap and the maximum cross-correlation method to realize image streak artifact removal and cross section of the splicing reconstruction;(2) Based on morphological identification method empty area and gray level correction, and through the mean curve smoothing method and reference method respectively in the cross section and the section between the brightness non-uniformity correction;(3) Set up the section axial bearing voxel alignment methods between unity and channel;(4) Based on MATALB parallel toolbox design level section of fine-grained parallel scheme, achieve rapid pretreatment.Fluorescent protein labeled with PI staining of mouse whole brain datasets this method after treatment, all brain regions uniform brightness, image quality compared with before treatment significantly improved, the corrected image can clearly show the structure and cell neurite shape. Uniform brightness, high-quality portfolio can be combined with whole brain nerve tracks, cell division and cell morphology body detection method for neurite quantitative cell body and neuronal cell types of calculation and analysis, neuroscientists reveal the working mechanism of the brain to provide reliable basic data sets.