Research On The Imaging Method And System Of Neuron Somas In Large Volume Samples

The knowledge of brain function and mechanism acquired from brain research is vital to neurological diseases treatment and artificial intelligence development.To acquire this knowledge,a basic demand is to obtain the distribution information of neuron somas in brain.Acquiring this information includes brain imaging,registration to reference atlas,brain region segmentation and soma counting.As the brain research important foundation,researchers' demands for high-throughput acquiring the somas information in the whole brain keeps increasing.Traditional methods manually sliced,imaged and registered the brain,which time and labor consuming,and error-introduced.Micro-optical sectioning tomography acquired neuron fine structures,but sample preparation and imaging were time-consuming.Light sheet microscopy conducted rapid imaging at the cell resolution,but clearing process took days.Sample deformation by clearing solution increased the registration difficulty,and fluorescence quenching affected the cell counting.Besides,existing registration methods usually needed background channel to ensure the accuracy,which brought large data size and long processing time.From the current publications we can see that data acquisition and analysis could not complete within a day.Due to multiple processes,large volume,and large number of samples,how to obtain the soma distribution and analyze accurately in a high-throughput way has become a problem needed to be solved in brain research.Aimed at problems mentioned before,this paper systematically researched imaging methods and strategy,data process,established whole workflow including sample preparation,image acquisition and data processing,applied it to specific brain study and received convinced results.The main work includes:(1)Developed inclined light sheet imaging methods free from optical clearing,realized fast sample preparation and data acquisition.Studied the relationship between scanning thickness and imaging speed,and the scattering and absorption effect of the tissue on the signal variation along the imaging depth,a proper thickness of imaging layer was selected to ensure the imaging speed and quality.Then proposed a fast surface-scanning strategy based on inclined light sheet towards the unclear sample,and tomography imaging to conduct large volume fast imaging.Based on the imaging principle and strategy mentioned above,designed and developed a fluorescence tomography system based on inclined scanning and vibration sectioning.At a voxel size of 2.34×2.34×1.65 ?m,the sample preparation and imaging acquisition of a whole mouse brain can be achieved within 4-7 hours.(2)A light sheet data processing and analysis method was developed to achieve fast precise brain registration,brain region segmentation and cell counting.The data processing method solved the problem of image misalignment,reduced the brain data size to speed up data processing and analysis.A background enhancement methods was used to achieve fine-registration on single-channel data.A high axial resolution atlas was established to achieve fine segmentation of brain regions,especially small regions.The numbers of soma in different brain regions were counted.The method proposed here could acquire the neuron soma localization information in brain within 10 hours,compared with days to weeks costing by existing methods,which greatly promoted the throughput.So far this soma localization system has been applied for brain function research.100 mice brain data has been obtained at a rate of one mouse per day.Collaborated with biology researchers,two brain function research were selected as a demonstration to show the potential applications of the method.In conclusion,this method provided a powerful tool for rapid identification,samples screening,hypothesis verifying,and brain function resolving in neuroscience,which has wide application prospect.