Investigation On The Construction Of Three-dimensional Digital Atlas Of The Visual System In The Pigeon Brain

As a classic animal model for investigation learning,memory,and cognition,pigeons has been widely used in the research of avian neurobiology outside of the domain of the song system.In the traditional neural science study,researchers could only by reading the anatomical atlas from paper and mostly depend on the experience from experiment to complete the complicated surgery positioning and push the microelectrodes to the specific location,this method leads to high failure rate and extreme long time to finish the surgery.Compare with it,3d digital brain atlas is more intuitive,researchers can fully understand the target area and surrounding structures,obtain the precise coordinates before experiment,so it can improve the efficiency of the study of physiological experiment,and has attracted more and more anatomists' attention.MR image is a non-invasive medical imaging technology,it plays a more and more important role in the neuroscience research.In this paper,the improved non-local denoising methods are used to preprocessing MR image;Then,using a Coarse-to-Fine approach to registration conventional atlas to MR image.In this way to extract the ROI(Region of Interesting)in MR image.Finally,design the 3d digital atlas of pigeon brain with visual studio and VTK.The main task is as follows:(1)In order to remove the Rician noise in MR images sufficiently,the Normalized Cross Correlation(NCC)of local pixel is proposed to characterize the geometry of a similar nature,and combine it with the traditional method of using only pixel intensity to determine its similarity weight.Then the improved method is applied to non-local mean and non-local linear minimum mean square-error estimation algorithm.In order to realize adaptively denoising,the weighted value of pixel to be filtered or the similarity threshold in non-local algorithms is computed according to the local signal-to-noise ratios(SNRS)dynamically.The results show that the proposed algorithm can not only better inhibit the Rician noise in MR images,but also can effectively preserve image details.(2)For the existence disadvantage of MR image and conventional atlas,combine the Snake-based segmentation method and contour-based registration method to complete the coarse registration.Then,to realize nonlinear registration,use particle swarm optimization algorithm to optimize the registration parameters.(3)Using c# program language and the visualization toolkit(VTK)to design the 3d digital atlas system of pigeon brain.In this paper,the system can scaling or rotation the ROIs from any view,and according to need,it has other interaction operation,for example volume and surface area measurement.It can also match the atlas with the angle of pigeon head in the real experiment,and get the coordinate of ROI through mouse picking or other method,then use it in the experiment.The results show that the three-dimensional digital atlas of the visual system in the pigeon brain by this process is easy to implement,and effectively avoid the error.The operation is simple,can satisfy the requirement of neural electrophysiological experiments.