Investigation On Automatic Segmentation Of Macaque Brain Tissues And Anatomical Structures Based On 7T MR Images

Using the rhesus macaque as an experimental model to explore the brain mechanisms has been an important means in the field of neuroscience.Typically,the macaque brain atlas is playing a vital role in the spatial orientation of the macaque brain.However,macaque brain imaging research is basically done on a few individuals,so it is particularly important to construct an accurate subject-specific digital three-dimensional atlas of macaque brain in time.However,most MR images based on which the brain labeling are derived are acquired from at or below 4.7T.Compared to what ultra-high-field MRI can possibly enrich us,the existing MR template images of the macaque brain atlas may have limited spatial resolution and hereby lose detailed anatomical information during constructing process.Moreover,the MR brain images of rhesus macaques are easily influenced by noise,inhomogeneity of the grayscale,complexity and diversity of brain structure,etc.,which lead to a very difficult,time-consuming and laborious segmentation of macaque brain.In this paper,we propose a simple and easy method for the automated segmentation of macaque brain based on in vivo MR Images acquired at 7T.The main work of this paper is as follows:First,we summarized the existing segmentation methods over MR images,and then an image segmentation method based on atlas registration was determined and the algorithm prototype's advantages and disadvantages were analyzed.Secondly,six anesthetized macaques were imaged by Siemens MAGNETOM 7T whole-body scanner.After a series of pre-processing steps,the obtained T1w images composed the macaque brain template collection.Then,the atlases of macaque brain were obtained by using the image segmentation method based on the single atlas registration,and the label images of macaque brain was obtained.Finally,based on the labeled images of macaque brain,the MR images of macaque brain were segmented using the special segmentation method derived from multi atlas registration,whose processing steps formed a feasible approach for automatic segmentation of macaque brain.Typically,the label images of macaque brain have two modes:the anatomical label images which consisted of 224 different anatomical brain structures and the label images of tissues consisting of GM(gray matter),WM(white matter),CSF(cerebrospinal fluid)and BV(blood vasculature).