Pediatric White Matter Tract Analysis And Landmark Localization Of Human Brain

Magnetic resonance imaging(MRI)can obtain intra individual structural images in a non-invasive manner.Diffusion tensor imaging(DTI),as a special form of MRI,capturing the diffusion of water molecules,provides a noninvasive assessment of microstructural organization of white matter(WM).The quantitative analysis of the above two kinds of image data is of great significance to the cognition of the internal structure of the human body,the study of the state of individual development,the prevention and diagnosis of clinical diseases.Prior to quantitative analysis,the images acquired by different individuals or the same individual at different times need to be firstly registered into the same space generally.However,the traditional voxel-based grayscale domain registration method has a large amount of calculation,requires template prior and derives registration error.Therefore,a new mentality raised for quantitative analysis of medical images in view of the above problems: First,for the analysis of pediatric WM tract development,we propose to separate entangled WM fibers in DTI into one single type of fiber bundles,and then perform statistical analysis on a common type of bundles from different subjects in the native space so as to circumvent the smoothing from registration algorithms.In addition,regarding the registration of brain MRI,we utilize the geometric correlation between landmarks for establishing an implicit regression model to locate the brain landmarks.The model solves the problem of computational complexity and the need for template priors.As for segmentation of WM fibers,a recently developed algorithm,density peaks(DP)clustering,demonstrates great robustness to the complex structural variations of WM tracts without any prior templates.Nevertheless,the calculation of densities,the core step of DP,is time consuming especially when the number of WM fibers is huge.In this paper,we propose a fast algorithm that accelerates the density computation about 50 times over the original one.We convert the global calculation for the density as well as critical parameter in the process into local computations,and develop a binary tree structure to orderly store the neighbors for these local computations.Hence,the density computation turns out to direct access of the structure,rendering significantly computational saving.Performing experiments on synthetic point data and the JHU-DTI data set and comparing results of our fast DP algorithm and existing clustering methods,we can validate the efficiency and effectiveness of our fast DP algorithm.Finally,we demonstrate the application of the proposed algorithm on the analysis of pediatric WM tract development.Additionally,recently developed regression based(including deep networks)methods typically learn a mapping from input features to individual landmark positions or transform parameters.These methods neglect the geometric correlations among landmarks,thus resulting in inaccurate localization,especially for the parcellation functional regions whose boundaries are composed of a bunch of landmarks.We build a shape energy function for landmarks on 3D MRI features and learn the gradient regression for the energy.Our method accelerates the iterative gradient calculation and accurately detects brain landmarks.Performing experiments on locating anterior commissure(AC)and posterior commissure(PC),we can validate the effectiveness of our implicit regression model;Finally,we apply the proposed model on the OASIS T1-weighted MR data set to detect three brain functional regions.Experimental results demonstrate its efficiency and effectiveness by comparing with the state-of-the-art.