Abstract— White matter fibre tractography is a non-invasive method for reconstructing three dimensional trajectories of fibre pathways. Fast Marching is one of fibre tracking methods in which co-linearity of principal eigenvectors determines the speed of front's evolution. In this algorithm effect of tensor’s eigenvalues are not considered. In the current work, the speed function of standard fast marching was modified by considering the strength of tensor’s eigenvectors. The proposed speed function has an adaptive Fractional Anisotropy (FA) weighted factor which can be set by type of brain’s environments (i.e. isotropic and anisotropic regions). This modification was found to have high accuracy for detecting fibres by reducing false pathways. The proposed method has performed high accuracy in detection of fibre crossing. Index Term – Diffusion Tensor Imaging (DTI), fibre tractography, Fast Marching (FM), FA Weighted Fast Marching (FAW-FM), adaptive weighted factor, Fibre crossing. I. INTRODUCTION iffusion Tensor Imaging (DTI) is a non-invasive tool, which measures random motion of particles that has been called diffusion or “Brownian motion”. In isotropic environments, molecules spread out equally in all directions whereas diffusion is restricted with myelin sheath of axons in anisotropic regions such as brain white matter. Diffusion in parallel direction of white matter tracts is at least twice F. Dargi is with the Department of Medical Physics & Biomedical engineering, Medical Sciences/University of Tehran, and with the Research Center for Science and Technology in Medicine, Tehran, Iran. (email: dargi@razi.tums.ac.ir) M. A. Oghabian is with the Department of Medical Physics & Biomedical engineering, Medical Sciences/University of Tehran, and with the Research Center for Science and Technology in Medicine, Tehran, Iran. (email: oghabian@sina.tums.ac.ir) A. Ahmadian is with the Department of Medical Physics & Biomedical engineering, Medical Sciences/University of Tehran, and with the Research Center for Science and Technology in Medicine, Tehran, Iran. (email: ahmadian@sina.tums.ac.ir) H. Soltanian Zadeh is with the Control and Intelligent Processing Center of Excellence, Department of Electrical and Computer Engineering, University of Tehran, Tehran, Iran and the Image Analysis Laboratory, Department of Radiology, Henry Ford Heath System, Detroit, MI, USA. (emails: hszadeh@ut.ac.ir, hamids@rad.hfh.edu) M. Zarei is with the Department of Clinical Neurology, University of Oxford and with the Oxford Center for Functional Magnetic Resonance Imaging of the Brain. (email: mojtaba@fmrib.ox.ac.uk) A. Boroomand is with the Department of Medical Physics & Biomedical engineering, Medical Sciences/University of Tehran, and with the Research Center for Science and Technology in Medicine, Tehran, Iran. (email: boroomand@razi.tums.ac.ir) faster than perpendicular directions [1]. Diffusion properties of neural pathways can be obtained by this imaging technique, where a symmetric 2 nd -order tensor is assigned to each image voxel. Principal eigenvector of each voxel’s tensor represents direction of diffusion [2]. One of the most important applications of DTI is white matter tractography, which non-invasively reconstructs three dimensional trajectories of white matter fibre pathways. Until now several tractography algorithms have been used to reconstruct neural pathways. The first fibre tractography method has been called Principal Diffusion Direction (PDD) techniques, which start tracking from user defined seed voxel and follow its principal eigenvector direction to enter next voxel. Tracking line is propagated until some stopping criterions are met. Other same algorithms are streamlines algorithms that reconstruct three dimensional trajectories from main eigenvector of each voxel using Runge-Kutta and Euler integration methods for solving differential equations in vector field of DTI. These techniques have discretization error for reconstructing fibres. Mori et al. [4] implemented FACT algorithm which eliminates this error. Global energy minimization algorithms are other tractography methods. Fast marching algorithm [5] is the basic one in this classification. In this algorithm a front is propagated from a seed point or volume of interest using defined speed function. The spreading speed is largest in parallel direction of fibre pathways. In this method a narrow band is defined around selected voxels which its voxels are candidate for front to propagate to them. Flow-based surface evolution fibre tracking method is an extension of fast marching consists of using all diffusion tensor information. This modification uses HARDI data in multiple direction cases [6]. Advanced fast marching is another modification. In this algorithm, four different speed functions based on a predefined threshold for their anisotropic index are determined for connecting any two voxels [7]. In previous fast marching based algorithms, only directions of eigenvectors determined the speed value and the arrival time of entrance to next voxel. An FA threshold was considered during the tractography for not entering to isotropic areas. In this paper, we consider the effect of tensor’s strength as well as directions of eigenvectors in DTI data. Thus, if speed values for entering two candidate voxels are equal or close to each other, velocity will be determined by the product of Fractional Anisotropy (FA) values of both selected voxel and candidate one. Modified Fast Marching Tractography Algorithm and Its Ability to Detect Fibre Crossing F. Dargi, M. A. Oghabian, Member, IEEE, A. Ahmadian, Senior Member, IEEE, H. Soltanian Zadeh, Senior Member, IEEE, M. Zarei, and A. Boroomand D Proceedings of the 29th Annual International Conference of the IEEE EMBS Cité Internationale, Lyon, France August 23-26, 2007. ThC03.3 1-4244-0788-5/07/$20.00 ©2007 IEEE 319