Estimation of multiple fiber orientations from diffusion tensor MRI

A novel method has been investigated to identify the orientations of multiple fiber tracts in a voxel of diffusion tensor MRI (DT-MRI). Conventionally, white matter fiber tracking has been performed based on the assumption that the major eigenvector represents the orientation of the fiber tract. However, it has been acknowledged widely that the assumption is valid only if there is a unidirectional single fiber tract per voxel. Recent attempts to resolve the problem of multiple fibers often require long acquisition times and strong gradients with greater than 50 gradient-directions. Our approach is to keep the same imaging sequence as conventional DT-MRI with a relatively small (six) number of diffusion gradient directions, and to apply independent component analysis technique to separate the raw data into multiple data sets corresponding to multiple fiber tracts respectively. A simulation study was carried out to test the method, followed by application to human data.