Diffusion tensor magnetic resonance imaging including local weighted interpolation

An imaging method comprising:

• acquiring a three-dimensional apparent diffusion tensor map of at least a portion of the subject including anisotropic structures using magnetic resonance imaging;
• processing the apparent diffusion tensor at a voxel to obtain eigenvectors and eigenvalues;
• extracting a three-dimensional fiber representation using the eigenvectors and eigenvalues;
• during the extracting, locally interpolating voxels relative to a selected spatial dimension in a vicinity of the fiber representation, the interpolating including weighting the interpolated voxels by one of:
• a direction of a major diffusion tensor eigenvector,
• a functional combination of diffusion tensors, and
• a fractional anisotropy, to produce a three-dimensional fiber representation having a higher resolution than the acquired tensor map.

An apparatus for tracking fibrous structures in a subject, the apparatus comprising:

• a magnetic resonance imaging scanner configured to acquire diffusion-weighted imaging data;
• a reconstruction processor which reconstructs the acquired diffusion-weighted imaging data into diffusion-weighted image representations;
• a diffusion tensor mapping processor that constructs a diffusion tensor map by selectively combining selected diffusion-weighted image representations;
• an eigenvalues/eigenvector processor that determines ordered eigenvalues and eigenvectors of the diffusion tensor corresponding to voxels;
• a fibrous structure tracking processor that computes a fibrous structure representation based on voxel eigenvalues and eigenvectors and at least one starting voxel selection;
• an interpolation processor cooperating with the fibrous structure tracking processor that increases a spatial resolution of the fibrous structure representation by locally spatially interpolating voxels in a neighborhood of the fibrous structure during computation of the fibrous structure representation;
• and
• a display device that displays at least a portion of the fibrous structure representation in a human-viewable medium.

An apparatus for tracking fibrous structures in a subject, the apparatus comprising:

• a magnetic resonance imaging scanner configured to acquire diffusion-weighted imaging data;
• a reconstruction processor which reconstructs the acquired diffusion-weighted imaging data into diffusion-weighted image representations;
• a diffusion tensor mapping processor that constructs a diffusion tensor map by selectively combining selected diffusion-weighted image representations;
• an eigenvalues/eigenvectors processor that determines ordered eigenvalues and eigenvectors of the diffusion tensor corresponding to voxels;
• a fibrous structure tracking processor that computes a fibrous structure representation based on voxel eigenvalues and eigenvectors and at least one starting voxel selection, the fibrous structure tracking processor including:
• a directional processor that identifies a local fiber direction corresponding to an eigenvector direction associated with the largest eigenvalue of the selected voxel,
• an incremental processor that identifies a voxel near the selected voxel along the local fiber direction, and
• a looping processor that repetitively invokes the directional processor and the incremental processor to iteratively compute the fibrous structure representation;
• an interpolation processor cooperating with the fibrous structure tracking processor that increases a resolution of the fibrous structure representation by locally interpolating voxels in a neighborhood of the fibrous structure during computation of the fibrous structure representation;
• and
• a display device that displays at least a portion of the fibrous structure representation in a human-viewable medium.

A method for tracking fibrous structures in magnetic resonance images, the method comprising:

• generating magnetic resonance imaging data from a region containing fibrous structures to be tracked;
• reconstructing the magnetic resonance imaging data into an apparent diffusion tensor map including a three-dimensional arrangement of diffusion tensor voxels;
• selecting a starting voxel;
• beginning at the starting voxel, iteratively following an eigenvector corresponding to a largest eigenvalue from voxel to voxel to construct a three-dimensional fiber representation;
• and
• during the iterative following, locally spatially interpolating voxels, the locally spatially interpolated voxels being weighted combinations of diffusion tensors of nearby voxels.

An imaging apparatus for imaging a subject including anisotropic structures, the apparatus comprising:

• a magnetic resonance imaging scanner configured to acquire a three-dimensional apparent diffusion tensor map of at least a portion of the subject including at least some anisotropic structures;
• a means for processing the apparent diffusion tensor at a voxel to obtain eigenvectors and eigenvalues;
• a means for extracting a three-dimensional fiber representation using the eigenvectors and eigenvalues;
• a means for locally spatially interpolating voxels in at least a selected spatial dimension in a vicinity of the fiber representation during the extracting, the spatial interpolating means including a means for weighting the interpolated voxels by a parameter indicative of a local anisotropy, the interpolating resulting in a three-dimensional fiber representation having a higher resolution than the acquired tensor map;
• and
• a means for producing a human-viewable display of the three-dimensional fiber representation.

An imaging apparatus comprising:

• a magnetic resonance imaging means configured for acquiring a three-dimensional apparent diffusion tensor map of at least a portion of the subject including anisotropic structures;
• a means for processing the apparent diffusion tensor at a voxel to obtain eigenvectors and eigenvalues;
• a means for extracting a three-dimensional fiber representation using the eigenvectors and eigenvalues;
• a means for locally interpolating voxels relative to a selected spatial dimension in a vicinity of the fiber representation during the extracting, the interpolating including weighting the interpolated voxels by one of:
• a direction of a major diffusion tensor eigenvector,
• a functional combination of diffusion tensors, and
• a fractional anisotropy, to produce a three-dimensional fiber representation having a higher resolution than the acquired tensor map.

An imaging apparatus comprising:

• a MR imaging scanner which non-invasively examines a subject and generating an apparent diffusion tensor map including a three-dimensional arrangement of diffusion tensor voxels;
• a processor which tracks fibrous structures in the apparent diffusion tensor map, the processor being programmed to perform the steps of:
• selecting a starting voxel;
• beginning at the starting voxel, iteratively following an eigenvector corresponding to a largest eigenvalue from voxel to voxel to construct a three-dimensional fiber representation;
• and,
• during the iterative following, locally spatially interpolating voxels, the locally spatially interpolated voxels being weighted combinations of diffusion tensors of nearby voxels.