Retrospective selection and various types of image alignment to improve DTI SNR

A method of diffusion tensor magnetic resonance imaging comprising:

• a) collecting a static and a plurality of differently diffusion weighted magnetic resonance image data sets, each data set representing the same spatial region within a subject;
• b) repeating the collecting step a) to generate a plurality of the data sets;
• c) reconstructing the data sets into image representations;
• d) rejecting data sets whose static or like diffusion weighted image representations that fail to exhibit a pre-determined similarity to each other;
• e) aligning like diffusion weighted image representations;
• f) combining non-rejected and aligned static images and non-relected and aligned like diffusion weighted image representations;
• g) in at least a region of interest of the remaining images, calculating an apparent diffusion coefficient tensor;
• h) extracting eigenvalues and eigenvectors from the diffusion coefficient tensor;
• i) tracking an anisotropic structure with the eigenvalues and the eigenvectors;
• j) generating a human readable display of the tracked anisotropic structure.

A method diffusion tensor magnetic resonance imaging comprising:

• a) collecting a static and a plurality of differently diffusion weighted magnetic resonance image data sets, each data set representing the same spatial region within a subject;
• b) repeating the collecting step a) to generate a plurality of static data sets and a Plurality of diffusion weighted data sets corresponding to each diffusion weighting;
• c) reconstructing the data sets into image representations;
• d) adjusting the image representations for better spatial conformity;
• e) rejecting static image representations that fail to exhibit a pre-determined similarity to other static images;
• f) rejecting diffusion weighted image representations that fail to exhibit a predetermined similarity to other diffusion weighted images of like diffusion weighting by comparing the image representations with an average of like diffusion weighted image representations;
• g) combining remaining static images and like diffusion weighted image representations;
• h) in at least a region of interest of the remaining images, calculating apparent diffusion coefficient tensors for each voxel of the region of interest;
• i) extracting eigenvalues and eigenvectors from the diffusion coefficient tensors;
• j) tracking an anisotropic structure with the eigenvalues and the eigenvectors;
• k) generating a human readable display of the tracked anisotropic structure.

A method of diffusion tensor magnetic resonance imaging comprising:

• a) collecting a static and a plurality of differently diffusion weighted magnetic resonance image data sets, each data set representing the same spatial region within a subject;
• b) repeating the collecting step a) to generate a plurality of static data sets and a plurality of diffusion weighted data sets corresponding to each diffusion weighting;
• c) reconstructing the data sets into image representations;
• d) comparing static image representations;
• e) adjusting the static image representations based on the comparing step (d) for greater conformity;
• f) in at least a region of interest of the remaining images, calculating apparent diffusion coefficient tensors for each voxel of the region of interest;
• g) tracking an anisotropic structure based on diffusion coefficient tensors;
• h) generating a human readable display of the tracked anisotropic structure.

A magnetic resonance apparatus comprising:

• a main magnet assembly for generating a substantially uniform main magnetic field through an imaging region;
• a gradient coil assembly for superimposing gradient magnetic fields on the main magnetic field;
• a radio frequency coil assembly for transmitting radio frequency pulses into the imaging region;
• a radio frequency receiver for receiving and demodulating magnetic resonance signals from the imaging region;
• a data memory for storing static and diffusion weighted data sets of a common region of a subject in the imaging region;
• sets into a reconstruction processor that reconstructs the data sets into static and diffusion weighted images;
• a comparator that compares the static images with each other and like diffusion weighted images with each other;
• an image rejection processor that discards images dissimilar to other like diffusion weighted and static images by more than a predetermined threshold;
• an image combining means for combining the remaining static images and like diffusion weighted images into combined static and diffusion weighted images;
• a diffusion analysis processor that (1) calculates a diffusion tensor coefficient from the combined images for each voxel of at least a region of interest, (2) creates eigenvectors from the diffusion tensors, and (3) tracks an anisotropic structure with the eigenvectors;
• a video processor and monitor for generating a human viewable display of the tracked anisotropic structure.

A method of diffusion weighted magnetic resonance imaging comprising:

• generating a plurality of static image representations and for each static image representation generating a plurality of contemporaneous diffusion image representations with each of a plurality of diffusion weightings of the same region of interest in a subject;
• aligning each static image and the contemporaneous diffusion weighted images analogously;
• rejecting images that differ from other static and like diffusion weighted images;
• combining the non-rejected static images and non-rejected diffusion weighted images with like diffusion weighting;
• diffusion analyzing the combined static and diffusion weighted images to image an anisotropic structure.