Method and system which forms an isotropic, high-resolution, three-dimensional diagnostic image of a subject from two-dimensional image data scans
View Patent ↗MRI scans typically have higher resolution within a slice than between slices. To improve the resolution, two MRI scans are taken in different, preferably orthogonal, directions. The scans are registered by maximizing a correlation between their gradients and then fused to form a high-resolution image. Multiple receiving coils can be used. When the images are multispectral, the number of spectral bands is reduced by transformation of the spectral bands in order of image contrast and using the transformed spectral bands with the highest contrast.
1. A method of forming an isotropic, high-resolution, three-dimensional diagnostic image of a subject from two-dimensional image data, the method comprising:
(a) scanning the subject with an imaging scanner device, in a first direction relative to the subject, in order to take image data of a first plurality of slices, the image data of the first plurality of slices being two-dimensional image data and having a low resolution in the first direction and a high resolution in directions orthogonal to the first direction;
(b) scanning the subject with said imaging scanner device in a second direction relative to the subject, which is different from the first direction in order to take image data of a second plurality of slices, the image data of the second plurality of slices being two-dimensional image data having at least one dimension substantially in common with the image data of the first plurality of slices and having a low resolution in the second direction and a high resolution in directions orthogonal to the second direction;
(c) registering the first plurality of slices with the second plurality of slices to define a matrix of isotropic, high-resolution voxels in image space, wherein the matrix has unknown high-resolution voxel values; and
(d) solving for the unknown high-resolution voxel values in the matrix defined in step (c) in accordance with the image data taken in steps (a) and (b) in order to form the isotropic high-resolution three-dimensional diagnostic image in the image space.
2. The method of claim 1 , wherein the second direction is orthogonal to the first direction.
3. The method of claim 1 , wherein step (d) comprises treating the image as a linear combination of at least two low-resolution functions and deriving the unknown high-resolution voxel values from the image data of the first and second pluralities of slices.
4. The method of claim 3 , wherein the functions are derived through an iterative process using the image data of the first and second pluralities of slices as initial assumptions for the functions.
5. The method of claim 1 , wherein step (c) comprises maximizing a correlation based on the directionally scanned image data of the first and second pluralities of slices.
6. The method of claim 5 , wherein the correlation is maximized through a hill-climbing technique in which successively larger values of the correlation are located until a local maximum value of the correlation is reached.
7. The method of claim 6 , wherein the hill-climbing technique is a multiresolution hill-climbing technique.
8. The method of claim 7 , wherein: steps (a) and (b) are performed with an imaging scanner device having an in-plane resolution; and the multiresolution hill-climbing technique is performed with a plurality of resolutions including:
(i) a maximum resolution which is twice the in-plane resolution; and
(ii) a minimum resolution which is one-quarter of the in-plane resolution.
9. The method of claim 6 , wherein the hill-climbing technique is used to determine both a relative displacement and a relative rotation between the first and second pluralities of slices.
10. A system forming an isotropic, high-resolution, three-dimensional diagnostic image of a subject from two-dimensional image data, the system comprising:
scanning means for
(i) scanning the subject in a first direction relative to the subject in order to take image data of a first plurality of slices, the image data of the first plurality of slices being two-dimensional image data and having a low resolution in the first direction and a high resolution in directions orthogonal to the first direction, and
(ii) scanning the subject in a second direction relative to the subject which is different from the first direction in order to take image data of a second plurality of slices, the image data of the second plurality of slices being two-dimensional image data having at least one dimension substantially in common with the image data of the first plurality of slices and having a low resolution in the second direction and a high resolution in directions orthogonal to the second direction; and
computing means for
(i) registering the first plurality of slices with the second plurality of slices in order to define a matrix of isotropic, high-resolution voxels in image space, wherein the matrix has unknown high-resolution voxel values and
(ii) solving for the unknown high-resolution voxel values in the matrix defined by the computing means in accordance with the image data taken in the first and second directions by the scanning means and thereby form the isotropic, high-resolution, three-dimensional, diagnostic image in the image space.
11. The system of claim 10 , wherein the second direction is orthogonal to the first direction.
12. The system of claim 10 , wherein the computing means registers the first and second pluralities of slices by maximizing a directionally scanned correlation based on the image data of the first and second pluralities of slices.
13. The system of claim 12 , wherein the directionally scanned correlation is maximized through a hill-climbing technique in which successively larger values of the correlation are located until a local maximum value of the correlation is reached.
14. The system of claim 13 , wherein the hill-climbing technique is a multiresolution hill-climbing technique.
15. The system of claim 14 , wherein:
the scanning means has an in-plane resolution; and
the multiresolution hill-climbing technique is performed with a plurality of resolutions including:
(i) a maximum resolution which is twice the in-plane resolution; and
(ii) a minimum resolution which is one-quarter of the in-plane resolution.
16. The system of claim 15 , wherein the hill-climbing technique is used to determine both a relative displacement and a relative rotation between the first and second pluralities of slices.
17. The system of claim 10 , wherein the computing means treats the image as a linear combination of at least two low-resolution functions and deriving the unknown high-resolution voxel values from the image data of the first and second pluralities of slices.
18. The system of claim 17 , wherein the functions are derived through an iterative process using the image data of the first and second pluralities of slices as initial assumptions for the functions.