Method for enhanced voxel resolution in MRI image
View Patent ↗A method for providing an enhanced 3-D image for a volume imaging apparatus obtains 3-D volume images of an object at a first resolution and forms a 3-D matrix of voxels at a second, higher resolution. Two or more 3-D volume images that have different spatial orientations and that intersect at the voxel are registered. The voxel is mapped to the 3-D matrix. Inpainting fills in gap regions between mapped voxels in the 3-D matrix. A simulated first-resolution data value is generated according to one or more 3-D matrix values. The simulated first-resolution data value is compared with an observed first-resolution value from the 3-D volume images and, until the simulated first resolution data value is within a predetermined threshold value of the observed first-resolution value or until a predetermined number of iterations is reached, the 3-D matrix voxel values are adjusted and the simulated first-resolution data value recomputed.
1. A method for providing an enhanced 3-D image for a volume imaging apparatus comprising:
a) obtaining a plurality of 3-D volume images of an object, the images having voxels at a first resolution, wherein at least two of the images are obtained at different spatial orientations;
b) forming a 3-D matrix of voxels at a second resolution, higher than the first resolution, by iterations of 3-D modeling steps in the following sequence:
(i) registering two or more of the 3-D volume images that have different spatial orientations at the first resolution;
(ii) mapping the voxels at the first resolution onto the 3-D matrix;
(iii) executing a 3-D inpainting routine at the second resolution that iteratively convolves one or more 3-D directional Gaussian functions in order to fill in gap regions between mapped voxels in the 3-D matrix;
(iv) generating, by 3-D modeling, simulated first-resolution data values according to one or more 3-D matrix voxel values at the second resolution; and
(v) comparing the simulated first-resolution data values in each obtained spatial orientation with corresponding observed first-resolution data values from the plurality of 3-D volume images from the same obtained spatial orientation; and, until the simulated first resolution data values are within a predetermined threshold value of the observed first-resolution data value or until a predetermined number of iterations is reached, evaluating a residual error of the simulated and observed data values and minimizing the residual error by combining the residual error with the 3-D matrix voxel values at the second resolution and recomputing the simulated first-resolution data value from (iv); and
c) storing the 3-D matrix voxel values at the second resolution as the enhanced 3-D image.
2. The method of claim 1 wherein at least two of the 3-D volume images are obtained at relatively orthogonal spatial orientations.
3. The method of claim 1 further comprising correcting for inhomogeneity in one or more of the obtained 3-D volume images.
4. The method of claim 1 further comprising applying image standardization to one or more of the obtained 3-D volume images.
5. The method of claim 1 wherein registering two or more of the 3-D volume images comprises using DICOM data related to image orientation.
6. The method of claim 1 wherein registering two or more of the 3-D volume images comprises using intensity-based 3-D affine registration.
7. The method of claim 6 wherein the intensity-based 3-D affine registration accounts for translation, rotation, scale, and shear differences.
8. The method of claim 6 wherein the intensity-based 3-D affine registration is performed in a coarse-to-fine resolution manner.