Automatic system calibration method of X-ray CT
First Claim
Patent Images
1. A method of reconstructing a computed tomography (CT) image, the method comprising:
- i) obtaining an initial CT image;
ii) performing a reconstruction algorithm on the initial CT image to obtain a reconstructed image of the initial CT image;
iii) using the reconstructed image to adjust one or more parameters associated with the image comprising using a Locally Linear Embedding (LLE) method on each of the one or more parameters;
iv) using the adjusted one or more parameters to perform the reconstruction algorithm on the reconstructed image to obtain an updated reconstructed image; and
v) repeating steps iii)-iv), using the most-recently updated reconstructed image in each repeated step iii), and updating the same one or more parameters in each repeated step iii) until a threshold value of a predetermined characteristic is met;
wherein the LLE method comprises;
with the K nearest vectors, representing the original data vector linearly with its neighboring vectors;
1 Assignment
0 Petitions
Accused Products
Abstract
Systems and methods for geometric calibration and image reconstruction in computed tomography (CT) scanning using iterative reconstruction algorithms are provided. An iterative reconstruction algorithm can be used to reconstruct an improved image, and then the improved image can be used to adjust inaccurate parameters by using a Locally Linear Embedding (LLE) method. Adjusted parameters can then be used to reconstruct new images, which can then be used to further adjust the parameters. The steps of this iterative process can be repeated until a quality threshold is met.
-
Citations
30 Claims
-
1. A method of reconstructing a computed tomography (CT) image, the method comprising:
-
i) obtaining an initial CT image; ii) performing a reconstruction algorithm on the initial CT image to obtain a reconstructed image of the initial CT image; iii) using the reconstructed image to adjust one or more parameters associated with the image comprising using a Locally Linear Embedding (LLE) method on each of the one or more parameters; iv) using the adjusted one or more parameters to perform the reconstruction algorithm on the reconstructed image to obtain an updated reconstructed image; and v) repeating steps iii)-iv), using the most-recently updated reconstructed image in each repeated step iii), and updating the same one or more parameters in each repeated step iii) until a threshold value of a predetermined characteristic is met; wherein the LLE method comprises; with the K nearest vectors, representing the original data vector linearly with its neighboring vectors; - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
-
19. A method of reconstructing a computed tomography (CT) image, the method comprising:
-
i) obtaining an initial CT image; ii) performing a reconstruction algorithm on the initial CT image to obtain a reconstructed image of the initial CT image; iii) using the reconstructed image to adjust one or more parameters associated with the image comprising using a Locally Linear Embedding (LLE) method on each of the one or more parameters; iv) using the adjusted one or more parameters to perform the reconstruction algorithm on the reconstructed image to obtain an updated reconstructed image; and v) repeating steps iii)-iv), using the most-recently updated reconstructed image in each repeated step iii), and updating the same one or more parameters in each repeated step iii) until a threshold value of a predetermined characteristic is met; wherein the LLE method comprises; calculating, with the weight coefficients W=(wik), the global internal coordinate Y=(yi) by solving the equation; - View Dependent Claims (20, 21, 22, 23, 24)
-
-
25. A method of reconstructing a computed tomography (CT) image, the method comprising:
-
i) obtaining an initial CT image; ii) performing a reconstruction algorithm on the initial CT image to obtain a reconstructed image of the initial CT image; iii) using the reconstructed image to adjust one or more parameters associated with the image comprising using a Locally Linear Embedding (LLE) method on each of the one or more parameters; iv) using the adjusted one or more parameters to perform the reconstruction algorithm on the reconstructed image to obtain an updated reconstructed image; and v) repeating steps iii)-iv), using the most-recently updated reconstructed image in each repeated step iii), and updating the same one or more parameters in each repeated step iii) until a threshold value of a predetermined characteristic is met; wherein the LLE method comprises solving the following linear system of equations;
Au=bi,where u=(u1, u2, . . . , uJ) is an image represented as a J dimensional vector, J is the number of pixels, bi=(b1, b2, . . . , bL) is data, L is the number of vector elements, and A=(ajk) is a projection matrix related to the one or more (geometric) parameters, wherein performing the reconstruction algorithm comprises calculating a projection matrix, which is affected by the one or more (geometric) parameters;
A=A(P),where P is an estimated parameter vector, which includes the one or more parameters, wherein the projections of the projection matrix are calculated using a distance-driven method, wherein using the reconstructed image to adjust one or more parameters comprises minimizing the mean squared error between the projection data and re-projected projection data, which can be formulated as;
P=arg min∥
bi−
{tilde over (b)}ij∥
22s.t. A(P)u=bi,where bi is the projection vector obtained from the measurement along different projection views, {tilde over (b)}ij is the corresponding re-projected projection vector from a reconstructed image with sampled parameters, and P is the updated vector of parameters, and wherein the re-projected projection vector can be calculated within a densely sampled parametric range;
{tilde over (P)}j=(pj1,pj2, . . . ,pjn).- View Dependent Claims (26, 27, 28, 29, 30)
-
-
27. The method according to claim 25, wherein the LLE method further comprises:
finding the K nearest neighbors of a data vector bi in its dataset vectors {tilde over (b)}ij according to the Euclidean distance;
dij=∥
bi−
{tilde over (b)}ij∥
22.
-
28. The method according to claim 25, wherein the LLE method further comprises:
with the K nearest vectors, representing the original data vector linearly with its neighboring vectors;
-
29. The method according to claim 25, wherein the LLE method further comprises, calculating, with the weight coefficients W=(wik), the global internal coordinate Y=(yi) by solving the equation:
-
30. The method according to claim 29, wherein the solution of the equation that can be solved to calculate Y=(yi) is given by the bottom d+1 eigenvectors of the generalized eigenvalue problem:
-
MY=λ
Y,where λ
is the eigenvalue, and M=(I−
W)T(I−
W).
-
Specification