Material decomposition of multi-spectral x-ray projections using neural networks

1. A method of processing x-ray images, the method comprising:

training an artificial neural network to process multi-spectral x-ray projections to determine composition information in terms of equivalent thickness of at least one basis material;

providing a multi-spectral x-ray projection of an object, wherein the multi-spectral x-ray projection of the object contains energy content information describing energy content of the multi-spectral x-ray projection;

processing the multi-spectral x-ray projection with the artificial neural network to determine composition information about the object, wherein the composition information includes one or more basis coefficients mapping the object in terms of equivalent thicknesses of the at least one basis material; and

providing the composition information about the object.

2. The method of claim 1 wherein training the artificial neural network includes:

providing a set of calibration multi-spectral x-ray projections of known thicknesses of the at least one basis material as input to the neural network; and

utilizing a training algorithm to minimize the error between the composition information and the known thicknesses.

3. The method of claim 1 wherein the multi-spectral x-ray projection of the object is a 2-dimensional projection and the basis coefficients comprise a 2-dimensional array of coefficients.

4. The method of claim 1 wherein the multi-spectral x-ray projection of the object is a 3-dimensional computerized tomography dataset and the basis coefficients comprise a 3-dimensional array of coefficients.

5. The method of claim 1 wherein the artificial neural network is trained to determine the composition information in terms of equivalent thicknesses of at least two basis materials, and wherein the composition information of the object includes a set of basis coefficients describing the object in terms of the at least two basis materials.

6. The method of claim 5 wherein one basis material has a high atomic number and one basis material has a low atomic number.

7. The method of claim 1 wherein the artificial neural network is trained to determine the composition information in terms of equivalent thicknesses of at least three basis materials, and wherein at least one of the at least three basis materials is a contrast agent with an absorption K-edge in an energy range of the multi-spectral x-ray projection.

8. The method of claim 1 wherein the multi-spectral x-ray projection is created by one of projection imaging, a tomosynthesis, or computerized tomography.

9. The method of claim 8 wherein the multi-spectral x-ray projection is created by energy-resolved computerized tomography.

10. The method of claim 1 wherein the multi-spectral x-ray projection is created by detecting multiple x-ray projections wherein, for each projection, an x-ray tube voltage and/or a beam filtration is varied.

11. The method of claim 1 wherein the method further includes utilizing the composition information about the object to calculate an array of coefficients mapping the object in terms of atomic number and density.

12. A method of imaging an object comprising:

projecting multi-spectral x-ray beams through an object;

detecting a multi-spectral x-ray projection through the object, wherein the detecting includes detecting the energy content of the multi-spectral x-ray beams; and

processing the multi-spectral x-ray projection with an artificial neural network to determine composition information about the object, wherein the composition information includes a set of basis coefficients mapping the object in terms of equivalent thickness of at least one basis material.

13. The method of claim 12 wherein the at least one basis material includes water.

14. The method of claim 12 wherein the composition information includes a set of basis coefficients mapping the object in terms of equivalent thicknesses of at least two basis materials.

15. The method of claim 14 wherein the basis materials are polymethyl methacrylate (PMMA) and aluminum.

16. The method of claim 12 wherein the artificial neural network generates a set of basis coefficients mapping the object in terms of equivalent thicknesses of at least three basis materials, and wherein at least one of the basis materials is a contrast agent having an absorption K-edge in an energy range of the multi-spectral x-ray projection.

17. The method of claim 12 where the artificial neural network is trained by providing a set of calibration x-ray projections of known thicknesses of the at least one basis material as input to the neural network and utilizing a training algorithm to minimize the error between the composition information and the known thicknesses.

18. The method of claim 12 further comprising generating an object composition output to a user visually depicting a material composition of the object.

19. The method of claim 12 wherein the step of detecting a multi-spectral x-ray projection through the object includes detecting multiple projections of x-ray beams through the object, wherein, for each projection, an x-ray tube voltage and/or a beam filtration is varied.

20. A system for imaging an object, the system comprising:

an imaging device configured to produce a multi-spectral x-ray image of an object; and

an artificial neural network trained to process the multi-spectral x-ray image to produce composition information;

wherein the neural network is trained by providing a set of calibration x-ray projections of known thicknesses of at least one basis material as input to the neural network and utilizing a training algorithm to minimize the error between the composition information and the known thicknesses; and

wherein the composition information includes an array of basis coefficients mapping the object in terms of equivalent thicknesses of the at least one basis material.

21. The system of claim 20 wherein the imaging device includes one or more of a dual kV detector, a photon counting detector, or a dual sandwich detector.