System for iteratively reconstructing computed tomography images through three domains

1. A system for reconstructing CT images comprising:

A. A CT data acquisition system for generating raw data for each detector element at each view;

B. A raw domain processor for generating M sets of serial projection data for each detector at each view; wherein M is a predetermined number of bins of a discretized X-ray spectrum of the X-ray source corresponding to said CT data acquisition system;

C. A projection domain processor for generating M sets of serial image data; wherein said projection domain processor also generates iterated projection data;

D. An image domain processor for generating CT images; wherein said image domain processor also generates iterated image data;

E. A raw synthesizer for generating synthesized raw data; and,

F. A projection synthesizer for generating synthesized projection data,

wherein the input data of said projection domain processor comprises said serial projection data, said iterated image data, and said synthesized projection data; wherein the input data of said image domain processor comprises said serial image data; wherein the input data of said raw synthesizer comprises said iterated projection data; wherein the input data of said projection synthesizer comprises said iterated image data.

2. The system of claim 1 , wherein said raw domain processor also generates air corrected data; wherein each of said M sets of serial projection data is a polynomial function of said air corrected data with K serial coefficients; wherein K is the degree of said polynomial function.

3. The system of claim 2 , wherein said raw domain processor generates said serial coefficients iteratively by updating said serial coefficients with gradient terms at each iteration.

4. The system of claim 3 , wherein said raw domain processor also generates offset corrected data and uses said synthesized raw data, said offset corrected data, said discretized X-ray spectrum, said air corrected data, and said iterated projection data to generate said gradient terms at each iteration.

5. The system of claim 1 , wherein said projection domain processor generates said serial image data iteratively by minimizing a cost function comprising a data fidelity term with said synthetic projection data and said serial projection data.

6. The system of claim 5 , wherein said cost function includes a regularization term on said iterated image data.

7. The system of claim 1 , wherein said image domain processor uses M pre-determined energy dependent weighting coefficients to compute a weighted average of said serial image data as said CT images.

8. The system of claim 1 , wherein said raw synthesizer generates said synthesized raw data, denoted as R s (n,c,v), using said iterated projection data, denoted as P i (n,c,v,E), according to

R

s

⁡

(

n

,

c

,

v

)

=

∑

E

⁢

S

⁡

(

E

)

⁢

e

-

P

i

⁡

(

n

,

c

,

v

,

E

)

⁢

Δ

⁢

⁢

E

,

wherein S(E) is said discretized X-ray spectrum, E=E 0 , E 0 +ΔE, . . . , E 0 +(M− 1 )ΔE is said pre-determined M bins of X-ray energy, E 0 is the first energy bin, ΔE is the energy bin size, n and c are the row index and column index of each detector element respectively, and v is the index of each view.