Spectral CT Material Decomposition in the Presence of Poisson Noise: A Kullback-Leibler Approach

- A new method of material decomposition in the projection domain for spectral CT is proposed.
- The decomposition is made by minimizing a regularized cost function with a Gauss-Newton algorithm.
- The Kullback-Leibler distance is introduced as data fidelity term since it corresponds to the noise in the data.
- This method is compared with the more widely used weighted least squares term on a numerical phantom of a mouse.
- Results show that at a high photonic noise the Kullback-Leibler distance is outperforming the weighted least squares.