System and method for controlling noise in multi-energy computed tomography images based on spatio-spectral information

1. A method for producing computed tomography (CT) images, the method including steps comprising:

(i) obtaining medical image data sets acquired from a subject at multiple energies of irradiating radiation;

(ii) analyzing the medical image data sets for spatial and spectral features;

(iii) comparing the spatial and spectral features of the medical image data sets across the multiple energies to identify similarities by:

forming spatio-spectral blocks from the medical image data sets, each spatio-spectral block comprising a spatial patch centered on a pixel and stacked across each of the multiple energies;

calculating similarities between different ones of the spatio-spectral blocks; and

(iv) using the similarities, weighting pixels in the medical image data sets to generate images of the subject having reduced noise compared to images of the subject produced from the medical image data sets without weighting.

2. The method of claim 1 wherein step (i) includes reconstructing multi-energy images of the subject and step (ii) includes applying a non-local means (NLM) filter to search for similar pixels in the multi-energy images.

3. The method of claim 2 wherein step (iv) includes performing a weighted average of pixels identified in step (iii) as similar.

4. The method of claim 3 wherein step (iii) includes quantifying a similarity between pixels using a summed square difference (SSD) relative to spatial and spectral features, wherein pixels with lower SSD correspond to higher similarity.

5. The method of claim 4 wherein pixels with higher similarity are weighted heavier in step (iv) and spatial and energy resolutions are maintained.

6. The method of claim 1 wherein step (iii) includes determining differences in image noise levels at distinct energy level channels and adapting weighting according to the noise levels.

7. The method of claim 1 , wherein calculating the similarities between different ones of the spatio-spectral blocks in step (iii) is performed in an image space, a projection space, or both.

8. A computed tomography (CT) imaging system comprising:

at least one x-ray source configured to emit x-rays at one or more energy levels toward an object to be imaged;

at least one detector configured to receive x-rays that are attenuated by the object;

a data acquisition system (DAS) connected to the at least one detector to receive an indication of received x-rays at multiple energy levels;

a computer system coupled to the DAS to receive the indication of the received x-rays at multiple energy levels and programmed to:

(i) obtain medical image data sets acquired from a subject using the multiple energies of irradiating radiation;

(ii) analyze the medical image data sets for spatial and spectral features;

(iii) compare the spatial and spectral features of the medical image data sets across the multiple energy levels to identify similarities by:

forming spatio-spectral blocks from the medical image data sets each spatio-spectral block comprising a spatial patch centered on a pixel and stacked across each of the multiple energies;

calculating similarities between different ones of the spatio-spectral blocks; and

(iv) use the similarities, weighting pixels in the medical image data sets to generate images of the subject having reduced noise compared to images of the subject produced from the medical image data sets without weighting.

9. The CT imaging system of claim 8 wherein step (i) includes reconstructing multi-energy images of the subject and step (ii) includes applying a non-local means (NLM) filter to search for similar pixels in the multi-energy images.

10. The CT imaging system of claim 9 wherein step (iv) includes performing a weighted average of pixels identified in step (iii) as similar.

11. The CT imaging system of claim 10 wherein step (iii) includes quantifying a similarity between pixels using a summed square difference (SSD) relative to spatial and spectral features, wherein pixels with lower SSD correspond to higher similarity.

12. The CT imaging system of claim 11 wherein pixels with higher similarity are weighted heavier in step (iv).

13. The CT imaging system of claim 8 wherein step (iii) includes determining differences in image noise levels at distinct energy level channels and adapting weighting according to the noise levels.

14. The CT imaging system of claim 8 , wherein the computer system is programmed to calculate the similarities between different ones of the spatio-spectral blocks in an image space, a projection space, or both.