Adaptive anisotropic filtering of projection data for computed tomography
View Patent ↗CT imaging is enhanced by adaptively filtering x-ray attenuation data prior to image reconstruction. Detected x-ray projection data are adaptively and anisotropically filtered based on the locally estimated orientation of structures within the projection data from an object being imaged at a plurality of rotation positions. The detected x-ray data are uniformly low pass filtered to preserve the local mean values in the data, while the high pass filtering is controlled based on the estimated orientations. The resulting filtered data provide projection data with smoothing along the structures while maintaining sharpness along edges. Image noise and noise induced streak artifacts are reduced without increased blurring along edges in the reconstructed images. The enhanced image allows reduced x-ray dose while maintaining image quality.
1. A method for enhancing an image based on projection data comprising the steps of:
a) projecting an x-ray beam through an object of interest to an x-ray detector,
b) collecting x-ray attenuation data (projections) using an x-ray source and a detector at a plurality of positions around the object,
c) estimating the orientation of structures within the obtained projection data, and
d) adaptively filtering the x-ray attenuation data based on the locally estimated orientation of structures, wherein the adaptive filtering is in projection space, and
e) reconstructing an image using the adaptively filtered x-ray attenuation data from step d).
2. The method of claim 1 wherein high frequency x-ray attenuation data are anisotropically filtered according to structures in the measured projection data.
3. The method of claim 2 wherein high frequency components are locally reduced in directions of no apparent structure.
4. The method of claim 3 wherein low frequency x-ray attenuation data are uniformly filtered to maintain a local mean value of the projection data.
5. The method of claim 2 wherein low frequency x-ray attenuation data are uniformly filtered to maintain the local mean value of the projection data.
6. The method of claim 5 , wherein the step of reconstructing an image using adaptively filtered x-ray attenuation data from step d), comprises including uniformly low pass filtered x-ray data in combination with an anisotropically filtered high pass component.
7. The method of claim 5 wherein step c) includes use of a set of differently oriented filters.
8. The method of claim 1 wherein step c) includes use of a set of differently oriented filters.
9. The method of claim 1 wherein in step e) uniformly filtered low frequency x-ray attenuation data are used.
10. The method of claim 1 wherein the projection data are CT projection data.
11. The method of claim 1 wherein the projection data are tomosynthesis projection data.
12. The method of claim 1 , wherein the adaptively filtering the x-ray attenuation data provides a 3D anisotropic filtering based on the estimated orientation of structures within the obtained projection data.
13. The method of claim 12 , wherein the estimated local orientation is represented in the form of a tensor, which is utilized to control the 3D anisotropic filtering.
14. The method of claim 13 , wherein 2D projections from different angles are stacked to form a 3D data volume.