Method for generation of synthetic mammograms from tomosynthesis data
A method and related apparatus (VS) for synthesizing a projection image (S), in particular for use in mammography. It is proposed to compute a weight function from one image volume (V 1 ) and is then used to implement a weighted forward projection through another image volume block to compute a synthesized projection image (S) across block (V 2 ).
1. An image processing method for forming a projection image, comprising:
based on a first image, which is obtained from a first set of data acquired by a tomographic imaging system, computing a weight-function computed by computing any or a combination of the following measures: an information theoretic quantity, an edge measure quantity, an intensity measure quantity;
based on at least a second image, which is obtained from a second set of data acquired by a tomographic imaging system, forward projecting across the second image, along a desired projection direction, to produce a projection image, wherein the first or the second image is a tomosynthesis volume and the projection image is a synthetic image, wherein the forward projection is weighted as per said weight function, and wherein the weight function is re-sampled to match the sampling of the second image by applying an interpolation scheme.
2. The image processing method of claim 1 , wherein the first image and the second image are derived from at least two different energy channels of a spectral tomographic imaging system with an energy-resolving photon-counting detector unit, and the predefined projection geometry is chosen similar to a 2D mammographic projection geometry.
3. The image processing method of claim 1 , wherein the computing of the weight function comprises local filtering or smoothing.
4. The image processing method of claim 1 , wherein the first image is any one of: a spectral image reconstructed from data detected by a photon-counting detector, a reconstructed spatial frequency component image derived from a multi-scale decomposition.
5. The image processing method of claim 1 , wherein the second image is a reconstructed energy weighted image volume.
6. The image processing method of claim 1 , wherein the weight function is one from a plurality of resolution level specific weight functions, each computed for different resolution levels of the first image.
7. The image processing method of claim 6 , wherein the second image is derived from a multi-scale decomposition, wherein a resolution level of the second image corresponds to a resolution level of the resolution specific weight function.
8. The image processing method of claim 1 , wherein the weight function is computed by using a sequence of kernels with different sizes.
9. The image processing method of claim 6 , wherein the measure, on which the computation of the different weight functions is based, varies with the resolution level.
10. The image processing method of claim 1 , wherein the projection image is a synthetic mammogram.
11. The image processing method of claim 1 , wherein the images are dynamic 4D volumes and wherein the projection image corresponds to a dynamic 2D projection view.
12. An image processing apparatus comprising a data processor equipped to perform the method of claim 1 .
13. A mammography imaging system including the apparatus of claim 12 .
14. A non-transitory computer readable medium having stored thereon a program element for controlling an apparatus, which program element, when executed by a processing unit, is adapted to perform elements comprising:
based on a first image, which is obtained from a first set of data acquired by a tomographic imaging system, computing a weight-function computed by computing any or a combination of the following measures: an information theoretic quantity, an edge measure quantity, an intensity measure quantity;
based on at least a second image, which is obtained from a second set of data acquired by a tomographic imaging system, forward projecting across the second image, along a desired projection direction, to produce a projection image, wherein the first or the second image is a tomosynthesis volume and the projection image is a synthetic image, wherein the forward projection is weighted as per said weight function, and wherein the weight function is re-sampled to match the sampling of the second image by applying an interpolation scheme.