Enhancing angiograms

1 . A device for enhancing angiograms, the device comprising:

a processor configured to:

obtain at least one angiographic image showing a vascular structure of interest comprising a plurality of vessel branches;

determine at least one branch of interest from the angiographic image;

segment the angiographic image for identifying possible branches of the vascular structure;

select branches based on the identified possible branches and the determined branch of interest;

estimate attenuation values of the selected branches;

subtract at least a predetermined part of the estimated attenuation values from the angiographic image to generate a corrected angiogram; and

provide the corrected angiogram.

2 . The device according to claim 1 , wherein the processor is further configured to:

select branches not of interest based on the identified possible branches and the determined branch of interest; and

estimate attenuation values of the selected branches not of interest.

3 . The device according to claim 2 , wherein, for the subtraction, the data-processor is further configured to:

subtract the complete estimated attenuation values of the branches not of interest from the angiographic image for the generation of the corrected angiogram; or

subtract a predetermined part of the respective estimated attenuation values from the angiographic image for the generation of the corrected angiogram.

4 . The device according to claim 2 , wherein, for the estimation of the attenuation values of the selected branches not of interest, the processor is further configured to assume cylindrically shaped cross-sections for the branches.

5 . The device according to claim 1 , wherein, for the segmentation of the angiographic image for identifying possible branches of the vascular structure, the processor is further configured to:

obtain geometric image acquisition parameters comprising at least one of direction and angulation; and

based on the geometric image acquisition parameters, select expected segments from a lookup table identifying possible segments expected for one or more of the geometric image acquisition parameters.

6 . The device according to claim 5 ,

wherein the vascular structure of interest is a coronary artery tree; and

wherein the lookup table is based on established guidelines listing angiographic projections and optimal visualization of coronary artery segments.

7 . The device according to claim 5 ,

wherein the lookup table comprises different categories of preferences for the to be expected segments of the vascular structure; and

wherein a change in the selection of the categories provides a change of the sensitivity of the segmenting of the angiographic image for identifying the possible branches of the vascular structure.

8 . The device according to claim 1 , wherein the processor is configured to apply a neural network configured to at least one of:

provide semantic segmentation of the angiographic image for identifying the possible branches of the vascular structure by a deep learning procedure; and

provide the semantic segmentation of the angiographic image for identifying the at least one branch of interest by a deep learning procedure.

9 . The device according to claim 1 , wherein the processor is further configured to:

combine the estimated attenuation values as attenuation profile; and at least one of:

i) pre-process the attenuation profile comprising at least one of low pass filtering, bias correction, and temporal subtraction; and

ii) post-process the attenuation profile comprising at least one of sliding median filtering and robust spline fitting.

10 . The device according to claim 1 , further comprising a display configured to display the corrected angiogram.

11 . A medical imaging system for assessment of vascular structures, the system comprising:

an X-ray imaging device with an X-ray source and an X-ray detector; and

the device for enhancing angiograms according to claim 1 ;

wherein the X-ray imaging device is configured to provide the at least one angiographic image.

12 . A method for enhancing angiograms, the method comprising:

providing at least one angiographic image showing a vascular structure of interest comprising a plurality of vessel branches;

determining at least one branch of interest from the angiographic image;

segmenting the angiographic image for identifying possible branches of the vascular structure;

selecting branches based on the identified possible branches and the determined branch of interest;

estimating attenuation values of the selected branches;

subtracting at least a predetermined part of the estimated attenuation values from the angiographic image generating a corrected angiogram; and

providing the corrected angiogram.

13 . The method according to claim 12 , wherein:

in the selecting of branches, branches not of interest are selected based on the identified possible branches and the determined branch of interest; and

in the estimating of attenuation values, attenuation values of the selected branches not of interest are estimated.

14 . The method according to claim 12 , wherein the subtracting further comprises:

subtracting the complete estimated attenuation values of branches not of interest from the angiographic image for the generation of the corrected angiogram; or

subtracting a predetermined part of the respective estimated attenuation values from the angiographic image for the generation of the corrected angiogram.

15 . The method according to claim 12 , wherein the segmenting of the angiographic image for identifying possible branches of the vascular structure further comprises:

obtaining geometric image acquisition parameters comprising at least one of direction and angulation; and

based on the geometric image acquisition parameters, selecting expected segments from a lookup table identifying possible segments expected for one or more of the geometric image acquisition parameters.

16 . The method according to claim 12 , further comprising applying a neural network for at least one of:

providing semantic segmentation of the angiographic image for identifying the possible branches of the vascular structure by a deep learning procedure; and

providing the semantic segmentation of the angiographic image for identifying the at least one branch of interest by the deep learning procedure.

17 . A non-transitory computer-readable storage medium having stored instructions which, when executed by a processor, cause the processor to:

obtain at least one angiographic image showing a vascular structure of interest comprising a plurality of vessel branches;

determine at least one branch of interest from the angiographic image; segment the angiographic image for identifying possible branches of the vascular structure;

select branches based on the identified possible branches and the determined branch of interest;

estimate attenuation values of the selected branches;

subtract at least a predetermined part of the estimated attenuation values from the angiographic image to generate a corrected angiogram; and

provide the corrected angiogram.

18 . The non-transitory computer-readable storage medium of claim 17 , wherein the instructions, when executed by a processor, further cause the processor to, for the subtraction:

subtract the complete estimated attenuation values of branches not of interest from the angiographic image for the generation of the corrected angiogram; or

subtract a predetermined part of the respective estimated attenuation values from the angiographic image for the generation of the corrected angiogram.

19 . The non-transitory computer-readable storage medium of claim 17 , wherein the instructions, when executed by a processor, further cause the processor to:

obtain geometric image acquisition parameters comprising at least one of direction and angulation; and

based on the geometric image acquisition parameters, select expected segments from a lookup table identifying possible segments expected for one or more of the geometric image acquisition parameters.

20 . The non-transitory computer-readable storage medium of claim 17 , wherein the instructions, when executed by a processor, further cause the processor to apply a neural network configured to at least one of:

provide semantic segmentation of the angiographic image for identifying the possible branches of the vascular structure by a deep learning procedure; and

provide the semantic segmentation of the angiographic image for identifying the at least one branch of interest by the deep learning procedure.