How to process computed tomography (CT) data for filtered back projection (FBP)

Abstract

The present invention relates to a method for processing computed tomography (CT) data to suppress image cone-beam artifacts (CBAs) in CT images reconstructed from the CT data. A frequency decomposition method is used for reconstruction. However, straightforward use of this method can lead to an undesirable increase in residual low-frequency noise remaining in the basis images after applying image-domain denoising methods. This residual noise propagates fairly linearly in the spectral results. To avoid this noise increase, the method presented herein selectively yet effectively uses a frequency-domain denoising method. Thus, in a first aspect of the present invention, a method for processing computed tomography (CT) data to suppress image cone-beam artifacts (CBAs) in CT images to be reconstructed from the CT data is provided. The method includes steps S1, S2, S3, and S4, respectively, of acquiring CT data generated during a CT scan of a patient, S4, S5, S6, S7, S8, S9, S10, S11, S12, S13, S14, S15, S16, S17, S18, S19, S20, S21, S22, S23, S24, S25, S26, S27, S28, S29, S30, S31, S32, S33, S34, S35, S36, S37, S38, S39, S40, S41, S42, S43, S44, S45, S46, S47, S48, S49, S50, S51, S52, S53, S54, S55, S56, S57, S58, S59, S60, S61, S62, S63, S64, S65, S66, S67, S68, S69, S70, S71, S72, S73, S74, S75, S76, S77

Description

[Technical field]

[0001] The present invention relates to a method for processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from the CT data, a system for processing CT data to suppress image cone beam artifacts (CBA) in a CT image, a computer program element and a computer readable medium. [Background technology]

[0002] The introduction of cone-beam computed tomography (CT) systems offers several advantages: compared to single-slice CT systems, the time for data acquisition can be reduced, the X-ray tube power is used more efficiently, which simplifies its thermal management, and z-resolution no longer has to be compromised to achieve acceptable scan times. However, the main challenge of using cone-beam CT systems is finding a suitable reconstruction algorithm. A large number of publications have addressed the problem of reconstruction of cone-beam projections acquired during a helical path of the source-detector system.

[0003] For example, International Patent Application Publication No. WO2019 / 096600 proposes a single CT backprojection with one geometry calculation per voxel for multiple different types of projection data, thereby reducing the cost of the reconstruction system and improving processing time.

[0004] However, so far, exact or semi-exact algorithms exist only for PI and 3-PI acquisitions. However, from a practical point of view, approximation methods still have advantages for intermediate cone angles, especially when any amount of redundant data can be used. Using redundant data is known to significantly reduce sampling artifacts in the z-direction as well as reduce sensitivity to patient motion.

[0005] Moreover, most approximation methods allow the use of the entire detector area with any pitch value and therefore the total dose delivered to the patient, supporting the possibility of using X-ray offset techniques for resolution enhancement. Another advantage may result from a reduction in the computational effort. This is why the inventors of the present invention invented the frequency split method together with their co-inventors, which was published in 2004 in the paper “The frequency split method for helical cone beam reconstruction”, Gilad Shechter, Thomas Koehler, Ami Altman and Roland Proksa, Medical Physics 31, No. 8,2004. Since then, the frequency split (FS) method has gained popularity among CT practitioners and is a well-known technique for improving CT imaging.

[0006] The frequency-splitting (FS) method provides a practical method for suppressing cone-beam artifacts (CBA) in CT images reconstructed from non-gated helical scans. However, the use of this method increases the low-frequency noise content of the images. This noise increase becomes intractable in spectral CT images reconstructed after projection-domain decomposition (see the papers “Energy-selective reconstructions in X-ray computerized tomography”, Robert E. Alvarez and Albert Macovski, Phys. Med. Biol.21,p-733,1796 and “K-edge imaging in x-ray computed tomography using multi-bin photon counting detectors”, Ewald Roessl and Roland Proksa, Physics in medicine and biology,52,2007). This is because these images are known to be affected by strong anti-correlated noise, as described in the paper “Quantization of Liver Tissue in Dual kVp Computed Tomography using Linear Discriminant Analysis”, J. Eric Tkaczyk et al, Proceedings of SPIE 7258, 2009.

[0007] This noise increase is even more pronounced in the presence of advanced image-domain denoising methods that can wipe out almost all high frequency noise components from an image.

[0008] Thus, the present inventors have recognized a need to further reduce the problem of CBA in wide cone angle spectral CT. Summary of the Invention [Problem to be solved by the invention]

[0009] Therefore, there may be a need for practical methods and systems for suppressing CBA in wide cone angle spectral CT. [Means for solving the problem]

[0010] The object of the present invention is solved by the subject matter of the independent claims, further embodiments are contained in the dependent claims. It is noted that the below described aspects of the invention equally apply to methods, systems, program elements and computer readable media for processing computed tomography (CT) data so as to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from said CT data.

[0011] According to a first aspect of the present invention, there is provided a method of processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from the CT data, the method comprising: - acquiring CT data generated during a CT scan of a patient (step S1); - decomposing the acquired CT data in the projection domain to generate a plurality of decomposed sinograms (step S2); - distributing noise and / or inconsistencies (mismatches) non-uniformly between the decomposed sinograms (step S3); and - applying a filtered backprojection (FBP) to the decomposed sinogram to which the non-uniform variance was applied in step S3, thereby reconstructing one or more basis images using a frequency division (FS) method (step S4); has.

[0012] In other words, a practical method is proposed here for suppressing image CBA in CT. Contrary to the original FS method applied today by those skilled in the art based on the aforementioned paper introducing the FS method, the method of the present invention suppresses low-frequency noise in these CT images, as will be explained in more detail below.

[0013] It should be noted that the method of the present invention is preferably performed on a computer normally used to reconstruct CT images, such as a reconstruction server.

[0014] Furthermore, potential applications of the methods presented here are in a variety of wide cone-angle spectral CT systems, such as dual-layer detectors, dual-energy systems such as fast kVp switching or dual sources, as well as all non-gated helical scans in photon counting systems.

[0015] Furthermore, in the context of the present invention, a "CT image to be reconstructed" is a "base image" or "basis image," as will be explained in more detail below. A basis image reconstructed from the originally acquired CT data by using the method of the present invention can be seen, for example, in FIG. 2, where it is designated by reference character 206.

[0016] Concerning the step of decomposing the acquired CT data in the projection domain to generate a number of decomposed sinograms, it is noted that the skilled person is familiar with this method step, since this has been frequently done in the field of CT reconstruction for many years, and details on this are disclosed, for example, in the above-mentioned publications and will also be explained below in the context of specific embodiments of the invention.

[0017] The result of the above described method steps S1-S3 are sinograms in which noise and / or inconsistencies that may lead to image cone beam artifacts are non-uniformly distributed. Preferably, said noise and / or inconsistencies are distributed as non-uniformly as possible, i.e. with a maximal non-uniform distribution of said noise and / or inconsistencies that may lead to image cone beam artifacts, as will be explained in more detail below.

[0018] Furthermore, the sinogram thus provided according to the invention is then used as data to which the well-known frequency partitioning (FS) method is applied, i.e. after steps S1 to S3 have been applied. Thus, steps S1 to S3 already provide improved data, which can then be used and processed in the reconstruction of one or more base images by applying a filtered backprojection (FBP) to the decomposed sinogram to which the non-uniform variance has been applied in step S3. In this filtered backprojection (FBP), the frequency partitioning (FS) method can be used.

[0019] In the context of the present invention, the "frequency split (FS) method" should be understood as a well-known method, as commonly performed by those skilled in the art in CT reconstruction, and described in detail in the article "The frequency split method for helical cone beam reconstruction", Gilad Shechter, Thomas Koehler, Ami Altman and Roland Proksa, Medical Physics 31, No. 8, 2004. When the sinograms / data obtained from steps S1-S3 of the method of the present invention are used in said method, improved CT images can be reconstructed, as will be explained in more detail below.

[0020] The frequency division (FS) method used in the context of the present invention can be understood as a filtered backprojection (FBP) in which at least some low frequencies of a base image are reconstructed without using all data of the supplied CT data, while higher frequencies of the base image are reconstructed using more, preferably all, of the CT data. Thus, the method presented herein can be considered as a method of processing a CT for filtered backprojection.

[0021] Thus, the method described herein above using steps S1-S3 provides a practical method for suppressing CBA in wide-cone-angle spectral CT. Contrary to the original FS method, the method presented here helps to suppress low-frequency noise in these images, as will become apparent from this disclosure.

[0022] In one exemplary embodiment, a unitary basis transformation can be applied to the different sinograms decomposed in the projection domain first. In this way, the noise (and / or mismatch leading to image CBA) can be distributed as non-uniformly as possible between these sinograms. Then, each sinogram can be reconstructed using the FS method, thereby reducing the aggressiveness of the FS method for sinograms that are known to be noisier and / or less contaminated by cone angle mismatch than others.

[0023] In other words, this embodiment suggests selective use in the sense that the FS method is selectively applied differently to different sinograms. This selective use of the FS method can also be applied to dual source spectrum systems.

[0024] Of course, such non-uniform distribution required in step S3 can be achieved in a variety of different ways, and some exemplary embodiments are described in more detail below with specific mathematically formulated embodiments.

[0025] Generally, the presented method solves the problems of the prior art for reconstructing CT images, where a frequency decomposition method is applied, which brings about the drawback of an increase in the low-frequency noise components of the image. This increase in noise is even more troublesome in the spectral CT images reconstructed after projection domain decomposition. These disadvantages are overcome by the provision of sinograms according to the above-mentioned steps S1-S3, since after processing them with the FS method, the increase in the low-frequency noise components in the images to be reconstructed is avoided or at least reduced.

[0026] The inventors of the present invention have also been able to demonstrate the advantages provided by the method presented herein in the tests shown in Figures 5(a)-(c). The sagittal brain images shown in Figures 5(a)-(c) are reconstructed from a simulated dual energy helical scan with a beam aperture of 8 cm at the center of rotation, a pitch factor of 0.3, and 340 mA. As shown in Figure 5(a), if the FS method is not applied to the reconstruction, an unacceptable band of CBA appears around the cerebellum. In Figure 5(b), the inventors have used the FS method to reconstruct the original basis element L α=1,2 was reconstructed using a low-frequency filter (LPP) that rolls off to zero at 4 line pairs per cm. Here, the suppression of the CBA is sacrificed by an increase in residual low-frequency noise, which impairs the distinction between gray and white matter and corrupts the image more than the original CBA. The residual low-frequency noise (in xy) also causes blurring of the image around the cerebellum. In FIG. 5(c), the inventors have non-uniformly distributed noise and / or inconsistencies among the decomposed sinograms, which would lead to image cone-beam artifacts. As an example, U βα The FS method configuration was used to obtain the new basis, L β=2 While reconstructing the sinogram of L β=1 The sinogram was reconstructed without using the FS method. With the selective use of FS in the test of FIG. 5(c), no increase in residual low-frequency noise is observed with respect to FIG. 5(a). At the same time, a clear suppression of CBA is still achieved. Thus, as confirmed by these tests, the method of processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from the CT data provides improved reconstruction results compared to the prior art.

[0027] An industrial dual-energy CT scanner consists of collecting signals at two energy bins. In a photon-counting CT system, n binThe number of bins, denoted by n, can be increased, for example, to 5. By projected domain decomposition, n m n equivalent paths are numerically decomposed, where m ≦n bin This can be done by inverting the analytical equation or maximizing the likelihood function. We call these equivalent paths, A α (α=1,2…n m For convenience, the above equivalent path is expressed as follows: α where μ α is the A at a fixed energy (e.g., 70 keV) α is the damping coefficient of the material corresponding to

number

[0028] Next, I α The basis images, denoted by, are reconstructed from the sinograms of these line integrals by the filtered back projection (FBP) method.

[0029] Thus, according to an exemplary embodiment, the projected domain decomposition is m is used to numerically decompose the equivalent paths.

[0030] According to another exemplary embodiment of the present invention, the noise and / or inconsistencies that would lead to image cone beam artifacts are distributed as non-uniformly as possible among the decomposed sinograms.

[0031] According to the present invention, the method further comprises a step (step S4) of applying a filtered back projection (FBP) to the decomposed sinogram to which the non-uniform variance has been applied in step S3, thereby reconstructing one or more base images by using a frequency division (FS) method.

[0032] In other words, one or more base images are reconstructed by applying a filtered back projection (FBP) such that at least some low frequencies of the base images are reconstructed without using all of the acquired CT data, while higher frequencies of the images are reconstructed using more, preferably all, of the CT data. Thus, according to this embodiment, low frequencies of the images are reconstructed using only a small portion of the projections, enhancing the 2D consistency between them.

[0033] The FS method is an approximation for utilizing redundant data in helical cone-beam CT. It is based on the observation that the original WEDGE method provides good image quality when slightly more than 180° of data is used for backprojection, but significant low-frequency artifacts appear when a large amount of redundant data is used. This degradation is compensated for by a frequency decomposition method. The low-frequency part of the image can be reconstructed using slightly more than 180° of data, while the high-frequency part is reconstructed using more, preferably all, of the data. The FS method causes little degradation in the signal-to-noise ratio of the reconstructed image and is known to operate robustly in the presence of motion.

[0034] According to another exemplary embodiment of the present invention, the frequency division (FS) method is characterized in that at least some low frequencies of the base image are reconstructed without using all of the acquired CT data, while higher frequencies of the base image are reconstructed using more, preferably all, of the CT data.

[0035] According to another exemplary embodiment of the present invention, the frequency division (FS) method is selectively applied to the plurality of sinograms.

[0036] Thus, for at least two different sinograms, parameters of the FS method are varied for the different sinograms, as will be explained in detail below.

[0037] According to another exemplary embodiment of the present invention, the aggressiveness of the Frequency Division (FS) method is varied for different sinograms.

[0038] As will be appreciated by those skilled in the art, the FS method (in contrast to previous reconstruction methods) applies at least two reconstructions to the acquired CT data, namely one reconstruction for high frequencies where all or much of the acquired CT data is used, and one reconstruction for low frequencies where the entire CT data is not used (because here redundant data is ignored). Thus, the "aggressiveness" of the FS method, in the context of the present invention, should be understood as a parameter / indicator that describes or indicates how much redundant data is ignored by applying the FS method.

[0039] In the first example, the boundary that can be set by the user or the system and that separates high and low frequencies to which different reconstructions are applied is a way to adjust the "aggressiveness" of the FS method. If the user or the system sets, for example, a high boundary value and thus classifies many frequencies as "low frequencies" where a certain amount of redundant data is ignored during reconstruction, this will be understood by those skilled in the art as an aggressive way of applying the FS method. Thus, a fairly low boundary value that separates high and low frequencies of the provided CT data will be understood by those skilled in the art as applying a fairly low "aggressiveness" value to the FS method.

[0040] In a second example, which can of course be combined with the first example above, the user / system can set the "aggressiveness" of the FS method using an angular parameter that limits the use of the angular spectrum of the provided CT data in the low-frequency reconstruction. Assuming that the provided CT data includes 360° of a patient scan, the angular parameter defines, for example, that only 180°, 185°, 190°, 200°, or 240° are used for the low-frequency reconstruction. Thus, this angular parameter, together with the previously mentioned boundary values, also determines how aggressively the FS method is applied.

[0041] According to another exemplary embodiment of the present invention, the aggressiveness of the Frequency Division (FS) method is reduced for at least one sinogram, preferably all sinograms, that have significant low frequency noise.

[0042] The user and / or the system executing the method may determine, for example by user input, a low-frequency noise level above which it is considered to be significant low-frequency noise. This configuration may be considered as part of an exemplary embodiment of the method. In this case, the method may automatically select sinograms that meet this criterion from sinograms in which the provided CT data is decomposed in the projection domain. This will be explained in more detail later in the context of a detailed embodiment.

[0043] According to another exemplary embodiment of the present invention, the aggressiveness of the Frequency Division (FS) method is reduced for at least one sinogram, preferably all sinograms, that have a low level of cone angle mismatch.

[0044] A user and / or a system executing the method can determine, for example by user input, a level of cone angle mismatch below which a low level of cone angle mismatch is considered. This configuration can be seen as part of an exemplary embodiment of the method. In this case, the method can automatically select sinograms that satisfy this criterion from sinograms in which the acquired CT data is decomposed in the projection domain. This will be explained in more detail later in the context of a detailed embodiment.

[0045] According to another exemplary embodiment of the present invention, the variation in aggressiveness of the FS method for different sinograms is controlled by varying the cutoff and / or shape of the low frequency filter used in the FS method.

[0046] This variation can be performed automatically by a computer, or by a user by individually controlling / adjusting the cutoff and / or shape of the low frequency filters used in the frequency division (FS) method for each sinogram or group of sinograms, as will be explained in more detail later in the context of the detailed embodiments.

[0047] According to another exemplary embodiment of the present invention, the variation in aggressiveness of the frequency division (FS) method for different sinograms is controlled by modifying the backprojection (BP) weighting configuration used to generate the low frequency images.

[0048] This modification of the backprojection (BP) weighting configuration can be performed automatically by a computer or by a user by individually controlling / adjusting the backprojection (BP) weighting configuration for each sinogram or group of sinograms, as will be explained in more detail later in the context of a detailed embodiment.

[0049] According to another exemplary embodiment of the present invention, the CT data is derived from a non-gated helical scan of the patient's body. Thus, the method presented may also include a step of performing a non-gated helical scan of the patient's body using a CT imaging device and generating the formed image data as "CT data" used in step S1 of the method presented herein.

[0050] According to a second aspect of the present invention, there is provided a system for processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from said CT data, the system comprising a computing unit comprising: - receiving CT data generated during a CT scan of a patient; - decomposing the received CT data in a projection domain to generate a plurality of decomposed sinograms; - distributing noise and / or inconsistencies that may lead to image cone beam artifacts unevenly among the decomposed sinograms; and - applying a filtered backprojection (FBP) to the decomposed sinogram to which the non-uniform variance has been applied, thereby reconstructing one or more base images using a frequency division (FS) method; It is configured as follows.

[0051] The system may be embodied as a computer typically used for the reconstruction of CT images, for example a reconstruction server, but it may also be constituted by a CT imaging device, in which the processing of CT data described in steps S1 to S3 is carried out.

[0052] Some or all elements of the system may be configured in hardware, such as in a suitably programmed FPGA (Field Programmable Gate Array) or as a hardwired IC chip. Some elements of the system may be configured in software, hardware, or both. One or more features disclosed herein may be configured or embodied as / by a circuit encoded in a computer readable medium, and / or a combination thereof. A circuit may include discrete and / or integrated circuits, application specific integrated circuits (ASICs), systems on chips (SOCs), combinations thereof, machines, computer systems, processors and memories, computer programs.

[0053] According to another exemplary embodiment of the present invention, the calculation unit is configured to apply a filtered back projection (FBP) to the decomposed sinograms to which the non-uniform variance of step S3 has been applied, thereby reconstructing a base image by using a frequency partitioning (FS) method (step S3), where the frequency partitioning (FS) method is preferably selectively applied to the multiple sinograms.

[0054] According to another aspect of the invention, there is provided a CT system for processing computed tomography (CT) data to suppress image cone beam artifacts (CBAs) in a CT image to be reconstructed from the CT data, the system comprising a CT imaging system and the system described above.

[0055] According to another aspect of the invention there is provided a computer program element which, when executed by at least one processing unit, causes the processing unit to: - acquiring CT data generated during a CT scan of a patient (step S1); - decomposing the acquired CT data in the projection domain to generate a plurality of decomposed sinograms (step S2); - distributing noise and / or inconsistencies that lead to image cone beam artifacts non-uniformly among the decomposed sinograms (step S3); and - applying a filtered backprojection (FBP) to the decomposed sinogram to which the non-uniform variance was applied in step S3, thereby reconstructing one or more basis images using a frequency decomposition (FS) method (step S4); The present invention is adapted to perform a method comprising the steps of:

[0056] Thus, said computer program elements may be stored on a computer unit, which may also be part of an embodiment of the present invention. This computer (computing) unit may be adapted to perform or direct the execution of the steps of the method described above. Furthermore, said computing unit may be adapted to operate the components of the apparatus described above. The computing unit may be adapted to operate automatically and / or to execute user commands. The computer program may be loaded into the working memory of a data processor. In this way, said data processor may be equipped to perform the method of the present invention.

[0057] This exemplary embodiment of the invention covers both computer programs that use the invention from the beginning, and computer programs that, through updates, turn existing programs into programs that use the invention.

[0058] Moreover, the computer program element may provide all the steps required to carry out the procedures of the exemplary embodiments of the methods described above.

[0059] According to another aspect of the invention, there is provided a computer readable medium having stored thereon the program elements described above.

[0060] According to another exemplary embodiment of the present invention, a computer readable medium, such as a CD-ROM, is presented, which stores the computer program elements described by the previous paragraphs.

[0061] The computer program may be stored and / or distributed on a suitable medium (particularly, but not necessarily, a non-transitory medium), such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless communication systems.

[0062] However, the computer program may also be provided over a network, such as the World Wide Web, and may be downloaded into the working memory of a data processor from such a network. According to another exemplary embodiment of the invention, a medium for making available a computer program element for downloading is also provided, the computer program element being configured to perform a method according to one of the aforementioned embodiments of the invention.

[0063] According to an exemplary embodiment of the invention, the acquired CT data set is a wide-cone angle spectral CT data set.

[0064] It should be noted that the embodiments of the present invention are described with respect to different subject matters. In particular, some embodiments are described with respect to method type claims, while other embodiments are described with respect to apparatus type claims. However, a person skilled in the art will understand from the above and following description that, unless otherwise stated, any combination of features belonging to a certain type of subject matter, as well as any combination between features related to different subject matters, is also considered to be disclosed by the present application. However, all features can be combined to provide a synergistic effect that goes beyond the mere collection of such features. [Brief description of the drawings]

[0065] [Figure 1] FIG. 1 shows a flowchart of a method for processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from the CT data, according to an exemplary embodiment of the invention. [Diagram 2] FIG. 2 shows a flow diagram of a method for processing CT data to suppress CBA in a CT image reconstructed from the CT data, according to an exemplary embodiment of the present invention. [Diagram 3] FIG. 3 shows a typical scatter plot of a noise component that may be used in an exemplary embodiment of the present invention. [Figure 4] FIG. 4 illustrates a CT system for processing CT data to suppress image cone beam artifacts (CBAs) in CT images, according to an exemplary embodiment of the present invention. [Diagram 5] 5(a)-(c) show the results of an exemplary embodiment of the method of the present invention versus the prior art for reconstructing CT images. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0066] FIG. 1 shows a flow chart of a method for processing CT data to suppress image cone beam artifacts (CBA) in a CT image reconstructed from the CT data. An FS method is used for the reconstruction. However, straightforward use of this method may lead to an undesirable increase in the residual low-frequency noise remaining in the basis image after applying the image domain denoising method. In this case, this residual noise propagates fairly linearly in the spectral results. To avoid this noise increase, the method shown in FIG. 1 selectively, yet effectively, uses the FS method, which is described in more detail below in combination with the other FIGS. 2 to 4.

[0067] The method of Fig. 1 comprises receiving / providing / acquiring CT data generated during a CT scan of a patient in step S1. As a further step, in step S2, the acquired CT data is decomposed in the projection domain resulting in a number of decomposed sinograms. Furthermore, in step S3, a non-uniform distribution among the decomposed sinograms is performed with respect to noise and / or inconsistencies that may lead to image cone beam artifacts. After step S3, the resulting data, i.e. the noise / inconsistency distributed sinograms, are used in step S4. In this step, a reconstruction of one or more base images is performed by applying a filtered back projection (FBP) to the decomposed sinograms to which a non-uniform distribution was applied in step S3, thereby using a frequency partitioning (FS) method in step S4.

[0068] In other words, a practical method is proposed here for suppressing image CBA in wide-cone-angle spectral CT. Contrary to the original FS method, the inventive method helps to limit low-frequency noise in these images, as will be explained in more detail below, especially in the context of the embodiment shown in FIG.

[0069] It should be noted that the method of Fig. 1 is preferably executed on a computer, for example a reconstruction server typically used to reconstruct CT images. Furthermore, possible applications of the method presented here are various wide cone angle spectral CT systems, such as dual-layer detectors, fast kVp switching, dual-source, dual-energy systems, and all non-gated helical scan in photocounting systems.

[0070] The result of the above described method steps S1-S3 will be a sinogram in which noise and / or inconsistencies that may lead to image cone beam artifacts are non-uniformly, i.e. unevenly distributed. Preferably, the noise and / or inconsistencies are distributed as non-uniformly as possible, i.e. such that the non-uniform distribution of said noise and / or inconsistencies that may lead to image cone beam artifacts is maximized. This will be explained in more detail in the context of FIG. 2.

[0071] Again, the frequency division (FS) method used in the context of FIG. 1 can be understood as filtered back projection (FBP) in which at least some of the low frequencies of a base image are reconstructed without using all of the acquired CT data, while higher frequencies of the base image are reconstructed using more data, preferably all of the data, of the CT data, as described in the context of FIGS. 2 and 3.

[0072] Fig. 2 shows a flow diagram of a method for processing CT data to suppress CBA of a CT-based image reconstructed from the CT data according to an exemplary embodiment of the present invention. As will be apparent from the following description, Fig. 2 shows a practical method for suppressing image CBA in wide-cone-angle spectrum CT, since the method helps to suppress low-frequency noise in these images. The embodiment shown in Fig. 2 can be applied to all non-gated helical scans in various wide-cone-angle spectrum CT systems, such as dual-energy systems, fast kVp switching methods or dual sources, as well as photo-counting systems.

[0073] In the method 200 of FIG. 2, a pre-calculated basis 201 for the transformation is used to distribute the noise and / or inconsistencies that lead to image cone beam artifacts non-uniformly among the decomposed sinograms 202. This results in a change of basis, as explained in the detailed exemplary embodiment that follows. Of course, there are various ways to pre-calculate the basis transformation 201. For example, a priori knowledge can be used that is intended to distribute the noise in a non-uniform manner among different sinograms. The idea here is to reconstruct each sinogram into an image using the FS method selectively for each basic element, i.e., to reduce the aggressiveness of FS sinograms that are known to have more low-frequency noise than others. As another example, different a priori knowledge can be used with the aim of distributing the inconsistencies due to wide cone angles in a non-uniform manner among these different sinograms. In this case, sinograms should be reconstructed with reduced FS aggressiveness for those that are known to contain less cone angle inconsistencies than others. Either of these two approaches, or a combination of both, can be used to obtain a satisfactory reduction in CBA while avoiding the increase of residual low-frequency noise in the spectral results.

[0074] Fig. 2 also shows several decomposed sinograms 204a, 205a and 206a where basis modification is applied to distribute noise and / or inconsistency unevenly among the decomposed sinograms 202. Fig. 2 also shows that the FS method is applied differently to different sinograms in terms of the aggressiveness of the FS method (see 204b, 205b, 206b). This results in a reconstructed basis image 206 that can be used to generate a spectral result. The principle of the embodiment shown in Fig. 2 will be elucidated below by other detailed embodiments. Detailed Exemplary Embodiments

[0075] An industrial dual-energy CT scanner consists of collecting signals at two energy bins. In a photon-counting CT system, n bin The number of bins, denoted by n, can be increased, for example, to 5. By projected domain decomposition, n m n equivalent paths are numerically decomposed, where m ≦n bin This can be done by either inverting the analytical equation or maximizing the likelihood function. Let us call these equivalent paths, A α (α=1,2…n m ) is expressed as

[0076] For convenience, the above equivalent path is expressed as follows: L α where μ α is the A at a fixed energy (e.g., 70 keV) α is the damping coefficient of the material corresponding to

number

[0077] Next, I α Basis images, denoted by , are reconstructed from the sinograms of these line integrals by the filtered back projection (FBP) method. These images provide all the spectral results, e.g. virtual monochromatic energy images, iodine density weighted images, K-edge material images, etc. To allow accurate quantitative imaging, these images must be free of CBA. Here, the FS method can be used. However, straightforward use of this method may lead to an undesired increase in the residual low-frequency noise remaining in Iα after applying image domain denoising methods. In this case, this residual noise propagates rather linearly in the spectral results. To avoid this noise increase, this embodiment proposes to selectively, yet effectively, use the FS method. For this purpose, in this exemplary embodiment, first L α Apply a unitary basis transformation to .

number

[0078] Basis transformation matrix U βα are different sinograms L β The idea here is to selectively use the FS method for each basis element β to generate a sinogram L β Image I β i.e., reconstructing L that is known to have larger low frequency noise than the others. β This means reducing the aggressiveness of FS against the sinogram.

[0079] Transformation matrix U βα can also be selected using different a priori knowledge aimed at distributing the mismatch due to wide cone angles in a non-uniform manner among these different sinograms. In this case, the aggressiveness of FS can be reduced for those known to contain fewer cone angle mismatches than others, making L β Reconstruct the sinogram.

[0080] Either of these two approaches, or a combination of both, can be used to obtain a satisfactory reduction in CBA while avoiding an increase in the residual low-frequency noise in the spectral result.

[0081] Control of FS aggressiveness according to this exemplary method is performed by varying the cutoff or shape of the low frequency filter used by the FS method, as described in the FS method paper cited above. Alternatively, the aggressiveness can also be changed by modifying the BP weighting configuration used to generate the low frequency image.

[0082] base L α From L β Since the conversion to is linear, I β From I α There is no need to apply the inverse transform afterwards to get I βThe spectral results can be extracted directly from the , a flow chart of which is shown in Figure 2. In the following an exemplary embodiment of this method is presented in more detail.

[0083] The method is designed for a variety of numbers of decomposed equivalent paths. Nevertheless, to visually illustrate this, we use n m =n bin Consider a dual energy system where L = 2. α In the basis L, L = 1 represents the line integral through a hypothetical material with an energy-dependent attenuation profile similar to that of the Compton scattering mechanism in water. α Select L α We complete the basis by choosing =2 as the line integral through a hypothetical material with an energy-dependent attenuation profile similar to the sum of the attenuation profiles of Rayleigh scattering and photoelectric mechanisms in water.

[0084] In Figure 3, L α Figure 1 shows typical scatter plots of the noise components of the . To obtain these, a simulation of the NCAT mathematical phantom scan is used. The noise components are the noisy L α The L of a given detector row is obtained after removing the noise-free simulation value from α=1 +L α=2 Figure 3 shows an aqueous sinogram of 100 x 100 mm. Two regions of interest (ROIs) are defined within the figure (see elliptical contours). Figure 3(b) shows the noise content of the measurements belonging to these ROIs. Figure 3 provides two important observations. First, the major principal axis is oriented at approximately -45°.

number

number

[0085] These findings m ≧2, we generalize to different L α Describe the noise correlation between,Σ αα´ We introduce the covariance matrix expressed as: βα But U T This helps to diagonalize ΣU. Its columns are αα´ If we arrange these in descending order according to their eigenvalues, we get the path L β The noise in the sinogram of reduces β as designed.

[0086] covariance matrix Σ αα´ According to the matrix U βα Selecting the sinogram L of different paths β This approach is based on the underlying assumption that the amount of wide-cone angle mismatch distributed across the spectral results is roughly the same across all Ls. However, as briefly mentioned above, other approaches use a priori knowledge of the patient's anatomy to generate Ls that contain different amounts of wide-cone angle mismatch. β The first step is to carefully construct the sinograms of the CT scan. An example of such a priori knowledge is the fact that most of the image CBA comes from a strong gradient in the patient's cortical bone content along the rotation axis z. Here, we ignore the air-soft tissue interface in the diaphragm region. The X-ray attenuation of the cortical bone is, within the approximation contained in the two basis model, determined by the dual energy basis element L α=1,2 According to this input, the basis transformation matrix given by Eq. (5) can be written as a linear combination of the two virtual materials defined above.

number

number

number

[0087] The inventors of the present invention have also been able to demonstrate the advantages provided by the method presented herein in the tests shown in Figures 5(a)-(c). The sagittal brain images shown in Figures 5(a)-(c) are reconstructed from a simulated dual energy helical scan with a beam aperture of 8 cm at the center of rotation, a pitch factor of 0.3, and 340 mA. As shown in Figure 5(a), if the FS method is not applied to the reconstruction, an unacceptable band of CBA appears around the cerebellum. In Figure 5(b), the inventors have used the FS method to reconstruct the original basis element L α=1,2 was reconstructed using a low-frequency filter (LPP) that rolls off to zero at 4 line pairs per cm. Here, the suppression of the CBA is sacrificed by an increase in residual low-frequency noise, which impairs the distinction between gray and white matter and corrupts the image more than the original CBA. The residual low-frequency noise (in xy) also causes image blurring around the cerebellum. In FIG. 5(c), the inventors have non-uniformly distributed noise and / or misalignment among the decomposed sinograms, which leads to image cone-beam artifacts. As an example, U βα The FS method configuration was used to obtain the new basis, L β=2 While reconstructing the sinogram of L β=1 The sinogram was reconstructed without using the FS method. With the selective use of FS in the test of FIG. 5(c), no increase in residual low-frequency noise is observed with respect to FIG. 5(a). At the same time, a clear suppression of CBA is still achieved. Thus, as confirmed by these tests, the method of processing computed tomography (CT) data to suppress image cone beam artifacts (CBA) in a CT image to be reconstructed from the CT data provides improved reconstruction results compared to the prior art.

[0088] 4 shows a CT system 400 for processing CT data to suppress image cone beam artifacts (CBAs) in CT images, according to an exemplary embodiment of the present invention. The system 400 includes CT imaging systems 401, 402 and a system 403 with a computing unit 406, which: - receiving CT data generated during a CT scan of a patient; - decomposing the provided CT data in the projection domain to generate a plurality of decomposed sinograms; and - distributing noise and / or inconsistencies non-uniformly among the decomposed sinograms; It is configured as follows.

[0089] Also shown is a computer program element 404 which, when executed by at least one processing unit (PU), is adapted to cause the processing unit (PU) to perform the methods described herein.

[0090] Such a calculation unit 406 can solve the problems of the prior art of reconstructing CT images, in which a frequency decomposition method is applied but which brings about the drawback of increasing the low-frequency noise components of the image. This noise increase is even more troublesome in the spectral CT image reconstructed after projection domain decomposition. These drawbacks are overcome by the calculation unit 406, because after processing the sinogram with the FS method, the increase of the low-frequency noise components in the image to be reconstructed is avoided or at least reduced.

[0091] While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the dependent claims.

[0092] In the claims, the word "comprising" does not exclude other elements or steps, and the singular does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Reference signs in the claims should not be interpreted as limiting the scope.

Claims

1. 1. A method for processing computed tomography (CT) data to suppress image cone beam artifacts in a CT image to be reconstructed from the CT data, comprising: acquiring CT data generated during a CT scan of a patient; decomposing the acquired CT data in a projection domain to generate a plurality of sinograms; distributing noise and / or inconsistencies that lead to image cone beam artifacts non-uniformly among the plurality of sinograms; and applying filtered backprojection to the sinograms to which the non-uniform dispersion has been applied in the dispersing step, thereby reconstructing one or more basis images using a frequency decomposition method. The method comprising:

2. 2. The method of claim 1, further comprising applying a unitary basis transformation to the different sinograms in the projection domain, such that noise and / or inconsistencies that lead to image cone beam artifacts are distributed as non-uniformly as possible between the sinograms.

3. 2. The method according to claim 1, characterized in that the frequency division method is such that at least some low frequencies of the basis image are reconstructed without using all of the data of the acquired CT data, while higher frequencies of the basis image are reconstructed using more, preferably all, of the CT data.

4. The method of claim 1 or claim 3, wherein the frequency division method is selectively applied to the plurality of sinograms.

5. The method according to claim 1 , further comprising the step of varying the aggressiveness of the frequency division method for different sinograms.

6. The method of claim 5 , wherein the aggressiveness of the frequency division method is reduced for at least one sinogram with significant low frequency noise.

7. The method of claim 5 , wherein the aggressiveness of the frequency division method is reduced for at least one sinogram having a low level of cone angle mismatch.

8. 6. The method of claim 5, further comprising controlling variation in the aggressiveness of the frequency division method for different of the sinograms by varying the cutoff and / or shape of a low frequency filter used in the frequency division method.

9. 6. The method of claim 5, further comprising controlling variation in aggressiveness of the frequency division method for different sinograms by modifying a backprojection weighting configuration used to generate low frequency images.

10. The method of claim 1 , wherein the CT data is derived from a helical scan of the patient's body.

11. 1. A system for processing computed tomography (CT) data to suppress image cone beam artifacts in a CT image to be reconstructed from the CT data, the system comprising: a computing unit, the computing unit comprising: acquiring CT data generated during a CT scan of the patient; decomposing the acquired CT data in a projection domain to generate a plurality of sinograms; Distributing noise and / or inconsistencies that lead to image cone beam artifacts non-uniformly among the plurality of sinograms; and applying filtered backprojection to the non-uniformly dispersed sinograms, thereby reconstructing one or more basis images using a frequency decomposition method; system.

12. 1. A computed tomography (CT) system for processing CT data to suppress image cone beam artifacts in a CT image to be reconstructed from the CT data, comprising: CT imaging system, and A system according to claim 11 for processing CT data. A CT system having:

13. A computer program which, when executed by at least one processing unit, causes the processing unit to carry out the method of any one of claims 1 to 5 for processing CT data.

14. A computer readable medium storing a computer program according to claim 13.

Images