Method and system for motion compensation in ct reconstruction

Abstract

This invention relates to a method for motion compensation in CT reconstruction, comprising the following steps: providing a projection image (B) of a CT scan, creating a repacked intermediate image (Z), and calculating a slice image (S) using the following steps: a) selecting an initial motion state (MI); b) calculating a reference voxel position (x) with respect to the initial motion state (MI). M c) Calculate the column image position (p) of the voxel (V); d) Determine the changed motion state (M) of the voxel (V) at the column image position (p); e) Calculate the changed voxel position (x') with respect to the changed motion state (M); f) Calculate the changed image position (p', q') of the voxel; g) Perform backprojection using the image values ​​of the intermediate image (Z) at the changed image positions (p', q'). Furthermore, the present invention also relates to a corresponding system, control device, and computed tomography system.

Description

Technical Field

[0001] This invention relates to a method and system for motion compensation in CT reconstruction, specifically the reconstruction of layer images in computed tomography (CT), particularly taking into account the repackaging process. Background Technology

[0002] Movement during image acquisition in CT scans can cause artifacts, which can severely degrade image quality and reduce the diagnostic value of the resulting images. Movement can lead to inconsistencies in the acquired data, resulting in artifacts such as blurred, streaked, or ghosting images.

[0003] This movement is typically patient movement. Patient movement can be involuntary, such as organ movement or tremor, but it can also be intentional, such as in uncooperative patients, emergency situations, or pediatric imaging. However, artifacts can also be caused by movement of the scanner itself, such as in mobile CT systems with, for example, movable or slidable gantry systems.

[0004] The ideal solution to this problem is to completely prevent any form of movement during the scanning process; unfortunately, this is usually not a realistic option. Therefore, a solution is needed that can compensate for artifacts caused by movement and improve image quality.

[0005] The commonly used motion compensation method in CT reconstruction is The method proposed here is based on the assumption that motion present during CT scanning is known. Motion correction is applied during reconstruction in the backprojection step. Each voxel (3D image point) of the volume to be reconstructed is virtually moved according to the motion present at the time of capture. This method includes the reconstruction algorithm of Feldkamp, ​​David, and Kress ("Practical Cone-Beam Algorithm", J. Opt. Soc. Am. A 1, 612-619; 1984). Following the proposal of new CT scan motion estimation methods (e.g., the method of Bruder et al. ("Compensation for Skull Motion and Respiratory Motion in CT Using Data-Based and Image-Based Metrics, respectively", Proc. SPIE 9783, Medical Imaging 2016: Physics of Medical Imaging, 97831E; March 22, 2016)), The methods used by [authors' names] are frequently used as motion compensation methods.

[0006] Another method for motion compensation in CT reconstruction is based on partial-angle reconstruction (see J. Hahn et al., “Motion Compensation of Coronary Regions Based on Partial-Angle Reconstructions of Short-Scan CT Data,” Medical Physics, 44(11); 2017). In this case, motion compensation is performed on multiple partial-angle reconstructions separately, which are incomplete reconstructions based on subsets of data. A complete motion-compensated reconstruction is obtained by combining the motion-compensated partial-angle reconstructions. Motion compensation itself is related to… The methods used by others are very similar, namely, moving the voxel volume according to the current movement.

[0007] However, a drawback of the reconstruction method is that artifacts still exist when combined with reconstruction algorithms that include a repacking step, which reduces the quality of the motion compensation results. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for motion compensation in CT reconstruction, which avoids the aforementioned drawbacks and, in particular, enables optimal motion compensation after repackaging.

[0009] This objective is achieved by the method, system, control device, and computed tomography system according to the present invention.

[0010] The method for motion compensation in CT reconstruction according to the present invention includes the following steps:

[0011] – Provides projection images from CT scans.

[0012] – Create intermediate images, where intermediate images are created from the reboxing of columns of the projected image.

[0013] – The layer image is calculated by back-projecting an intermediate image onto a predetermined voxel of the layer image, wherein the following steps are performed for each intermediate image and each voxel based on a preset motion profile (with motion data) during the capture of the projected image:

[0014] a) Select the initial movement state based on the intermediate image.

[0015] b) Calculate the reference voxel position from the original voxel position and the movement profile regarding the initial movement state (i.e., movement data with the movement profile).

[0016] c) Calculate the column image position of the voxel that will be mapped to the reference voxel position in the intermediate image.

[0017] d) Determine the movement state of the voxels after the changes based on the calculated column image positions.

[0018] e) Calculate the changed voxel position from the original voxel position and the movement profile regarding the changed movement state.

[0019] f) Calculate the changed image position of the voxel that will be mapped to the changed voxel position in the intermediate image.

[0020] g) Use the image values ​​of the intermediate image at the changed image position for backprojection.

[0021] The provision of projection images from a CT scan can be achieved by capturing images using a CT scanner, simulating CT images, or accessing projection images stored therein. It should be noted that the term "projection image" does not refer to a reconstructed image, but rather to an image produced by projecting X-rays passing through the object being examined onto a detector. It is preferably the raw data from the CT scan, but can also be, for example, preprocessed data with noise reduction.

[0022] When an X-ray source and detector move along a circular or spiral track, a conical beam is typically used to capture projected images. However, this is essentially irrelevant to the present invention.

[0023] Intermediate images are created from the projected images using a repackaging process. These intermediate images are also projected images; however, for better distinction, the term "intermediate images" is used to indicate that these images are only used as an intermediate stage in the reconstruction. It should be noted that this step can also be performed before downloading the projected images from the database. Therefore, it is entirely feasible to take images, repack the projected images, and temporarily store the resulting intermediate images in the database.

[0024] In the repackaging of projected images, columns of the projected image are assigned to different intermediate images. Hereinafter, column coordinates are denoted by "p" and row coordinates by "q". In the case of cone X-rays, projected images are captured using the ray cone. Thus, the intensity information of the X-ray source is obtained at a single pixel on the detector (essentially corresponding to a bin), where, due to the cone shape of the X-rays, different detector pixels detect rays from different angles. However, some commonly used reconstruction methods require parallel ray geometry. Other projection methods may require different ray geometries.

[0025] To convert into an intermediate image of parallel "capture," rays are selected from various projection images (captured at different positions on the measuring device) that are parallel to the rays of the other projection images at the time of capture. Since the detector and X-ray source rotate during capture, and the capture position (in space) is known, the angle in space for each ray can be calculated from its capture position and the corresponding image position. In the case of parallel projection, the bin will not be a single pixel, but a single column of the projection image. For example, in a simple example, the first intermediate image can be formed by the values ​​of row p=1 of the first projection image, row p=2 of the second projection image, and so on. For the second intermediate image, it would be, for example, p=2 of the first projection image, p=3 of the second projection image, and so on. Parallel projection images are obtained in this way, where the column image position (p coordinate) simultaneously corresponds to the time progression, because the rows originate from different captures at different capture times, and thus also correspond to a certain movement state.

[0026] Optionally, the intermediate images can be filtered after repackaging. For example, convolution can be performed. During a preferred convolution step, the projected image is filtered to obtain better reconstruction results. However, as an alternative (or addition) to convolution, multiplication in Fourier space can also be performed. Here, the intermediate images are Fourier transformed and multiplied with an adaptive filter in Fourier space. This is more efficient than convolution of the intermediate images. This filtering is essentially a prior art technique.

[0027] Now, the layer image is calculated using the back projection of the intermediate image. This is known in the prior art (see, for example, the above). (Articles by Stierstorfer et al., mentioned below). However, a specific motion compensation step is additionally performed during this calculation, which takes into account repackaging. During the calculation, voxels (3D image points or volume pixels) corresponding to image points in the region where the object is being examined or in the layer image to be reconstructed are considered. Thus, the voxel values ​​ultimately reflect the image points in the reconstructed layer image, where each pixel of the layer image corresponds to a voxel of the layer's captured volume.

[0028] In (albeit known) backprojection, intermediate images are used to calculate the actual layer image. The purpose of the backprojection step is to reconstruct the voxel volume representing the patient or scanned object, which is formed by the aforementioned voxels. This is done by observing each voxel in each intermediate image separately (after repackaging and possible filtering). For each voxel, the ray passing through the voxel and the corresponding position in the intermediate image is calculated. The information of the corresponding position in the intermediate image is then used for backprojection. Once this operation has been completed for all voxels in the volume, the process is repeated for the next intermediate image until all intermediate images have been considered. By means of the present invention, the exact value in the current intermediate image will be used for backprojection.

[0029] Each voxel has a voxel position, which can be a three-dimensional coordinate or a three-dimensional vector. The following explanation will use coordinates, which thus include vectors. As mentioned above, a voxel represents a volume element of the object being photographed, and thus lies in a virtual space corresponding to the photographing area. Therefore, the voxel position is its location within this virtual space.

[0030] Motion compensation calculations assume a known motion profile. The motion profile pertains to the shooting area (the area where the object is located during shooting), and thus involves the movement of voxels. Since voxels are defined by their individual voxel positions, the motion profile involves changes in the individual voxel positions of each voxel. Therefore, the motion profile includes motion data for each voxel, or a collection of motion data for each voxel.

[0031] Creating motion profiles is existing technology. For example, a patient can be filmed by a camera during an imaging session, and a motion profile can be derived from the camera footage. However, motion estimation methods also exist for CT scans, such as those detailed in the paper by Bruder et al. (“Compensation for Skull Movement and Respiratory Movement in CT Using Data-Based and Image-Based Metrics, Separately”). Motion can be measured using sensors, such as motion sensors attached to the scanner to measure scanner movement or 3D cameras used to measure patient movement.

[0032] In simple terms, a motion profile is designed such that it (along with its motion data) reflects the motion of each individual voxel during the capture time, or at least allows for interpolation. Since this typically requires very large storage in practice (approximately 25GB in size with 512 voxels per image axis, 30 images, and two bytes per voxel coordinate), the motion profile can also be based on approximations, subscans, and / or interpolation, and / or include predefined regions without motion. For example, it is sufficient when the motion of certain points in the wall of the heart is known, and the motion of voxels in the heart volume is calculated based on these points, where only those voxels located within the motion regions of the wall must be considered. As described in further detail below, the motion profile can be based on approximations or models of motion. For example, rigid motion can be assumed, for which only translational, rotational, and displacement motion data of the center of rotation of the entire scanned object are needed at multiple time points (and at each relevant time point if necessary). Thus, in this case, the entire volume moves in the same way, and only one set of motion parameters is needed for the entire volume, rather than for each individual voxel. Therefore, this method requires very little storage space.

[0033] Using the motion profile, further reconstruction steps are now performed for each intermediate image and each voxel (with a previously known initial voxel position x). The aim here is to assign the correct image information to each voxel (here referring to voxels on the ray path, which is required for backprojection). Although it is known, as described above, to reconstruct layer images from intermediate images, the imaging of voxels is often not in their proper positions (in the absence of motion) within the intermediate images due to their motion. Therefore, in the prior art without motion compensation, voxels are sometimes assigned image content that does not actually correspond to the images of these voxels. This results in artifacts in the layer images, which is undesirable. Through the following special steps for motion compensation, the image information where the voxels should be found in the intermediate images is essentially determined, and this image information is accurately applied to the voxels.

[0034] To better understand the following implementation, it should be noted that the movement of a voxel is essentially only a change in its voxel position during the shooting time. During this shooting time, multiple projection images are captured at different shooting points, so that each projection image displays the voxel at its respective current voxel position. Here, each intermediate image also simultaneously displays the time progression in its column, since the rows come from different projection images. Thus, the column image position (p-coordinate) essentially corresponds to the time coordinate, and each p-coordinate also corresponds to a different movement state of the voxel.

[0035] Because the precise capture time is known for each P-coordinate in the intermediate images (it is known which projection image the image information at that P-coordinate originates from and when that projection image was captured), the motion state at that point is also known. Using the motion state (or capture time point), the current voxel position can be calculated from the motion profile. Essentially, the variables "motion state" and "capture time point" are equivalent in meaning for voxel positions, as a specific capture time point has a specific motion state. For simplicity, it is assumed below that voxel positions can be determined directly from the known motion state in the motion profile, where this also applies to the known capture time. It should be noted that the column image coordinates p of different intermediate images do not necessarily have to be the same as the motion state, because, for example, the center of an intermediate image corresponds to the same p-coordinate but typically to different motion states. However, as mentioned above, since the image content of the intermediate images originates from which projection image, the corresponding motion state can be derived from each p-coordinate of each intermediate image, and thus can be used as a "global parameter" for the p-coordinate.

[0036] In short, the motion profile includes motion data for different motion states. For a known motion state, the desired motion data can be selected from the motion profile and used to calculate the voxel position according to step b) or e). The motion data can also be presented as a function of the motion state. For the simple case of a simple translation z(t) with a translation function z and a motion state t, the voxel position x' after the change in the known motion state T can be easily calculated from x' = z(T).

[0037] For motion compensation, an initial motion state is first selected based on intermediate images, preferably with different initial motion states for different intermediate images. This initial motion state can be chosen arbitrarily, but is preferably located at the center of each considered intermediate image. If the centers of the intermediate images are considered separately, a simple and standardized calculation can be performed. Although each intermediate image is thus treated as a different motion state (the capture time of the center column image position is typically different for each intermediate image), this does not adversely affect the results, as all ultimately trace back to the (static) initial voxel volume.

[0038] However, it should be noted regarding the term "center" that there may be multiple reference systems that can define the "center" in the intermediate image. In some cases, this is the midpoint of the image, located at the halfway point of the column. However, the situation now is that the detector does not always have to be constructed symmetrically. If the center is defined by the rotation center of the gantry, it is possible that the detector may spend more time to the left of that center during image capture than to the right. This is then reflected in the p-coordinate of the intermediate image accordingly. In this case, the center is not necessarily located at the midpoint of the intermediate image, but may be slightly offset to the left or right.

[0039] However, it is also particularly preferred to assume that the center point is the midpoint of each image of the projected image (half of the column). The image midpoint can be considered as the center of the intermediate image, i.e., the column in the intermediate image where the image midpoint of the projected image exists. At this particular point, the rotation angle of the frame in the conical ray geometry is the same as the virtual rotation angle of the parallel ray geometry, so that the movement state based on the projected image and the corresponding intermediate image is the same at this point. As mentioned above, the image midpoint of the projected image does not necessarily have to be located at the image midpoint of the intermediate image. However, for the sake of simplicity and easier understanding of the invention, for the following embodiments, it is conceivable that the center is at the image midpoint of the projected image.

[0040] When calculating the reference voxel positions (voxel positions with respect to the initial movement state) from the movement profile, the voxel positions with respect to the selected initial movement state are now calculated. The movement profile is designed accordingly, as described above.

[0041] For example, in a simple but storage-intensive implementation, the required precision of the voxel position can be determined simply by looking up a lookup table for the corresponding motion state (shooting time point). However, the voxel position can also be calculated based on a function of the corresponding motion state.

[0042] For example, a motion profile can be preset to translate by z and rotate by R (R being a matrix, for example) around a rotation point c. Therefore, the motion state (denoted here by "t" to indicate its proximity to the shooting time) can be a function of z(t), c(t), and R. T The dataset. Then, according to formula x... M =R T (xc(T))+c(T)+z(T) calculates the voxel position x with respect to the initial movement state T from the original voxel position x. M It should be noted that this assumes the voxel to be reconstructed moves during the imaging process as the patient or the object being scanned moves (or possibly due to errors in the scanner's movement process).

[0043] That is, for a voxel at voxel position x, calculate the reference voxel position x of its initial movement state. M In the next step, we will calculate where the voxel (located at the reference voxel position) will be seen in the intermediate image.

[0044] In this regard, at least the calculation at the reference voxel position x MThe voxel at a given position will be mapped to either the column image position p or the 2D image position (p, q) in the intermediate image. Here, the position in the q direction is not a problem and has little relevance to the basic method. Since the ray path is known (which is the basis for repackaging), it is easy to determine where a ray with a known direction will be incident on the intermediate image through that voxel. For example, for parallel ray geometry, only the projection of the voxel onto the intermediate image needs to be considered.

[0045] However, it should be noted that in the intermediate images, each column image position corresponds to a specific shooting time point, and thus to a specific "changed motion state," which no longer needs to (and usually does not) correspond to the initial motion state. For this changed motion state (at the determined column image position p), the voxel may be located at another voxel position due to its movement.

[0046] In short, if a voxel is not directly displayed in the center of the intermediate image, there is another movement state that is different from the initial movement state, and the voxel may be located in a different position at that "point in time".

[0047] Therefore, the changed movement state of the voxel is now determined based on the calculated column image position p. The changed movement state can be calculated directly from the column image position.

[0048] Based on the calculation of the column image position p above, the changed column image position p' is now calculated. However, the changed row image position q' is also calculated here, resulting in two-dimensional coordinates (p', q'). The calculation of the changed row image position q' can be simply performed by calculating the angle at which the ray passing through the voxel at the changed voxel position x' will be incident on the intermediate image. However, it should be noted that if a conical ray geometry exists during the imaging process, the conical shape in the q direction will be preserved even in the case of parallel repackaging.

[0049] Finally, the image value at the transformed image position (p', q') is obtained in the intermediate image and used as the backprojection value (and assigned to a voxel or a set of voxels if necessary).

[0050] This is done for all voxels of the intermediate image, and then again for each voxel of another intermediate image, until all relevant intermediate images have been processed.

[0051] Essentially, the process can be summarized as follows: instead of processing at the first optimal voxel position, an iterative step is performed first. While this method may seem inaccurate, since the voxel at the "time point" of the "changed column image position p'" might be located elsewhere, it has proven effective in significantly improving image quality with a single iterative step. However, as described below, further iterative steps can certainly be performed, each requiring additional computational effort.

[0052] Therefore, this invention integrates motion compensation into backprojection and is particularly advantageous for algorithms such as the weighted and filtered backprojection algorithm (WFBP algorithm) proposed by K. Stierstorfer et al. (“Weighted FBP—A Simple Approximate 3D FBP Algorithm for Multislice Spiral CT with Good Dose Use Applicable to Arbitrary Pitch.” Physics in Medicine and Biology, 49(11), 2209; 2004), and can be advantageously used for CT reconstruction. However, the motion compensation method is essentially advantageous for any CT reconstruction method that uses repacking and backprojection steps.

[0053] The system for motion compensation in CT reconstruction according to the present invention is preferably designed to perform the method according to the present invention. The system includes the following components:

[0054] – Data interface, designed to receive projected images from CT scans.

[0055] – A repacking unit, designed to create intermediate images, where intermediate images are created from repackaging columns of the projected image.

[0056] – A reconstruction unit, designed to compute layer images by backprojecting intermediate images onto predetermined voxels of the layer image.

[0057] The reconstruction unit (in addition to the elements known in the prior art for reconstruction) includes the following additional components:

[0058] – The movement module is designed to select the initial movement state of the intermediate image, or to determine the changed movement state of the voxels based on the calculated column image positions of the voxels in the intermediate image.

[0059] – A positioning module designed to calculate the voxel position from the voxel's original voxel position and a movement profile with respect to the selected or calculated movement state.

[0060] – An imaging module designed to calculate the column and row image positions that will map voxels in the intermediate image.

[0061] – The reuse module is designed to reuse the image values ​​of the intermediate image at the changed image position for backprojection.

[0062] Data interfaces are known in the prior art. They can be, for example, network data interfaces and are designed, for example, to receive images via a PACS (Picture Archiving and Communication System). However, a data interface can also be the data bus of a CT system through which projected images captured by CT scans are transmitted.

[0063] The repacking unit is used to create intermediate images, such as images with parallel ray guidance, by repacking as described above.

[0064] Reconstruction units are generally known in the prior art for computing layer images. The reconstruction units used herein include elements of conventional reconstruction units and additionally include other modules listed above, which will be described in more detail below.

[0065] The characteristic of a mobility module is that it predefines a mobility state. The mobility state can be determined in two modes, and for this purpose, the mobility module can (but does not necessarily have to) have two sub-modules.

[0066] The first mode is to select the motion state. This is based on user presets, pre-adjustments, or the currently used intermediate image. For example, selection can be made by choosing the motion state corresponding to a specific column position p (e.g., its center) of the intermediate image. However, it is also possible to select a motion state associated with a specific shooting time point. The corresponding submodule can be called the "motion selection module".

[0067] The second mode determines the voxel's movement state based on the calculated column image position p. This only requires specifying the column position and selecting the movement state associated with that column position (in the corresponding intermediate image). The corresponding submodule can be called the "movement position module".

[0068] Basically, a simple movement module can have an assignment table or assignment function, through which a corresponding movement state can be assigned to the image coordinates of each intermediate image.

[0069] The positioning module communicates with the movement profile, for example, through communication with a database or storage area. The movement profile contains information about voxel movement, for example in the form of functions or tables, and the positioning module is designed to invoke this information and, in particular, write it to a predetermined storage area.

[0070] The voxel positions can be calculated as described above. Here, the localization module can calculate not only the reference voxel position but also the changed voxel position, since the calculation principle is the same. However, it can also have two sub-modules to calculate these variables separately.

[0071] The imaging module calculates image positions in the manner detailed above. Regarding the reference voxel position, only the column image position of the voxel in the intermediate image needs to be calculated, but its row image position can also be calculated additionally. For the changed voxel position, both the column image position and its row image position of the voxel in the intermediate image must be calculated so that appropriate image information can be assigned to the voxel later. The imaging module could ideally have two sub-modules to calculate the variables independently.

[0072] The reuse module can easily reuse the image values ​​of the intermediate image by copying the image information at the previously calculated image coordinates and writing it into the dataset used for backprojection.

[0073] These modules may be included in the control unit of the CT system. The control unit for controlling a computed tomography system according to the invention is designed to perform the methods according to the invention and / or include the system according to the invention.

[0074] The computed tomography (CT) system according to the invention includes a control device or system according to the invention and is designed to perform the method according to the invention. The CT system itself is known in the prior art.

[0075] Most of the aforementioned components of the system or control device can be implemented, wholly or partially, as software modules in the processor of the corresponding computing system. This largely software-based implementation has the advantage that previously used control devices can be easily modified through software updates to operate according to the invention. In this regard, this…

[0076] The objective of the invention can also be achieved by a corresponding computer program product having a computer program, which can be directly loaded into the computing system or storage device of the control device of a computed tomography system, having program segments for executing all steps of the method according to the invention when the program is run in the computing system or control device. In addition to the computer program, such a computer program product may, if necessary, include additional components such as documentation and / or other related elements.

[0077] Or additional components such as hardware components, for example, hardware keys (donkeys, etc.) used to use software.

[0078] For transmission to and / or storage on or within a computing system or control device, a computer-readable medium, such as a memory stick, hard disk, or any other removable or fixedly embedded data carrier, may be used, on which program segments of a computer program that are readable and executable by the computing unit of the computing system or control device may be stored. For this purpose, the computing unit may, for example, have one or more cooperating microprocessors.

[0079] Other particularly advantageous designs and modifications of the invention are described below, wherein claims of one claim class may be modified in a manner similar to claims of another claim class and description portion, and in particular individual features of different embodiments or modifications relating to new embodiments or new variations may be combined with each other.

[0080] In the preferred method, the initial movement state corresponds to a column image position p that is substantially located at the center of the intermediate image. "Substantially located at the center of the intermediate image" means that the column image position P is no more than 5% of the total number of columns in the intermediate image, preferably no more than 1% of the total number of columns.

[0081] The particularly preferred initial movement state corresponds to column image position p, which corresponds to those columns in the intermediate image that are exactly in the middle of one of the projected images.

[0082] Here, it is preferable to select such an initial movement state for each intermediate image, and more preferably to select an initial movement state for each intermediate image corresponding to the column image position that is exactly located at the center of the corresponding intermediate image.

[0083] According to a preferred method, in addition to calculating the column image positions of the mapped voxels in the intermediate image, the row image positions of the mapped voxels in the intermediate image are also calculated.

[0084] Regarding the calculation of motion profiles and voxel positions, it should be noted in practice that, in terms of hardware, image reconstruction typically involves two working memories: a smaller, faster one and a larger, slower one. Computations are usually performed in the faster working memory. It is advantageous to write the data required for the computation into the faster working memory, so that the computation is not unnecessarily slowed down by access time to the slower working memory. As mentioned above, since the motion profile can be very large, it is usually stored in the slower working memory (or fixed memory). Therefore, it is advantageous to load a certain amount of motion data from the motion profile into the faster working memory for each intermediate image, and interpolate the temporally located motions between these motion states based on the downloaded motion data. This saves computation time. Experiments show that, in addition to the initial motion state, two other motion states are sufficient for effective motion compensation.

[0085] According to a preferred method, in the backprojection of intermediate images, for each of the multiple intermediate images, particularly for each intermediate image in the motion profile, motion data (regarding voxel movement) is pre-selected from the motion profile for an initial motion state and for at least one earlier motion state and at least one later motion state. Thus, for example, there exists motion data for the initial motion state at the p-coordinate of the center of the intermediate image and corresponding motion data for the motion states at the p-coordinates to the left and right of the center.

[0086] In this case, the changed voxel position x' is preferably calculated based on these pre-selected movement data and the determined changed movement data by interpolating "suitable" movement data for the changed movement state. For this purpose, it is preferable to use the movement data of the most recent movement state.

[0087] Therefore, interpolation is preferably based on those movement states that are closest to the changed movement state. This means determining the relationship between each next pre-selected movement state and the reference voxel position x. M The distance to the column image position p is used, and the changed voxel position x′ is calculated using (pre-selected) movement data from the movement profiles of adjacent movement states. Here, the distance between the image positions of the movement states is preferably considered in a weighted form. Reference voxel position x M The closer the column image position p is to the adjacent moving state, the more its moving data is considered in the calculation of the voxel position.

[0088] In short: Select some movement states (e.g., three), determine movement data for these movement states based on the movement profile, and for a new movement state between the selected movement states, interpolate the corresponding movement data based on the movement data of the adjacent movement states. Then, use the interpolated movement data to calculate the voxel position.

[0089] Therefore, for the three movement states—the initial movement state at the center (p0) and the two other movement states at the other two P coordinates pmin and pmax—which P coordinates are closest (e.g., pmin and p0) will be determined given the calculated column image position p. The distances from p to these positions (e.g., A1 to pmin and A2 to p0) will be determined, and then the movement will be weighted using weighting factors a1 and a2 (e.g., a1 = A1 / (A1+A2) and a2 = A2 / (A1+A2)). The movement data R is used for the "new" movement state T2 and the two adjacent known movement states T and T1, as well as the movement data R for these two known movement states. T R T1 A preferred method for c(T), c(T1), z(T), and z(T1) is:

[0090] –Based on R Tand R T1 Interpolation calculation of rotation R T2 .

[0091] – Calculate the rotation point c(T2) based on interpolation of c(T) and c(T1), for example, c(T2) = a1·c(T 1) + a2·c(T).

[0092] – Calculate the translation z(T2) based on interpolation of z(T) and z(T1), for example, z(T2) = a1·z(T 1) + a2·z(T).

[0093] – Calculate the changed voxel position using the following formula:

[0094] x′=R T2 (xc(T2))+c(T2)+z(T2)

[0095] It should be noted that when the center is not the midpoint of the middle image, the outer regions pmin (always negative) and pmax (always positive) do not necessarily have to be symmetrically distributed around p=0, and pmin ≠ -pmax.

[0096] According to a preferred method, the following steps are traversed at least once before backprojection using image values:

[0097] – Calculate the changed column image position p′ of the voxel that will be mapped to the changed voxel position x′ in the intermediate image.

[0098] – Determine the movement state of the voxel after the change based on the changed column image position p′.

[0099] – Calculate the changed voxel position x' of the voxel from the original voxel position x and the movement profile with respect to the changed movement state.

[0100] Therefore, at least one new, modified voxel position will be calculated iteratively. This takes into account the case where each p-coordinate in the intermediate image corresponds simultaneously to both the individual's shooting time and their movement state.

[0101] According to a preferred method, when creating an intermediate image, the columns of the projected image are repacked such that those columns of the projected image, which are photographed with correspondingly parallel X-rays in a plane orthogonal to the columns, are used for the intermediate image. This is known in the prior art; however, the method according to the invention is particularly suitable for this parallel repacking.

[0102] According to a preferred method, in particular, filtering the intermediate image using convolution and / or Fourier transform yields better reconstruction results. In the preferred convolution, the filtered intermediate image is computed using a predefined kernel, particularly a ramp filter such as the Shepp-Logan kernel. In an alternative filtering method, the intermediate image undergoes a Fourier transform and is multiplied by an adaptive filter in Fourier space. Filtering and Fourier transform are largely known in the prior art.

[0103] According to a preferred method (a special iterative method), the following steps are performed after calculating the layer image from the intermediate image:

[0104] i) Reconstruct and compare intermediate images from layer images based on the moving profile.

[0105] ii) Compare the intermediate images with each other.

[0106] iii) Create a modified layer image based on this comparison;

[0107] iv) Iteratively repeat steps i) to iii) using the corresponding previously created layer image.

[0108] According to a preferred method, when reconstructing the intermediate image for each image point of each intermediate image, the image points of the intermediate image are determined according to the following steps:

[0109] – Calculate the ray that passes through the image voxels, starting from an image point in the intermediate images.

[0110] –Motion data that determines the motion state by comparing the image positions in intermediate images in the motion profile.

[0111] – The positions of the ray and image voxels are shifted relative to each other based on the motion data in the motion profile regarding the motion state, specifically where the ray is moved according to the motion data.

[0112] –As the ray and image voxels shift relative to each other, the values ​​of the image voxels along the ray are accumulated.

[0113] – The accumulated value is used to compare the positions in the intermediate images.

[0114] Therefore, the pixel at position (p, q) in the intermediate image should be filled with a value. For this, a ray passing through the image voxel and being absorbed accordingly is calculated. The (correct) value of the ray should then be the accumulation of the image voxel's value along the ray. Now (according to back projection) the situation is such that the voxel volume can be located in different places at different times. However, since the column position p is known (it is chosen), the corresponding movement state is also known, and thus the corresponding movement data from the movement profile is also known, which can be directly derived from the movement profile or interpolated. Here, interpolation works accordingly, as in back projection. Then, using the corresponding movement data, the start and end points of the ray can be shifted according to the movement (especially in rigid movements), or alternatively, the image voxel can be shifted (especially in non-rigid movements). For the ray after this shift (relative to the image voxel), the value is accumulated along the ray, and this value is written to the original position (p, q). Here, this is clearly not the new position of the ray, but rather the pixel starting from it according to the iteration of the intermediate image pixels.

[0115] Therefore, in the preferred system, the reconstruction unit additionally includes the following modules:

[0116] – The ray simulation module is designed to calculate rays that travel from a point in an image through a voxel.

[0117] – The motion simulation module is designed to determine motion data for the motion state based on the column image positions in the comparative intermediate images from the motion profile, and to move the positions of the rays and image voxels relative to each other based on the motion data from the motion profile regarding the motion state.

[0118] – The simulation carry-over module is designed to accumulate the value of the image voxel along the ray as the ray and the image voxel shift relative to each other, and carry over the accumulated value at the position in the comparison intermediate image.

[0119] This iterative motion compensation method can be incorporated into the reconstruction process to obtain motion-compensated voxel volumes. If a forward projection step is performed to obtain a comparative intermediate image without considering motion, the resulting comparative intermediate image will also be motion-compensated. However, it will be compared with the (uncompensated) original intermediate image. Therefore, during the forward projection step, the same motion compensation calculations as in the back projection should be performed, only in "other directions." In the forward projection, the motion removed in the back projection step (all based on the motion profile) is added back. In the subsequent back projection step, the motion is compensated again.

[0120] In short, the forward projection process calculates (compares) intermediate images from the layer images, compares these intermediate images with the original intermediate images, then backprojects the differences between the intermediate images back, and uses the previous layer images to perform offset calculations on the results to obtain a more accurate result. This new, more accurate layer image is then forward projected again, and so on.

[0121] The components of this invention are preferably used as a "cloud service." This cloud service is used for data processing, particularly with the aid of artificial intelligence, but can also be a service based on traditional algorithms or a service subject to manual evaluation in the background. Typically, a cloud service (hereinafter also simply "the cloud") is IT infrastructure in which, for example, storage space or computing power and / or application software is provided via a network. Here, communication between the user and the cloud occurs through a data interface and / or data transmission protocol. In the present context, it is particularly preferred that the cloud service provides not only computing power but also application software.

[0122] Within the scope of preferred methods, data is provided to cloud services via a network. This includes computing systems, such as computer clusters, which typically do not include the user's local computer. The cloud can be provided, particularly by medical institutions that also provide medical technology systems. For example, image acquisition data is sent to the (remote) computing system (cloud) via a RIS (Radiation Information System) or PACS. The cloud's computing system, network, and medical technology system preferably constitute a network in a data technology sense. Here, the method can be implemented using a set of instructions within the network. The data computed in the cloud ("result data") is then sent to the user's local computer via the network.

[0123] Compared to existing technologies, the disclosed invention significantly improves motion compensation results for known movements, particularly when combined with the WFBP algorithm. After motion compensation reconstruction, the results are significantly better when combined with the WFBP algorithm than, for example, in… The method of [Authors' Name] leaves fewer motion artifacts. The main difference is that, in the disclosed invention, the repackaging process is taken into account in motion compensation, which leads to a more accurate representation of the motion states present in each step, thereby reducing residual artifacts in the final result.

[0124] The significant reduction in motion artifacts improves image quality, thereby enhancing the diagnostic value of the resulting images. Therefore, improvements in image quality in motion-distorted CT scans are highly desirable for clinical practice, as they enable greater accuracy in the diagnostic process.

[0125] Furthermore, this invention avoids the need for repeated CT scans due to motion artifacts. Repeated CT acquisitions involve increased patient dose, which is highly disadvantageous. Moreover, re-acquiring scans is very time-consuming and may require significant resources.

[0126] For the mobile CT scanner project, the disclosed motion compensation method can contribute to improved image quality, making the scanner more attractive for clinical practice. Attached Figure Description

[0127] The invention will now be explained in more detail with reference to the accompanying drawings and embodiments. Here, the same parts are referred to by the same reference numerals in different drawings. The drawings are generally not drawn to scale. Wherein:

[0128] Figure 1 A rough schematic diagram of a computed tomography system with control devices is shown, which has a system according to the invention for performing the method.

[0129] Figure 2 A flowchart illustrating a possible process of the method according to the present invention is shown.

[0130] Figure 3 A supplementary flowchart illustrating a possible flow of the iterative method is shown;

[0131] Figure 4 A flowchart of the preferred iterative method is shown.

[0132] Figure 5 A schematic diagram of the preferred system is shown.

[0133] Figure 6 An image of the moved voxel is shown.

[0134] Figure 7 The repackaging was shown.

[0135] Figure 8 The selection of the movement state is shown.

[0136] Figure 9 Comparison results of the method according to the present invention are shown. Detailed Implementation

[0137] In the following description, it is assumed that the imaging system is a computed tomography (CT) system. However, in principle, this method can also be used with other imaging systems.

[0138] Figure 1A computed tomography (CT) system 1 with a control device 10 for performing the method according to the invention is shown schematically. The CT system 1 typically includes a scanner 2 with a gantry, wherein X-ray sources 3 rotate and pass through a patient, who is pushed into the measurement chamber of the gantry by means of a bed 5, so that the rays are incident on detectors 4 corresponding to the X-ray sources 3. It should be clearly noted that this embodiment is only one example of CT, and the invention can also be used in any CT design, such as one with a ring-shaped fixed X-ray detector and / or multiple X-ray sources.

[0139] Similarly, only components essential to the explanation of the invention are shown in control unit 10. Essentially, this CT system and its associated control unit are known to those skilled in the art and therefore do not require detailed explanation.

[0140] Here, the core component of the control device 10 is the processor 11, on which various components are implemented as software modules. In addition, the control device 10 also has a terminal interface 14, where a terminal 20 is connected, allowing the user to operate the control device 10 and thus the computed tomography (CT) system 1. Another interface 15 is a network interface for connecting to the data bus 21 to establish a connection with a RIS (Radiological Information System) or PACS (Picture Archiving and Communication System).

[0141] The scanner 2 can be controlled by the control device 10 via the control interface 13, for example, controlling the rotation speed of the gantry, the displacement of the patient bed 5, and the X-ray source 3 itself. Raw data RD is read from the detector 4 via the acquisition interface 12. Furthermore, the control device 10 has a memory 16, which stores various measurement protocols in particular.

[0142] As a software component, an image data reconstruction unit 18 is implemented on the processor 11, which reconstructs the required image data from the raw data RD received through the data acquisition interface 12. This reconstruction unit 18 is a module of a system 9 for motion compensation in CT reconstruction. The system 9 itself includes:

[0143] Data interface 6 is designed here to receive raw data RD (or projected image B from data acquisition interface 12).

[0144] The repacking unit 7 is designed to create an intermediate image Z, which can be created from the repacking using parallel ray guidance, such as... Figure 7 As shown.

[0145] Here, reconstruction unit 8 also includes the following modules:

[0146] The motion module 30 is designed to select an initial motion state MI of the intermediate image Z, or to determine the changed motion state M of voxel V based on the calculated column image positions p, p' of voxel V in the intermediate image Z. Therefore, the motion states M and MI can be determined using two modes. The first mode is to select the motion state MI. This is, for example, based on a preset made using the currently used intermediate image Z (see example...). Figure 8 For example, the selection can be made as follows: select the motion state MI corresponding to the center of the intermediate image Z. The second mode determines the motion state M based on the calculated column image position p. For this, only the column position needs to be specified, and the motion state M associated with that column image position p (of the corresponding intermediate image Z) is selected accordingly.

[0147] Positioning module 31 is designed to calculate the voxel position x', x' of voxel V from the original voxel position x and the movement profile BP with respect to the selected or calculated movement states MI, M. M Here, the positioning module 31 is associated with the motion profile BP, which resides in storage unit 16. The calculation of the voxel position is as follows. Figure 2 The details are explained in more detail below, in which the positioning module 31 is able to calculate the reference voxel position x. M And the position of the changed voxel x'.

[0148] Imaging module 32 is designed to calculate the column image positions p, p′ and row image positions q, q′ of voxel V in the intermediate image Z. Here, regarding the reference voxel position x... M Only the column image position of voxel V in the intermediate image Z needs to be calculated, but its row image position q can also be calculated additionally. For the changed voxel position x′, both the column image position p′ and the row image position q′ of the voxel in the intermediate image Z must be calculated so that appropriate image information can be assigned to voxel V later.

[0149] The reuse module 33 is designed to reuse the image values ​​of the intermediate image Z at the changed image positions p′, q′ for backprojection. The reuse module 33 can achieve the reuse of the image values ​​of the intermediate image Z, for example, by copying the image information at the previously calculated image coordinates p′, q′ and writing it into the dataset used for backprojection.

[0150] Figure 2 A flowchart illustrating a possible process for a method of motion compensation in CT reconstruction according to the present invention is shown.

[0151] In step I, a projection image B from a CT scan is provided, wherein the image is derived from... Figure 1The X-ray source 3 of the CT scanner 2 irradiates the detector 4 with X-rays and takes a series of projected images B.

[0152] In step II, an intermediate image Z is created, which is generated by repackaging the columns of the projected image B. Here, the parallelized intermediate image Z is created from the projected image B, which was captured with a conical ray.

[0153] In step III, the layer image S is calculated by back-projecting the intermediate image Z onto a predetermined voxel V of the layer image S, wherein the following steps are performed for each intermediate image Z and each voxel V based on a preset moving profile BP of voxel V during the capture of the projected image B:

[0154] In step IIIa, an initial motion state MI is selected based on the intermediate image Z. The initial motion state may correspond, for example, to coordinates that are exactly located at the center (e.g., the middle of the image) of the intermediate image Z.

[0155] In step IIIb, the reference voxel position x of voxel V is calculated from the original voxel position x of the voxel and the movement profile BP with respect to the initial movement state MI. M Voxel V may have moved to the reference voxel position x in the initial movement state MI with respect to the original voxel position x. M This movement has been considered.

[0156] In step IIIc, the value x will be calculated as the reference voxel position mapped in the intermediate image Z. M The column image position p of voxel V is given. The complete image positions p and q are calculated here, where the row image position q is not necessarily needed. This can be done, for example, by using the projection of voxel V onto the intermediate image Z.

[0157] In step IIId, the movement state M of voxel V after the change is determined based on the calculated column image position p.

[0158] In step IIIe, the changed voxel position x′ of voxel V is calculated from the original voxel position x of voxel V and the movement profile BP with respect to the changed movement state M.

[0159] In step IIIf, the changed image positions p′ and q′ of voxel V, which will be mapped to the changed voxel position x′ in the intermediate image Z, are calculated.

[0160] In step IIIg, the image values ​​of the intermediate image Z at the changed image positions p′ and q′ are used for backprojection.

[0161] The dashed arrows indicate that steps IIId to IIIf can be repeated or performed multiple times in the form of iterative improvements to the results, where the changed column image position p′ in step IIId forms the basis.

[0162] Figure 3 A supplementary flowchart illustrating a possible flow of the iterative method is shown, wherein, based on... Figure 2 After calculating the layer image S from the intermediate image Z, perform the following steps:

[0163] In step IIIh, the intermediate image ZV is reconstructed from the layer image based on the moving profile BP. This step is divided into four sub-steps.

[0164] In step IIIh 1 In this process, a ray is calculated that starts from an image point in the intermediate image ZV and travels through the image voxel VB.

[0165] In step IIIh 2 In this process, the movement data of the movement state M is determined based on the column image position p in the comparison intermediate image ZV from the movement profile BP, and the positions of the ray and the image voxel VB are shifted relative to each other based on the movement data of the movement profile BP regarding the movement state M, in particular, the ray is moved according to the movement data.

[0166] In step IIIh 3 In the process, as the relative displacement of the ray and the image voxel VB increases, the value of the image voxel VB is accumulated along the ray.

[0167] In step IIIh 4 In the comparison, the cumulative values ​​of positions p and q in the intermediate image ZV are used.

[0168] In step IIIi, the intermediate image ZV is compared with the intermediate image Z.

[0169] In step IIIj, a modified layer image S is created based on this comparison.

[0170] Steps IIIh, IIIi, and IIIj are iteratively repeated using the corresponding previous layer image S created.

[0171] Figure 4 A flowchart of a preferred iterative method is shown.

[0172] As described above, a layer image S is generated from the intermediate image Z (denoted as Z0 in the formula below) using backprojection Q. This backprojection Q combines convolution and backprojection themselves in one step. The first layer image S (in formula f0) is iteratively improved using each round (fk). Forward projection (in formula P) calculates a comparison with the intermediate image ZV from the layer image S (according to P(f...)).k Then, calculate and compare the difference between the intermediate image ZV and the original intermediate image Z: P(f k The result is then back-projected: Q(P(fk)-Z0), and the result is subtracted from the previous layer image by a coefficient a: fk-aQ(P(fk)-Z0). k This will yield the next layer image f. k+1 Then repeat these steps.

[0173] In forward projection, the motion is added instead of removed as in back projection. After motion-compensated back projection, the layer image S ideally has no motion. In forward projection, for each intermediate comparison image, ZV is no longer iterated by voxels, but by the position (p, q) of the intermediate comparison image ZV. For each position in the intermediate comparison image ZV, the corresponding ray is calculated using the corresponding voxel volume, and the voxel volume along the calculated ray value is written to the position in the intermediate comparison image ZV.

[0174] Since the position in the intermediate comparison image ZV is known, a movement state M belonging to the corresponding column image position p can be determined to account for movement. For example, this movement state M could be used to move a ray through the voxel volume instead of the entire voxel volume itself. However, in principle, both are feasible. The value along the newly calculated ray can now be written to the corresponding position in the intermediate comparison image ZV. Thus, an intermediate comparison image ZV that also incorporates movement will be calculated.

[0175] Figure 5 A schematic diagram of the preferred system 9 is shown. The basic structure of system 9 has been described. Figure 1 As shown, however, besides data interface 6, repackaging unit 7 and Figure 1 The preferred embodiment includes modules such as the moving module 30, positioning module 31, imaging module 32, and reuse module 33. It also includes other modules in the reconstruction unit 8. The system 9 utilizes these modules and is designed for applications such as... Figure 3 The method is shown. These additional modules include:

[0176] Ray simulation module 34 is designed to calculate a ray that starts from an image point and travels through an image voxel VB.

[0177] The motion simulation module 35 is designed to determine motion data of motion state M based on column image position p in the comparative intermediate image ZV from the motion profile BP, and to shift the positions of ray and image voxel VB relative to each other based on the motion data of motion profile BP regarding motion state M, in particular, wherein the ray is moved according to the motion data.

[0178] The simulation uses module 36, which is designed to accumulate the value of the image voxel VB along the ray as the ray and the image voxel VB are relatively shifted.

[0179] Figure 6 The image of the moving voxel V is shown. The solid-line cube shows 27 voxels V, where voxels V are imaged on the upper right rear of their original voxel position x. This image is taken in the case of projected image B and in an iterative method (see...). Figure 3 The image shown is a comparison of the intermediate image ZV. As can be seen, the ray (solid arrow) passes through multiple voxels in its path to the image, so the image values ​​on the image do not necessarily match the image values ​​of the voxels (they usually do not match, thus requiring complex back projection).

[0180] The diagram can also show the intermediate image Z by reversing the arrow.

[0181] Now, during the shooting process, the cube moves along the dashed arrow to another position indicated by the dashed cube. After the movement (depending on the movement states MI, M), the voxel is located at the reference voxel position x. M Or the changed voxel position x', and imaged on the image at a completely different location (dashed arrow).

[0182] The movement shown here is a very simple one, in which all voxels remain within their original connections afterward. In reality, such movements are usually more complex, and each voxel can move independently, but in normal examination, there are always closed connections where the movements of adjacent voxels are very similar. For example, although a beating heart moves, its walls (fortunately) always form a closed body.

[0183] Figure 7 The diagram shows a repackaging of the projected image B (top), which is captured as an intermediate image Z (bottom) using a conical ray profile along a spiral trajectory (see the simplified diagram on the right). Based on calculations performed during repackaging, the movement state M of the intermediate image Z varies with the column (direction p) but not with the row (direction q). The center of the intermediate image Z contains the same movement state M as the projected image B before repackaging. A previous movement state M exists when moving right in the image, and a subsequent movement state M exists when moving left in the image. The exact position of each movement state M in the intermediate image Z must be calculated individually, as it depends on several parameters.

[0184] If you observe the simplified diagram in the lower right corner, it is clear that the middle image Z in this example does not have perfectly parallel geometry, but only semi-parallel geometry. The rays are only perfectly parallel when viewed along the p direction. This is not the case along the q direction. However, this is not a problem for the present invention, because the q coordinate is only needed for selecting image values ​​and can be easily determined.

[0185] Figure 8 The selection of the movement state M is shown, and the determination of the movement of the voxel position x' is sketched.

[0186] The most accurate motion compensation method uses the precise motion for each motion state of each intermediate image Z (i.e., at position p in each column image) for the backprojection step. However, this leads to storage issues in current implementations of reconstruction algorithms. A compromise can be used to achieve good results with less memory consumption.

[0187] Therefore, for multiple intermediate images Z, particularly for each intermediate image Z from the motion profile BP of the initial motion state MI, and for at least one earlier and at least one later motion state M, motion data is pre-selected from the motion profile BP in the back projection of the intermediate images Z. Here, in addition to the initial motion state MI with number 20, it also includes motion states M 27 and 13.

[0188] Therefore, for each intermediate image Z, three motion states MI, M are used: one in the center, one on the left, and one on the right. For example, the motion in the center is interpolated using linear interpolation of pre-selected motion data at positions 13, 20, and 27. However, it must be considered that the left and right motion states M do not necessarily have to be at the far left or far right, as these areas of the intermediate image Z may not be reachable depending on parameters such as the field of view. Taking these parameters into account, the optimal positions can be calculated individually.

[0189] Therefore, the movement state M (at p above the curly braces) can now be calculated, and the voxel position x′ after the change of voxel V can be calculated based on pre-selected movement data for movement states 27 and 13 and based on interpolated movement data. Here, proximity to adjacent movement states M can be considered in a weighted form.

[0190] Figure 9 Comparative results of the method according to the present invention are shown. The image shows a simulated scan of the clock phantom reconstructed using the WFBP algorithm.

[0191] On the left (0), an image of a stationary object that should be imaged as a reference, or its ideal reconstruction result after reconstruction, is shown. On the right (>0), an image of a moving object (during translation), or its reconstruction result without motion compensation, is shown. Severe artifacts caused by motion are visible. This imaging will be improved by the present invention.

[0192] The third image (SdT) illustrates the existing technique combined with the WFBP algorithm (according to...). Reconstruction using motion compensation methods (e.g., motion compensation methods). Although artifacts can be significantly reduced, some artifacts still exist because incorrect image information is assigned to voxels during reconstruction.

[0193] The reconstruction according to the method of the invention is shown in the right-hand image (E), where the method is applied only once (and without additional iterative steps). Motion artifacts are barely perceptible. It can be considered that the right-hand and left-hand images appear identical at first glance, but subtle differences become apparent upon close inspection (e.g., subtraction). In the case presented here, the movement of the object is 100% known. Even in the method according to the invention, more complex movements may produce slight artifacts; however, these are reduced by the number of iterations that can be achieved using the method.

[0194] Finally, it should be reiterated that the methods described in detail above and the computed tomography system 1 shown are merely embodiments, which can be modified in various ways by those skilled in the art without departing from the scope of the invention. Furthermore, the use of the indefinite articles "a" or "an" does not preclude the possibility that the described features may appear more than once. Similarly, the terms "unit" and "module" do not preclude the related component from being composed of a plurality of interacting sub-components that may also be spatially distributed. The word "a plurality" can be understood as "at least one".

Claims

1. A method for motion compensation in CT reconstruction, comprising the following steps: – Provides projection images (B) from CT scans. – Create an intermediate image (Z), wherein the intermediate image (Z) is created from the repacking of the columns of the projected image (B). – The layer image (S) is calculated by back-projecting the intermediate image (Z) onto a predetermined voxel (V) of the layer image (S), wherein the following steps are performed for each intermediate image (Z) and each voxel (V) based on a preset motion profile (BP) of the voxel (V) during the capture of the projected image (B): a) Select an initial movement state (MI) based on the intermediate image (Z). b) Calculate the reference voxel position (x) of the voxel (V) from the original voxel position (x) of the voxel (V) and the movement profile (BP) with respect to the initial movement state (MI). M ), c) Calculate the location in the intermediate image (Z) that will be mapped to the reference voxel position (x). M The column image position (p) of the voxel (V) at position ). d) Determine the changed movement state (M) of the voxel (V) based on the calculated column image position (p). e) Calculate the changed voxel position (x') of the voxel (V) from the original voxel position (x) of the voxel (V) and the movement profile (BP) with respect to the changed movement state (M). f) Calculate the changed image position (p', q') of the voxel (V) that will be mapped to the changed voxel position (x') in the intermediate image (Z). g) Use the image values ​​of the intermediate image (Z) at the changed image positions (p', q') to perform the back projection.

2. The method of claim 1, wherein the initial movement state (MI) corresponds to a column image position (p) that is substantially located at the center of the intermediate image (Z).

3. The method according to claim 2, wherein in the intermediate image (Z), the center corresponds to the column located exactly in the middle of a projection image of the projection image (P).

4. The method according to any one of claims 1 to 3, wherein in the back projection of the intermediate image (Z), for a plurality of the intermediate images (Z), for the initial motion state (MI) and for at least one earlier motion state and at least one later motion state (M), motion data is pre-selected from the motion profile (BP), motion data with respect to the changed motion state (M) is interpolated from these motion data, and the changed voxel position (x') of the voxel (V) is calculated based on the interpolated motion data with respect to the changed motion state (M).

5. The method according to any one of claims 1 to 3, wherein the following steps are traversed at least once between steps e) and f) before using the image values ​​used for the back projection: – Calculate the changed column image position (p') of the voxel (V) that will be mapped to the changed voxel position (x') in the intermediate image (Z). – Determine the changed movement state (M) of the voxel (V) based on the changed column image position (p'). – Calculate the changed voxel position (x') of the voxel (V) from the original voxel position (x) and the movement profile (BP) with respect to the changed movement state (M).

6. The method according to any one of claims 1 to 3, wherein when creating the intermediate image (Z), the columns of the projection image (B) are repacked such that the intermediate image (Z) uses those columns of the projection image (B) that are photographed with correspondingly parallel X-rays in a plane orthogonal to the columns.

7. The method according to any one of claims 1 to 3, wherein the intermediate image (Z) is filtered by means of convolution and / or Fourier transform to obtain a better reconstruction result, wherein in the convolution, the filtered intermediate image (Z) is computed by means of a predefined kernel, and / or the intermediate image (Z) is subjected to Fourier transform and multiplied with an adaptive filter in Fourier space.

8. The method according to any one of claims 1 to 3, wherein after calculating the layer image (S) from the intermediate image (Z), the following steps are performed: i) Reconstruct the comparison intermediate image (ZV) from the layer image based on the moving profile (BP). ii) Compare the intermediate image (ZV) with the intermediate image (Z). iii) Create a modified layer image (S) based on the comparison; iv) Repeat steps i) to iii) iteratively using the corresponding previously created layer image (S).

9. The method of claim 8, wherein when reconstructing the intermediate comparison image (ZV) for each image point of each intermediate comparison image (ZV), the image points of the intermediate comparison image (ZV) are determined according to the following steps: – Calculate the ray that starts from an image point in the intermediate image (ZV) and passes through an image voxel (VB). – Based on the motion profile (BP), motion data for determining the motion state (M) is obtained according to the column image position (p) in the comparison intermediate image (ZV). – The positions of the ray and image voxels (VB) are shifted relative to each other based on the movement data in the movement profile (BP) regarding the movement state (M). – As the ray and the image voxel (VB) shift relative to each other, the value of the image voxel (VB) is accumulated along the ray. – The accumulated values ​​at positions (p, q) in the intermediate comparison image (ZV) are used.

10. The method of claim 1, wherein the initial movement state (MI) corresponds to a column image position (p) substantially located at the center of the intermediate image (Z), and such an initial movement state (MI) is selected for each intermediate image (Z).

11. The method of claim 4, wherein the interpolation is based on the movement state (M) that is closest to the changed movement state (M).

12. The method of claim 4, wherein the distances of the image locations of the plurality of said movement states (M) are considered in a weighted manner.

13. The method of claim 9, wherein the ray is moved according to the movement data.

14. A system (9) for motion compensation in CT reconstruction, designed to perform the method according to any one of claims 1 to 13, said system (9) comprising: – A data interface (6) is designed to receive projection images (B) from CT scans. – A repacking unit (7) is designed to create an intermediate image (Z), wherein the intermediate image (Z) is created from the repacking of columns of the projected image (B). – A reconstruction unit (8), designed to compute the layer image (S) by back-projecting the intermediate image (Z) onto a predetermined voxel (V) of the layer image (S), comprising: – A movement module (30) is designed to select an initial movement state (MI) of an intermediate image (Z), or to determine the changed movement state (M) of the voxel (V) based on the calculated column image position (p, p') of the voxel (V) in the intermediate image (Z). – A positioning module (31) is designed to calculate the voxel position (x', x') of the voxel (V) from the original voxel position (x) of the voxel and the movement profile (BP) with respect to the selected or calculated movement state (MI, M). M ), – An imaging module (32) is designed to calculate the column image positions (p, p') and row image positions (q, q') that will map voxels (V) in the intermediate image (Z). – A reuse module (33) is designed to reuse the image values ​​of the intermediate image (Z) at the changed image positions (p', q') for the back projection.

15. The system of claim 14, wherein the reconstruction unit (8) additionally comprises: – A ray simulation module (34) is designed to calculate a ray that starts from an image point and travels through an image voxel (VB). – A motion simulation module (35) is designed to determine motion data of motion state (M) based on the motion profile (BP) by comparing column image positions (p) in the intermediate image (ZV), and to shift the positions of the ray and image voxels (VB) relative to each other based on the motion data of the motion profile (BP) regarding the motion state (M). – A simulation carry-over module (36) is designed to accumulate the value of the image voxel (VB) along the ray as the ray and the image voxel (VB) shift relative to each other, and carry over the accumulated value at position (p, q) in the comparison intermediate image (ZV).

16. A control device (10) for controlling a computed tomography system (1), comprising the system according to claim 14 or 15.

17. A computed tomography system, comprising the control device (10) according to claim 16.

18. A computer program product having a computer program that can be directly loaded into a storage device of a control device (10) of a computed tomography system (1), the computer program having program segments for performing all steps of the method according to any one of claims 1 to 13 when the computer program is run in the control device (10) of the computed tomography system (1).

19. A computer-readable medium storing a program segment that is readable and operable by an arithmetic unit, such that when the program segment is run by the arithmetic unit, all steps of the method according to any one of claims 1 to 13 are performed.

Images