Computer-implemented method and evaluation device for evaluating a set of projection images, computer program and electronically readable data carrier

Abstract

The invention relates to a computer-implemented method for evaluating a set of projection images of an X-ray device (1) in maskless angiography in an examination area with a blood vessel structure, comprising the following steps - Providing the recorded set of projection images showing the passage of a contrast agent through the blood vessel structure in a covered passage time period in a number of projection images, - Selection of a reference image from the projection images, - Determining a weight for each projection image, whereby the weight for the projection images not corresponding to the reference image is determined depending on a similarity of the projection image to the reference image and the weight for the reference image is not lower than the maximum of the weights for the other projection images, - Determination of at least one evaluation image from the projection images, wherein the projection images are included in the determination according to the determined weights.

Description

The invention relates to a computer-implemented method and an evaluation device for evaluating a projection image set of an X-ray device during maskless angiography in an examination region having a blood vessel structure. In addition, the invention relates to a computer program and an electronically readable data carrier.The accurate visualization of blood vessel structures and, in some cases, quantification of their static parameters and flow parameters is important for a variety of medical applications. Angiography imaging to represent such blood vessel structures, for example, vessel trees, usually uses, especially in the case of X-ray imaging, a contrast agent that is introduced into the patient's blood vessel system and then transported through the blood vessel structure. In order to be able to obtain evaluation images of only the blood vessel structure from a projection image set, i.e. two-dimensional X-ray images, as far as possible, digital subtraction angiography has been proposed. Therein, at least one mask image without contrast medium is first recorded, which shows the anatomical background, and is subtracted from the filling images with contrast medium. 3D-DSA has also already been proposed. However, the acquisition of mask images has some disadvantages, for example possible noise amplification, additional dose and possible movement of the patient between the acquisition with and without contrast medium.Maskless angiography ("mask angiography"), also known as kinetic or temporal-dynamic imaging, uses the temporal dynamics provided by the administered contrast agent in order to derive evaluation images of the vascular structure without anatomical background. Since these approaches do not require subtraction of a mask image, a higher signal-to-noise ratio can be achieved therewith, the radiation dose can be reduced and less contrast medium can be used. Examples of maskless angiography, which thus derives evaluation images directly from a projection image set without a dedicated mask image, include digital variance angiography (DVA) and virtual mask angiography (VMA).The DVA is described, for example, in an article by Victor Imre Orias et al., "Digital variance angiography as a paradigm shift in carbon dioxide angiography", Investigative Radiology 54.7 (2019), pages 428-436, with further references. It is proposed therein to record a plurality of, usually under-exposed, projection images and to derive evaluation images by statistical analysis, and therefore statistical images whose image values contain, for example, the standard deviation, the variance and other time-derived parameters of the X-ray attenuation for each pixel. Statistical images using standard deviation and variance as statistics and thus image values contain functional motion related information that allows for an improvement in visualization of blood vessel structures. A DVA method is further described in EP 2 628 146 A2.DE 10 2022 209 890 A1 describes a VMA method, specifically a method for generating a virtual mask image. A plurality of projection images of an object are acquired by means of an angiography apparatus. For at least a part of all pixel positions, an extreme pixel value of the pixels of the plurality of projection images in the respective pixel position is determined in each case, in particular by maximum intensity projection (MIP) over time in the case of iodine as contrast medium. Finally, the virtual mask image is created from the extreme pixel values at the respective pixel positions. The virtual mask image consequently has a respective absolute or statistically formed minimum or maximum value of a part or all pixel values of all projection images at this respective pixel position at each pixel position. It can be provided that a noise-reduced virtual mask image is obtained by summing at least two (optionally all) of the plurality of images including the virtual mask image pixel by pixel with a basic weighting which more strongly emphasizes the extreme pixel value.However, it is problematic in such approaches that patient movement also introduces temporal dynamics and can cause artifacts in the resulting evaluation images. At present, no specific procedures for reducing patient movement for maskless angiography, in particular DVA and VMA, are known. Known motion correction methods based on 2D-2D registration of the projection images with one another could be used. However, such two-dimensional approaches have the disadvantage of having limitations on motion in multiple layers and non-rigid 3D motion. In other words, motion patterns that do not correspond to single-slice translation motion within the image plane result in residual motion artifacts.The object of the invention is therefore to specify a possibility for improving the image quality of evaluation images of maskless angiography, in particular with regard to patient movement.To achieve this object, a computer-implemented method, an evaluation device, a computer program and an electronically readable data carrier according to the independent patent claims are provided according to the invention. Advantageous further developments are evident from the dependent claims.In a computer-implemented method according to the invention for evaluating a projection image set of an X-ray device during maskless angiography in an examination region having a blood vessel structure, the following steps are provided according to the invention:providing the recorded projection image set which shows the passage of a contrast agent through the blood vessel structure in a covered passage period in a projection image number of projection images,selecting a reference image from the projection images,determining a weight for each projection image, wherein the weight for the projection images not corresponding to the reference image is determined as a function of a similarity of the projection image to the reference image and the weight for the reference image is determined not to be lower than the maximum of the weights for the other projection images,determining at least one evaluation image from the projection images, wherein the projection images enter into the determination according to the determined weights.At the beginning of the method, a plurality of projection images, in particular radiographs, of a projection image set, for example 20 to 40 projection images, are therefore provided, which have already been recorded beforehand by the examination region. In this case, a contrast agent was administered in particular to the patient, so that the projection images recorded over a transit time period of the contrast agent show the temporal dynamics of the contrast agent flooding, the filling and the outflow over the transit time period in terms of their temporal progression. However, the projection images can also show movements of the patient over time in the examination region. The projection images are two-dimensional x-ray images that have been recorded by means of an x-ray device, in particular an angiography device.From this set of projection images, a reference image is now selected which represents a reference movement state. As will be explained in more detail, the reference image is in particular such a projection image in which the vascular structure can be seen well, since it is actually the case of the most good possible extraction thereof and thus a good reference is given. Firstly, with respect to the remaining, other projection images, weights are now determined which ultimately indicate how well the respective other projection image matches the reference image, for example as a similarity of the movement states. If these weights, which can also be referred to as motion weights, are then taken into account in the determination of the evaluation image, portions of similar motion states are therefore emphasized in particular and artifacts due to differences in the motion states are thus suppressed.This is based on the idea that although both the contrast medium flow and movements of the patient are time-dynamic effects which are reflected in the projection images, the contrast medium change, since they mainly relate to the intensity, are weaker effects which thus enter the background compared to actual movements of the patient which result in hard edge displacements. In other words, the similarity of the images, in particular with regard to the movement states, clearly predominates in the actual movements of the patient, so that the weights actually indicate which movement state deviations are present. Nevertheless, it should be noted that, in order to increase this dominance of the actual movement (against changes in the contrast medium concentration), it is also conceivable in exemplary embodiments that image points lying outside a vessel mask which indicates where the vessel structure is imaged in the projection images, that is to say pixels, of the projection images are used for determining the weights. A vessel mask, which for better buffering of movements can also comprise a tolerance range around an actually determined vessel course, can be determined, for example, by means of a so-called vesselness filter, which will be discussed in more detail below.Generally speaking, the present invention thus introduces a weighting scheme for reducing motion artifacts in maskless angiography imaging. In this way, the influence of the temporal dynamics, which arises as a result of patient movement, can be significantly reduced overall. In particular, a reduction of motion artifacts is possible, so that the image quality is significantly increased.It should be noted here that weighting schemes for reducing motion artifacts have already been proposed in classical digital subtraction angiography for a multiplicity of recorded mask images. The present invention allows the advantages of such motion artifact reduction to be combined with those of maskless angiography, particularly higher signal-to-noise ratio and reduced dose. In particular, the realization of the dominance of actual patient movement with regard to temporal dynamics when the similarity of projection images is considered is a substantial ingredient which only allows the technique to be transferred on the basis of contrast-agent-free combination techniques for ascertaining mask images.Depending on the approach specifically used, for example VMA or DVA, different types of evaluation images can be determined. It is particularly expedient if the evaluation image or at least one of the evaluation images is a blood vessel structure image with at least reduced background features. Both VMA and DVA basically allow comparable evaluation images to be obtained with digital subtraction angiography, in which the anatomical (and possibly other) background is at least substantially subtracted out, which therefore essentially only show the vascular structure or the contrast medium located therein and / or the dynamics thereof. Other types of evaluation images, for example those which reproduce the dynamics by color coding, can also be significantly improved in their quality by the procedure according to the invention.A large number of conceivable possibilities exist for ascertaining the reference image. Thus, practical refinements of the present invention provide that the projection image of maximum contrast medium filling, in particular determined from an average value and / or histogram of the image values of the projection images and / or on the basis of a user input, and / or the reference image is determined by applying a trained selection function, as the reference image. The maximum contrast medium filling means that the largest possible proportion of the vascular structure is actually clearly visible in the reference image. After angiography aims at displaying the vascular structure, it is particularly expedient to use one as a reference image in which it can be recognized particularly completely, which is the case with maximum contrast medium filling. While it is conceivable in principle to select the reference image on the basis of a user input, automatic selection by a selection function, which can be trained, that is to say can be based on machine learning, is also possible. In the case of automatic selection by a selection function, the selection function can, for example, evaluate mean values and / or histograms of the image values and / or comprise a so-called vesselness filter which ultimately provides information about which feature represented in the image is vessels (after these have a particularly high "vesselness"). For example, the projection image can be found which shows the most vessels, which corresponds substantially to the maximum contrast medium filling. If artificial intelligence, i.e. a trained selection function, is used, it can implement the Vestesy filter, but can also be trained in another way, for example based on a user selection from different projection image sets.With regard to possible vestness filters, reference may be made by way of example to the article by Alejandro F. Frang et al, "Multiscale Vest Enhancement Filtering", Medical Image Computing and Computer-Assisted Intervention-MICCAl'98, MICCAl 1998. Lecture Notes in Computer Science, Vol. 1496, pages 130-137, in which reference is also made to other conceivable approaches in addition to the multiscale approach described therein.In a particularly advantageous embodiment of the present invention, it can be provided that the weights are determined in an optimization process for minimizing a deviation function for deviation images, which are calculated by subtracting a sum of the other projection images not corresponding to the reference image, determined using a test set of weights, from the reference image. In this case, the aim is ultimately to simulate the reference image from the other projection images, with just those which have the greatest similarities therewith contributing most. Thus, a natural way is given to describe the similarity in relation to the state of motion, which is also easily implementable algorithmically on account of the many known techniques for solving optimization problems. High quality weights can be derived.Here, it is particularly preferable that the deviation function in the form of a convex square program of a dimension corresponding to the projection image number minus one is minimized. It is proposed to convert the optimization problem into a quadratic program which allows the computation effort to be reduced by a factor of 100 to 1000 by reducing the dimensionality to the projection image number of less than one. At the same time, a robust determination of optimum weights is also possible. Thus, there is a highly efficient way to solve the weighting problem. The optimization problem is transformed into a low-dimensional quadratic program, which can be solved with less computing effort. There are already freely available approaches for quadratic programs, so that the effort for the specific implementation is also significantly reduced. In particular, it is avoided to create a general algorithm which is efficient with respect to computing time for solving non-linear, limited problems. Experiments have shown that the approach according to the invention allows weights of the other projection images to be determined in a robust manner in a very low, real-time-compatible calculation type, for example on laptop computers as computing devices.In a particularly preferred, specific embodiment of the present invention, an, in particular weighted, L2normal is used as the deviation function. This L2 norm of the respective deviation images for the test sets is to be minimized in the optimization process. Conveniently, the quadratic program may be formulated by multiplying the subtraction terms for the deviation images and separating the quadratic and linear terms into a vector of the weights.In the following, let P be the number of pixels in the projection images D, the reference image R and the deviation image, and let N be the number of projection images (number of projection images), so that N-1 other projection images D i exist next to the reference image R. Let d, r and a be the vectored projection, reference and deviation images (with P real valued entries, the image values, respectively). The weight for the ith other projection image Di is α i, each weight being in an interval of 0 to 1, and the sum of the weights of the other projection images is 1. Then, the optimization problem can be written asLet M now furthermore be a P x (N-1) matrix which contains the vectorized other projection images d i as columns, M = (d 1... d N-1), and α be the vector of the mask weights α i with (N-1) entries. The following are: and, finally, the quadratic programThis quadratic program comprises only matrices and vectors of dimensionality N-1, i.e. the number of other projection images. No evaluation of the original projection images of dimming significance P is required during the optimization process, wherein N-1 is very much smaller than P, i.e. N-1<<P. The complete projection images need only be processed if Q and c are predicted. By design, Q is positive (semi-)definite, so the optimization problem is convex.As Equation (4) shows, both Q and c can be expressed as scalar products of the vectorized projection images d i, r of the projection image set: for the quadratic term and for the linear term.In words, it can thus be said that, for forming the quadratic matrix of the quadratic term of the deviation function in the form of the quadratic program, dot products of the vectorized other projection images are expediently formed with one another and, for forming the vector of the linear term of the deviation function in the form of the quadratic program, dot products of the vectorized reference image are expediently formed with the vectorized other projection images.For such convex quadratic optimization problems, there are already special algorithms in the prior art that can be executed more quickly than general algorithms for limited nonlinear optimization. In an expedient embodiment of the present invention, it can be provided that the optimization process takes place by means of a gold color Idani method. In particular, a "gold-color idnanie active-set dual method" can be used, which is available, for example, in tool boxes called QuadProg. The gold color Idnani method is basically described in an article by D. Goldcolor and A. Idnani, "A numerically stable dual method for solving statistically convex quadratic programs", Mathematical Programming 27.1 (1983): 1-33. For positive semidefinite problem positions, reference is also made to the article by N. L. Boland, "A dual-active-set algorithm for positive semidefinite quadratic programming", Mathematical Programming 78 (1996): 1-27.While it is preferred according to the invention to use an optimization process, in particular with respect to a quadratic program, it is also conceivable in principle within the scope of the present invention to determine the weights in another way. Thus, for example, it can be provided that the weights are determined at least partially depending on a correlation measure and / or comparison measure of the other projection images with respect to the reference image. In this case, it is possible in principle to use known correlation measures and / or comparison measures which are based on the comparison of different images. For example, it is then conceivable to determine the weights by normalizing the corresponding dimensions to one.Finally, it is also necessary to determine a weight for the reference image itself. Since the reference image is naturally very similar, according to the invention, however, no excessively high values for the weighting are sought, since, despite the weighting, the image contents of many projection images should be incorporated as significantly as possible during the determination of the evaluation image. Thus, a particularly advantageous development of the present invention provides that the weight for the reference image is determined as the maximum of the weights for the other projection images. In the above-mentioned formalism, the weights w i for all projection images can therefore result as where N R is the index of the reference image.It should also be noted here that the weights as a whole, i.e. also taking into account the weight of the reference image, do not necessarily have to be normalized to one in their sum, but it is quite conceivable and expedient if the sum of the weights is renormalized to one after addition of the weight for the reference image.Depending on how the at least one evaluation image is determined, the weights can be applied differently in order to make the projection images of the projection image set contribute differently. In general, it can be provided that the weights are used in the formation of at least one weighted sum and / or in a selection of projection images for a determination step in the determination of the evaluation image, wherein in particular only projection images lying above a limit value for the respective weight are selected. If a sum is formed, for example in the case of pixel-by-pixel averaging or other combinations of projection images, the weights can be introduced particularly easily in the sense of a weighted sum, as is known in principle. However, they can also be used in other cases, for example when selecting specific projection images for a determination step in which, for example, all projection images whose weight above a limit value are selected. Specifically, it can be provided that the limit value is determined as 0.5 divided by the projection image number, i.e., as a formulaIn such a selection, it has been shown that a sufficient number of projection images is usually selected. If, however, the limit value ensures that too few images are selected, various specific approaches are conceivable within the scope of the present invention.Thus, it can be provided that if, on the basis of a predetermined limit value, less than a minimum number of projection images, in particular 10 to 30% of the projection images, were selected, either the limit value is lowered until the minimum number of projection images is selected or the determination of the weights is repeated, wherein a regularization technique is used to broaden the distribution of the determined weights. It is therefore possible first to lower the limit value, for example by lowering μ in equation (9), in order to ensure the selection of at least the minimum number of projection images.In a particularly advantageous alternative for this case, however, provision is made to use a regularization technique. In this case, it is possible in principle to use known regularization techniques, for example the Tikhonov regularization known from linear regression or variants thereof. Specifically, in the minimization in the form of a quadratic program as described above, the present invention may provide that in the minimization of the deviation function in the form of a convex quadratic program, the unit matrix multiplied by a regularization factor is added to the matrix of the quadratic term.In this case, the optimization problem, compare formula (5), can thus be written asThe parameter (regularization factor) λ≅ ∈ R + can be selected empirically, for example, or else can be increased until the minimum number is selected.Some specific application cases, in particular with respect to specific types of evaluation images, are explained in more detail below.Thus, it can be provided in concrete terms that a statistical image of a digital variance angiography (DVA) is determined as at least one of the at least one evaluation image, wherein the weights are applied both in the pixel-by-pixel formation of an average value over time and in the pixel-by-pixel determination of a statistical value used as an image value of the statistical image, in particular a variance and / or a standard deviation over the run-through period. Such a variant of DVA can also be referred to as DWVA ("digital weighted variance angiography"). Expressed in formulas, a DWVA image can be determined pixel by pixel asFor the application in the VMA, it can be provided that the determination of at least one of the at least one evaluation image is carried out using a virtual mask image, wherein the weights are used in the selection of projection images for an extreme value projection over time and / or, as an additional weighting to a base weighting, in the determination of a noise-reduced virtual mask image and / or, in the determination of the evaluation image, as a weighted sum of subtraction images of the projection images and of the virtual mask image.Thus, first of all, when determining the original mask by extreme value temporal projection, for example maximum value temporal projection (MIP) when using iodine as contrast medium, a part of the projection images can be excluded by selecting only projection images with a movement phase close to the reference image, wherein 0.5 divided by the number of projection images can be used as the limit value for selection, for example, as described above. If a noise-reduced mask is determined, as likewise described in DE 10 2022 209 890 A1, the motion weights can add multiplicatively to the basic weight in the case of the weighted sum as an additional weight. Independently of the determination of a noise-reduced mask image, even if the VMA evaluation image is determined as the sum of subtraction images, a weighted sum can likewise be formed using the movement weights. The inclusion of the weights when using virtual mask images can also be referred to as WVMA ("weighed virtual mask angiography").The movement weights can also be used in a different context, for example with regard to evaluation images to be displayed, which are intended to visualize facts, for improving the image quality. Thus, for example, it can be provided that a flow information image is determined as at least one of the at least one evaluation image, in which, in particular pixel by pixel, regions are colored in this region depending on the time of arrival of the contrast agent, wherein projection images whose weight is below a limit value are excluded for generating the flow information image. Such display techniques are known, for example, by the term "iFlow". By excluding projection images which differ greatly with regard to the state of movement, a better image quality can also be achieved here. It should be noted here that such flow information images can of course also be determined on the basis of the last-described DWVA images and / or WVMA images.In a particularly preferred embodiment of the present invention, it can be provided that a movement correction process is carried out on the projection images before the determination of the weighting, in particular comprising a rigid or affine 2D-2D registration of the projection images with respect to one another for the determination of correction factors. In this way, a further improvement in the image quality can be achieved, on the one hand, since movement artifacts are already reduced in such a step and, on the other hand, the number of projection images with comparable movement states is increased. It is particularly advantageous if the 2D-2D registration takes place at least between the reference image and the other projection images, after a part of the movement differences between the reference image and the other projection images is then already omitted and a greater similarity is given.In an advantageous development, it can be provided that a plurality of reference images are used and the at least one evaluation image is determined for each reference image, wherein in a selection process one of the evaluation images is selected as a final evaluation image depending on a trained evaluation function and / or an image quality measure, in particular a maximum vesselness, and / or user selection information. Instead of a single reference frame, weights for multiple reference images may also be determined. This is possible in particular in the case of a solution of the optimization problem in the form of a quadratic program in a computationally efficient manner, since only entries in Q and c have to be replaced and, in addition, the quadratic program can be solved with little computational effort. For example, different DWVA images and / or WVMA images may be determined and the best of them selected. This selection can be made manually by the user, but also automatically, for example by means of a trained evaluation function. Such a trained assessment function can work, for example, on the basis of image quality measures, such as the maximum vesselness, and / or on the basis of annotations.In addition to the method, the invention also relates to an evaluation device for evaluating a projection image set of an X-ray device during maskless angiography in an examination region having a blood vessel structure, wherein the evaluation device comprises:an input interface for receiving the recorded projection image set, which interface shows the passage of a contrast medium through the blood vessel structure in a covered passage time period in a projection image number of projection images,a selection unit for selecting a reference image from the projection images,a determination unit for determining a weight for each projection image, wherein the weight for the projection images not corresponding to the reference image is determined as a function of a similarity of the projection image to the reference image and the weight for the reference image is determined not to be lower than the maximum of the weights for the other projection images, anda determination unit for determining at least one evaluation image from the projection images, wherein the projection images enter into the determination according to the determined weights.All embodiments with respect to the method according to the invention can be transferred analogously to the evaluation device according to the invention and vice versa, so that the already mentioned advantages can also be obtained with the evaluation device.The evaluation device can have at least one processor and at least one storage means. Functional units can be formed by hardware and / or software in order to carry out steps of the method according to the invention. In addition to the aforementioned functional units-selection unit, determination unit and determination unit-there are naturally also further functional units for other steps, for example a correction unit for motion correction by means of 2D-2D registration, and / or subunits, for example an optimization subunit for minimizing the deviation function in the optimization process, provided as a subunit of the determination unit. The evaluation device can also have an output interface via which the at least one determined evaluation image can be output.The evaluation device can be particularly advantageously part of a control device of the X-ray device with which the projection image set is recorded. An X-ray device comprising an evaluation device according to the invention, in particular as part of a control device of the X-ray device, is therefore conceivable. In this case, the interfaces may be internal interfaces.The X-ray device can preferably be an angiography device and / or an X-ray device with a C-arm, on which an X-ray detector and an X-ray radiator are arranged opposite one another. Such C-arm X-ray devices are frequently used in angiography laboratories and have the advantage that they allow different projection directions of the projection images to be easily adjusted.A computer program according to the invention can be loaded directly into a storage means of an evaluation device and has program means such that, when the computer program is executed, the evaluation device is caused to carry out the steps of a method according to the invention. The computer program can be stored on an electronically readable data carrier according to the invention, which therefore comprises control information stored thereon, which comprises at least one computer program according to the invention and are configured such that, when the data carrier is used in an evaluation device, said data carrier is configured to carry out the steps of a method according to the invention. The data carrier can be a non-transient data carrier, for example a CD-ROM.Further advantages and details of the present invention are evident from the exemplary embodiments described below and on the basis of the drawings. The following are shown: FIG. 1 shows a flow chart of an exemplary embodiment of the method according to the invention, FIG. 2 shows a schematic diagram of an X-ray device, and FIG. 3 shows the functional structure of a control device of the X-ray device.FIG. 1 shows a flow chart of an exemplary embodiment of the method according to the invention. In this case, in a step S 1, a projection image set is provided which comprises a number of projection images, referred to as a number of projection images, of an examination region of a patient in whom a contrast medium has flowed through a blood vessel structure in a passage period of an angiography imaging process. In the present case, no mask image is recorded, that is to say maskless angiography is carried out. For example, they can be 20 to 40 projection images.In a step S 2, a reference image is selected from the projection images of the projection image set. In the present case, the projection image with the maximum contrast medium filling is selected as the reference image, which can be done manually by a user, but preferably is done by a selection function, in particular a trained selection function, which implements a Vestesy filter and / or evaluates mean intensities (image values) and / or histograms thereof, so that the projection image with most image portions identified as vessels can be selected.In an optional step S 3, a first motion correction is carried out in that a rigid or affine 2D-2D registration is carried out between the other projection images and the reference image in order to reduce the differences in the motion state between these images and already to carry out a first step for avoiding motion artifacts. It should be noted that it is also conceivable in principle to carry out this correction step already before the selection of the reference image, wherein the projection images are then likewise all shifted towards a reference movement state by the 2D-2D registration.In a step S 4, weights for the projection images of the projection image set are determined based on the similarity of the other projection images to the reference image. First, the weights for the other projection images, i.e., all projection images except the reference image, are obtained. In this case, an optimization process is used to minimize a deviation function for deviation images which are calculated by subtracting from the reference image a sum of the other projection images which do not correspond to the reference image and which is determined using a test set of weights. The test set of weights is determined as motion weights, which minimizes the value of the deviation function.Here, the optimization takes place based on the quadratic program according to the formula (5). Q and c (formula (4)) are calculated in advance for this purpose by evaluating the scalar products according to equations (6) and (7). The actual optimization is then carried out by a gold color Idani method. For test sets of motion weights, deviation images are determined and evaluated as a deviation measure by the L2 norm.After the determination of the weights for the other projection images, according to equations (8) a weight for the reference image is also added, in the present case the maximum of the weights for the other projection images.In a step S 5, at least one evaluation image is determined using the weights determined in step S 4. In this case, the weights can be used both when selecting projection images for a determination step in such a way that only projection images whose weight is above a limit value are selected (compare formula (9)) and in the case of summing in such a way that a weighted sum is formed.If too few projection images, in particular less than a minimum number (for example 5 or 6) of projection images, are selected when using a limit value, step S 4 is repeated in the present case in such a way that a broader distribution of the weights results. To this end, according to equation (10), a regularization term is added in the quadratic component of the optimization problem. This is possible without problems, since the optimization process can be carried out in a computationally efficient manner. Alternatively, it is also possible to reduce the limit value, for example by reducing.mu., until the minimum number is reached.In step S 5, as at least one of the at least one evaluation image, for example, a DWVA image (compare, for example, equation (11)) or a WVMA image can be determined, wherein in the latter, an initial mask image is first determined by extreme value projection of projection images selected on the basis of the limit value and, thereafter, a noise-reduced virtual mask image is determined by weighted summation of the mask image and the projection images. The weighting is composed of a basic weighting, which weights the mask image much more strongly, and a motion weighting defined by the motion weights determined in step S 4 as an additional weighting. Finally, the WVMA image is determined by weighted summation of respective subtraction images formed from the projection images and the noise-reduced mask image, again using the weights from step S 4.Both of these evaluation images can be the basis for a flow information image in which flow information is shown in a color-coded manner and projection images to be used for ascertaining the flow information can in turn be selected on the basis of the weights on the basis of the or a further limit value.Finally, the at least one determined evaluation image can be output.Although the embodiment is described for a single reference image, it is also possible to select a plurality of reference images and determine weights and evaluation images for each reference image. The best one can be selected from these evaluation images, for example manually, by means of a judgment function and / or on the basis of an image quality measure.FIG. 2 shows a schematic diagram of an X-ray device 1, which in the present case has a C-arm 2 with an X-ray radiator 3 and an X-ray detector 4, which are situated opposite one another. The C-arm 2 held on a stand 5 is movable, so that different projection geometries can be set with respect to a patient arranged on a patient couch 6.The operation of the X-ray device 1 is controlled by means of a control device 7 which comprises an evaluation device according to the invention and whose functional structure is to be explained in more detail with reference to FIG. 3. The control device 7 comprises a storage means 8 which is also used by the evaluation device 9 according to the invention.A recording unit 10 of the control device 7 controls the recording operation, in particular also the projection images of the projection image set, and can provide this via an internal input interface 11 of the evaluation device 9 to a selection unit 12 for selecting the reference image according to step S 2. In an optional correction unit 13, movement correction can be carried out by 2D-2D registration according to optional step S 3. A determination unit 14 is configured to determine the weights according to step S 4 and accordingly also comprises an optimization subunit 15. In a determination unit 16, the determination of the at least one evaluation image according to step S 5 is possible.The at least one determined evaluation image can be provided via an output interface 17 of the evaluation device 9, for example to a display unit 18 of the control device 7, and can be output on a display device of the X-ray device 1. Of course, determined evaluation images can also be stored in the storage means 8 or otherwise forwarded / held.In the foregoing description, irrespective of the grammatical sex of a certain term, persons having male, female or other sex identity are intended to be included.

Claims

Computer-implemented method for evaluating a projection image set of an X-ray device (1) during maskless angiography in an examination region having a blood vessel structure, comprising the following steps - providing the recorded projection image set which shows the passage of a contrast medium through the blood vessel structure in a covered passage period in a projection image number of projection images, - selecting a reference image from the projection images, - determining a weight for each projection image, wherein the weight for the projection images not corresponding to the reference image is determined as a function of a similarity of the projection image to the reference image and the weight for the reference image is determined not to be lower than the maximum of the weights for the other projection images, - determining at least one evaluation image from the projection images, wherein the projection images enter into the determination according to the determined weights.Method according to Claim 1, characterized in that the projection image of maximum contrast medium filling, in particular determined from an average value and / or histogram of the image values of the projection images and / or on the basis of a user input, is selected as the reference image and / or the reference image is determined by application of a trained selection function.Method according to Claim 1 or 2, characterized in that the weights are determined in an optimization process for minimizing a deviation function for deviation images which are calculated by subtracting from the reference image a sum of the other projection images which do not correspond to the reference image and which is determined using a test set of weights.A method according to claim 3, characterized in that the deviation function is minimized in the form of a convex square program of a dimension corresponding to the projection image number minus one.Method according to Claim 4, characterized in that an L2 standard is used as the deviation function.The method of claim 5, characterized in that the quadratic program is formulated by multiplying the subtraction terms and separating the quadratic and linear terms into a vector of the weights.Method according to Claim 5 or 6, characterized in that, in order to form the quadratic matrix of the quadratic term of the deviation function in the form of the quadratic program, scalar products of the vectorized other projection images are formed with one another and, in order to form the vector of the linear term of the deviation function in the form of the quadratic program, scalar products of the vectorized reference image are formed with the vectorized other projection images.Method according to one of Claims 4 to 7, characterized in that the optimization process takes place by means of a gold-colour Idnani method.Method according to one of the preceding claims, characterized in that the weights are used in the formation of at least one weighted sum and / or in a selection of projection images for a determination step in the determination of the evaluation image, wherein in particular only projection images lying above a limit value for the respective weight are selected.Method according to claim 9, characterised in that if, on the basis of a predetermined limit value, less than a minimum number of projection images, in particular 10 to 30% of the projection images, were selected, either the limit value is lowered until the minimum number of projection images is selected or the determination of the weights is repeated, wherein a regularisation technique is used to broaden the distribution of the determined weights.Method according to one of the preceding claims, characterized in that before the determination of the weighting a movement correction process is carried out on the projection images, in particular comprising a rigid or affine 2D-2D registration of the projection images with respect to one another for the determination of correction factors.Method according to one of the preceding claims, characterized in that a plurality of reference images are used and the at least one evaluation image is determined for each reference image, wherein in a selection process one of the evaluation images is selected as a final evaluation image as a function of a trained evaluation function and / or an image quality measure, in particular a maximum vesselness, and / or user selection information.Evaluation device (9) for evaluating a projection image set of an X-ray device (1) during maskless angiography in an examination region having a blood vessel structure, wherein the evaluation device (9) comprises: - an input interface (11), configured to receive the recorded projection image set, which shows the passage of a contrast agent through the blood vessel structure in a covered passage period in a projection image number of projection images, - a selection unit (12), configured to select a reference image from the projection images, - a determination unit (14), configured to determine a weight for each projection image, wherein the weight for the projection images not corresponding to the reference image is determined as a function of a similarity of the projection image to the reference image and the weight for the reference image is determined not to be lower than the maximum of the weights for the other projection images, and a determination unit (16), configured to determine at least one evaluation image from the projection images, wherein the projection images enter into the determination according to the determined weights.Computer program which, when executed on an evaluation device (9), causes it to carry out the steps of a method according to one of Claims 1 to 12.Electronically readable data carrier on which a computer program according to claim 14 is stored.

Images