Ensuring quality of different contrast agent events in 3D dual energy x-ray imaging

Abstract

A system (SYS) for image processing and related methods. The system comprises an input interface (IN) for receiving a volumetric image reconstructed from spectrally processed projection data acquired by a spectral imaging device (IA) during acquisition when a contrast agent is present in a field of view of the spectral imaging device (IA). A divider (SC) segments the volumetric image for the contrast agent contribution to obtain a segmented volumetric image. A forward projector (FP) forwards projects the segmented volume image onto the spectrally processed projection data. A quality checker (QC) establishes quality indicator data indicative of a mismatch between the forward projected segmented volumetric image and the spectrally processed projection data. The output data may be used to assess the degree to which the reconstructed data is composed of artifacts.

Description

Technical Field

[0001] The present invention relates to a system, an imaging arrangement, a related method, a computer program unit, and a computer-readable medium for medical image processing. Background Art

[0002] As a non-invasive first-line diagnostic tool for suspected coronary artery disease, coronary CTA (Computed Tomography Angiography) imaging is becoming increasingly important. Specifically, CT perfusion or CT-derived IMR assessment has recently gained the most interest as these methods can support the diagnosis of microvascular dysfunction, thus addressing long-standing problems.

[0003] Such imaging setups use contrast agents to enhance the contrast of structures that are otherwise nearly transparent to X-radiation. Dual-layer detectors for C-arm systems may also be used in the near future to enable interventional spectral imaging.

[0004] Projection data acquired using a dual-layer detector (or any other technique that allows for energy-resolved projection data to be collected from the same view) enables material decomposition on the projection. If iodine or any other contrast agent (CA) is administered before or during the acquisition, one possible result of the material decomposition is a 2D CA map that indicates, for each pixel of the projection, how much CA is present in the X-ray beam. If projections are recorded along a 3D acquisition trajectory, a 3D CA map can be reconstructed based on the entire collection of 2D contrast agent maps. The 3D CA map represents the spatial distribution of the quantitative CA concentration of the scanned object. Once the correct 3D CA map is available, a virtual non-contrast (VNC) image can be derived.

[0005] If all parts of the scanned object that contribute to the 2D CA map are located within the reconstructible FOV (Field of View), the most accurate 3D CA map and VNC (Virtual Non-contrast) image or VC (Virtual Contrast only) image can be obtained. Potential applications of the quantitative 3D CA map / VNC image are to distinguish fresh hemorrhage and iodinated staining, both of which appear as high density in post-disposal CBCT stroke scans. This can also be used in other types of interventional CBCT imaging, such as multiphase liver tumor assessment and ablation, general oncology, or urology, etc.

[0006] It has been found that contrast spectral image reconstruction may be affected by artifacts that can potentially obscure relevant medical details and undermine the usefulness of the image. Summary of the Invention

[0007] Therefore, there may be a need for improved contrast agent-assisted tomography imaging.

[0008] The object of the present invention is achieved by the subject matter of the independent claims, wherein further embodiments are incorporated into the dependent claims. It should be noted that the aspects described below for the present invention apply equally to related methods, imaging arrangements, computer program units, and computer-readable media.

[0009] According to a first aspect of the present invention, there is provided a system for image processing, comprising:

[0010] An input interface for receiving a volume image when the system is in use, the volume image being reconstructed from spectrally processed projection data that is acquired by a spectral imaging device (IA) during acquisition while a contrast agent is present in the field of view of the spectral imaging device;

[0011] A segmenter component configured to segment the volume image for contrast agent contribution to obtain a segmented volume image;

[0012] A forward projector configured to forward project the segmented volume image onto the spectrally processed projection data;

[0013] A quality checker configured to establish quality metric data based on a mismatch between the forward projected segmented volume image and the spectrally processed projection data, the quality metric data indicating a change in the concentration of the contrast agent in the field of view during the acquisition, and

[0014] An output interface for providing output data including the quality metric data.

[0015] The spectrally processed projection data may include, for example, VC projection images, which are obtained by using any one of a series of spectral image processing techniques, such as any suitable kind of material decomposition algorithm.

[0016] The output data can be used by a user to evaluate the quality of the reconstructed volume image, specifically, the degree to which the reconstructed volume image is corrupted by image artifacts caused by fluctuations in the contrast agent concentration in the FOV of the imager during projection data acquisition.

[0017] In an embodiment, the system includes a graphical display generator or visualizer configured to generate a graphical display for display on a display device based on the output data, wherein in the graphical display, if there is a mismatch, the mismatch is located in at least part of the projection data or the volume image.

[0018] In an embodiment, the system includes a control interface configured to request, based on the output data, any one or more of the following: i) emitting an alert signal, ii) additional processing related to the projection data, iii) additional acquisition of new projection data.

[0019] In an embodiment, the alert signal indicates whether such additional processing or additional acquisition is necessary, or whether such additional processing or additional acquisition is recommended, and wherein the additional processing or additional acquisition is requested based on a user input.

[0020] In an embodiment, the alert signal includes any one or more of the following: i) an audio alert signal, ii) a haptic feedback signal, iii) a visual alert signal.

[0021] In an embodiment, the additional processing includes any one or more of the following: i) additional reconstruction operations by a reconstructor for reconstructing a second volume image based on the new projection data or based on a forward projection obtained by a forward projector; ii) correction of the volume image for a mismatch by a corrector based on the output data.

[0022] In another aspect, an imaging arrangement device is provided, including the system and further including one or more of the following: i) a display device, ii) an imaging device.

[0023] In another aspect, an image processing method is provided, including:

[0024] Receiving a volume image reconstructed from spectrally processed projection data acquired by a spectral imaging device (IA) during acquisition while a contrast agent is present in the field of view of the spectral imaging device;

[0025] Segmenting the volume image for contrast agent contribution to obtain a segmented volume image;

[0026] Forward projecting the segmented volume image onto the spectrally processed projection data;

[0027] Establishing quality metric data based on a mismatch between the forward projected segmented volume image and the spectrally processed projection data, the quality metric data indicating a change in concentration of the contrast agent in the field of view during the acquisition, and

[0028] Providing output data including the quality metric data.

[0029] In yet another aspect, there is provided a computer program unit which, when run by at least one processing unit, is adapted to cause the processing unit to perform the method.

[0030] In another aspect, there is provided at least one computer-readable medium having a program unit stored thereon.

[0031] It has been found that if the CA concentration varies during projection data acquisition, the 2D CA maps derived from the projection data deliver inconsistent CA information from different perspectives. These inconsistencies result in artifacts in the reconstructed 3D CA map, i.e., lead to locally incorrect quantitative CA concentrations. Consequently, the diagnostic value of the 3D CA map is significantly reduced. Thus the same applies to VNC images.

[0032] Accordingly, the proposed system and method allow quantification of the quality of reconstructed images in spectral imaging. The output quality metric data particularly represents the extent to which such a reconstructed image is corrupted or disturbed (if any) by artifacts caused by contrast agent concentration variations occurring during the acquisition of (spectral) contrast projection images.

[0033] Accordingly, system SYS allows assessment of whether the timing of projection data acquisition has an acceptable margin of error. Specifically, the proposed system SYS can allow assessment of the extent to which the acquisition time period is outside the steady state / saturation phase PP. Forward projection is used to quantify this quality by determining the mismatch between the forward projection reconstructed spectral image (e.g., 3D CA map) and the original spectral projection image (e.g., 2D CA map) that forms the basis for the reconstructed image. Corresponding quality metric data can be output, enabling a series of different image-assisted tasks and / or control operations, which include re-running the acquisition, reconstructing using updated or forward projection data instead of the original projection data, and / or using a correction algorithm in the reconstruction as needed.

[0034] The proposed system and method are preferably used in interventional imaging, such as in a trauma department or an interventional catheterization laboratory, where high-response results that need to be delivered quickly and efficiently by computational means in real time are required. Since the proposed quality inspection measures can be provided at low computational cost and in an efficient manner that saves memory and CPU time, the proposed method achieves the goal in this regard: avoiding them in some cases when impractical, complex reconstruction or correction schemes are unnecessary (no artifacts are found). Expensive computational and memory costs only occur when needed, i.e., when the quality checker finds that the reconstructed spectral volume is indeed caused by artifacts due to in-acquisition variations of the contrast agent concentration in the FOV.

[0035] "User" refers to a person who operates an imaging device or supervises an imaging process, such as medical staff or others. In other words, the user is generally not the patient.

[0036] With respect to "(tomographic) reconstruction", the references in this document to reconstructed "volumes" in the image domain include references to specific slices. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Exemplary embodiments of the present invention will now be described with reference to the following drawings, which are not to scale unless otherwise indicated, wherein:

[0038] Figure 1 A schematic block diagram of a contrast-agent-assisted medical imaging arrangement including a spectral imaging device is shown;

[0039] Figure 2 A computer-implemented system for facilitating spectral imaging is shown; and

[0040] Figure 3 Is a block diagram of a computer-implemented method for facilitating spectral imaging. DETAILED DESCRIPTION

[0041] First referring to Figure 1 The block diagram of Figure 1 A medical imaging arrangement MIA is shown schematically. It includes an imaging device IA ("imager") and a computing system SYS. The imaging device IA is preferably X-ray based and of a tomographic type. The imager IA is preferably configured for multi-energy imaging, also known as spectral imaging. Soft issue imaging is also contemplated.

[0042] The medical imaging arrangement MIA operates to produce medical images of a patient PAT. The medical arrangement is preferably configured for contrast-agent-assisted imaging in order to enhance the contrast of low-attenuation anatomical structures (such as soft tissues of interest) in, for example, cardiac or pulmonary imaging. The medical images V can be stored, viewed, or otherwise processed. The medical images can be used for medical purposes such as diagnosis, treatment, planning, etc.

[0043] Broadly, the imager acquires projection data, which is then processed by the computing system SYS into the medical images V. The computing system SYS is broadly configured herein to help avoid certain image artifacts that may be caused by the use of contrast agents. Such image artifacts are undesirable because they may render the images useless as key details may be lost. While X-ray-based imaging is the main focus herein, other modalities that can operate on contrast-agent-assisted protocols, such as nuclear imaging (S)PE(C)T or magnetic resonance imaging (MRI), are not excluded herein.

[0044] Before turning more specifically to the computing system SYS, first continue to refer to Figure 1 To explain the operation of the spectral imager IA in conjunction with its use in a contrast-agent-assisted protocol.

[0045] Preferably, the X-ray based imager allows to obtain cross-sectional images of the ROI, preferably 3D images. To this end, the imager IA is configured for multi-directional (schematically indicated as "·" in the figure) projection data acquisition around the ROI. The acquisition (path) or "scan" may not necessarily define a full 360° angular range, but an angular range of 180° or even less may be sufficient herein. The acquisition path may not necessarily define a circular arc, although in most cases this may be so, as other geometries are also envisaged. A helical path is also envisaged herein. Thus, the present disclosure herein is not limited to an axial path, but such a path is not excluded herein, such as in a "step-and-shoot" setting.

[0046] In some embodiments, the imaging device IA may be of a rotational type to enable multi-directional projection data acquisition operations. Such a rotational imaging device allows to collect projection images · along different projection directions P(·) in a scan around the region of interest (such as around the stenosis in the mentioned coronary vessel). The imaging device IA includes an X-ray source XS and an X-ray sensitive detector XD. In some embodiments, the X-ray source XS and the X-ray sensitive detector XD opposite thereto are preferably arranged on a rotatable gantry G. The gantry G may have a recess, and the examination area is defined in the recess, where the patient or particularly the region of interest resides during imaging. Thus, the rotatable gantry may have a "C", "U" or similar shape, such as in the C-arm imaging device envisaged herein. A biplane imaging setting with two detectors having intersecting imaging axes (e.g., at right angles) is not excluded herein. Instead of using a C-arm / U-arm system, in some embodiments any generation of CT scanners are also envisaged, but preferably at least the third generation, wherein the X-ray source XS and the detector XD are mounted such that they rotate together around the patient. However, for multi-directional projection data acquisition, higher generation CT settings are also envisaged, where there is not necessarily a mechanical rotation of the source and the detector. Instead, a fixed ring of detectors around the ROI and / or a fixed plurality of X-sources arranged in a ring around the ROI are used. A cone beam imaging or other divergent beam geometries are also envisaged herein, and thus imaging is performed using a helical scan path caused by a circular path of the X-ray source, for example using a translational movement along the patient longitudinal axis Z with respect to the imaging device IA and the ROI. In most cases, such a translational movement is caused by the patient bed, where the patient on the patient bed translates during projection data acquisition. However, a scan path having a geometry other than circular / helical is not excluded herein.

[0047] In addition, while digital source imaging such as flat panel detectors is preferred, where the image is acquired natively as digital data, such as settings are not necessary. The present text does not exclude more traditional cassette-operated analog imaging methods (based on X-ray film). Analog images can be digitized after acquisition, for example, by using a digital camera. Imaging with post-digitization based on X-ray image intensifiers etc. is also envisioned in some embodiments.

[0048] Generally, imaging / acquisition includes exciting an X-ray source XS such that an X-ray beam XB emanates from the focal spot of the source XS, passes through the examination region (where the region of interest is located) and interacts with the patient tissue substance. The interaction of the X-ray beam XB with the tissue substance causes the beam to be modified. The modified beam can then be detected at a detector XD. A conversion circuit (not shown) converts the detected intensity into a digital image, particularly a projection image. The image can be used by a user in a medical procedure. During imaging, the patient can reside on a patient support PS (such as an examination table). Soft tissues (such as the soft tissues constituting the coronary blood vessels) generally provide poor contrast in native X-ray images.

[0049] To enhance the contrast, a contrast agent (sometimes "CA" or simply referred to) such as an iodine solution, gadolinium or others is delivered into the patient's bloodstream through an access point AP before or during imaging. The contrast agent can be delivered via a CA delivery device DA (such as a motorized pump), but manual delivery is not excluded herein. The delivery device DA is operable to deliver a defined volume of the contrast agent to the patient. The volume of the contrast agent so delivered travels with the bloodstream. After a certain period of time, the contrast agent accumulates at the region of interest, such as in or around a stenosis. Image acquisition is preferably synchronized with the arrival of the contrast agent at the ROI. Preferably, a "contrast" image is acquired when the contrast agent is present at the ROI. Optionally, additional "non-contrast" images are acquired when the contrast agent is not present at the ROI.

[0050] In particular, contrast projection images can be acquired in a time series, with a changing concentration of the contrast agent accumulating at the region of interest. Image acquisition can start before the accumulation of the contrast agent at the ROI, can continue throughout the accumulation, and can terminate after the accumulation. Generally, the contrast agent concentration increases with time after delivery, will stabilize at a certain maximum value, and then will wash out or decrease with time until it completely disappears. While in some cases single or a specific discrete number of still images can be acquired, acquisition of a stream of images or frames at a suitable frame rate is preferred as this allows real-time dynamic observation. Specifically, the acquired stream of projection images can be visualized by a visualizer VIZ as a live video feed on a display device DD. Alternatively or additionally, still images are so visualized. The operation of the proposed system SYS is based on processing such still or stream images. The images can be stored in an image memory MEM or can be otherwise processed.

[0051] X-ray imaging is typically based on the attenuation coefficients of the (one or more) substances that make up the region of interest. The attenuation (coefficient) depends on the energy of the X-ray beam. To utilize this energy dependence, a spectral imager IA configured for spectral imaging can be used. This can include detector-side or source-side solutions. Detector-side solutions include photon-counting detector hardware or, for example, dual-energy setups with bilayer hardware. More than two detector layers can be used, but this is less common. Source-side solutions include multi-source setups, kVp switching, filtering, etc.

[0052] The imaging setup in spectral imaging allows for the acquisition of multi-dimensional image data. Such multi-dimensional image data includes projection images at different energy levels, rather than the energy-integrated images along different projection directions as provided by conventional (non-spectral) X-ray imaging setups.

[0053] Such a spectral CT imaging setup includes a spectral image processor SP. The spectral image processor SP can implement spectral image processing algorithms, such as material decomposition or others. A series of such spectral image processing algorithms are envisioned herein. The spectral processor SP processes the multi-energy projection images acquired by the source or detector-side imaging setup just described, and thereby calculates a spectral image different from the original input multi-energy projection images. Thus, a spectral projection image is computationally derived from the input multi-energy projection images.

[0054] Such spectral images in the projection domain or image domain can particularly include only contrast images ("VC") as mainly envisioned herein. Such VC images represent an approximation of the images that might have been obtained in the case where only the contrast agent is present in the field of view of the imager IA, excluding other materials. Thus, the image contrast in VC is specifically or mainly focused on the contrast agent.

[0055] The tomographic reconstructor RECON processes the projection data into a tomographic / volume image R(·)=V, as will be described in more detail below. The volume data V is spectral image data due to the spectral processor SP. The spectral processor SP can operate on the projection data as acquired in the projection domain, or can operate in the image domain after reconstruction. For simplicity, the symbols "V" or "R(·)" can be used herein to indicate the spectral volume image in the image domain, regardless of the domain in which the spectral processing SP occurs. It is such spectral volume data V that is processed by the proposed system SYS. The spectral processor SP can be integrated into the reconstructor RECON.

[0056] The spectral processor SP and the proposed system SYS can be implemented on different computing systems, possibly geographically separated, such as in a distributed "cloud" architecture. However, implementations on a single computing system are also envisioned herein. Imaging as envisioned herein is preferably attenuation-based, but this does not exclude other X-ray modalities such as dark field imaging and / or phase contrast imaging.

[0057] The reconstructor RECON is configured to reconstruct a volume image V = R(·) from the acquired projection data ·, which is acquired when at least some contrast agent is present in the field of view of the imager IA and thus in the region of interest including at least a portion of the vasculature of interest.

[0058] The projection domain is the space where the detected projection data is located at the detector. Tomographic reconstruction is performed by the reconstructor RECON, which is a transform operation R according to which the volume data is calculated. The reconstruction transform operation R is a transform from the projection domain to the image domain. The volume data V is spatially associated with the image domain. The image domain is a portion of the space defined between the X-ray source XD and the detector D, in which the region of interest (through which the contrast agent passes) resides during imaging. Thus, the region of interest specifically includes the portion of the vasculature VS to be imaged, and sometimes the contrast agent resides therein. Conceptually, the image domain can be considered to consist of a 3D grid of spatial points ("voxels").

[0059] The reconstruction operation includes calculating the image values at each grid point (voxel) to obtain the volume data R(·). A manifold of reconstruction algorithms is envisioned herein, including FBP, algebraic, or iterative or others. The reconstruction can be iterative. It iteratively assigns values to the image domain during each iteration cycle based on the projection images and can be based on the optimization of a function. During the iteration, the reconstruction converges to a solution in the image domain, i.e., converges to the final reconstructed volume V. This function can measure the consistency of various reconstructed volumes with the measured data. This function can also include additional functional terms that manipulate the convergence behavior of the iterative algorithm.

[0060] Turning now more specifically to the computing system SYS, which is configured herein for quality control in a contrast agent-assisted imaging protocol, thereby leveraging the material-specific imaging capabilities of spectral imaging. The system SYS can be configured to alert the user to certain image artifacts that may occur or have occurred in the reconstructed volume image, particularly in such contrast agent-assisted imaging applications. The system SYS can facilitate remediation, reduction, or avoidance of such artifacts.

[0061] After administration of a certain volume of contrast agent CA, it travels with the blood flow to accumulate at the region of interest ("ROI"), e.g., in a portion of a blood vessel of interest, such as in the coronary arteries of a patient's heart. The distribution of the liquid blood and contrast agent mixture conforms to the internal structure of the blood vessel, in particular its lumen. Spectral imaging allows mapping of the concentration distribution of the contrast agent and thus allows good and targeted image contrast of the anatomical structures of interest, such as blood vessels. This in turn allows for more precise treatment or diagnosis notification.

[0062] Specifically, the concentration distribution of the contrast agent over time t can be represented by a concentration curve c(t), as Figure 1A schematically shown. At a certain ROI, after a certain arrival time period after administration, in the rising phase RP, an increase in the contrast agent concentration is expected, followed by a stable or saturation phase PP, where the contrast agent concentration reaches a maximum and remains relatively constant over a certain time period ("stable period") until it decreases again in the falling phase DP due to washout. The contrast agent curve c(t) is a function of the location of the region of interest, and thus the behavior / shape of c(t) may vary depending on the location.

[0063] The timing of projection data acquisition is a consideration in contrast-assisted imaging protocols as used herein. Ideally, the imager IA should be controlled such that projection data acquisition starts at the correct time. Specifically, the projection data should be acquired during the stationary phase PP, during which the contrast agent concentration remains relatively stable due to saturation. Due to various clinical factors (stress level, patient compliance, user fatigue or experience, etc.), it may not always be possible to time the projection data acquisition such that it occurs entirely during the stationary phase. In these cases, the concentration may change during projection data acquisition in the rising phase RP or the falling phase DP.

[0064] Thus, the acquired contrast projection images may include attenuation contributions from the changing contrast agent concentration. It has been found that reconstructing based on such projection images using contributions from different contrast agent concentrations will result in image artifacts in the reconstructed image V. This is caused by certain consistency violation assumptions experienced by some types of tomographic reconstruction algorithms. For example, it can be expected that pixels acquired within a certain time interval or along different views ·, ·' represent the same contrast agent attenuation. If the concentration has changed within that time interval, this may not be the case. Thus, due to the concentration changes recorded in the projection domain, the reconstruction algorithm may not easily be able to unambiguously assign values to the corresponding voxels in the image domain. In particular, iterative reconstruction may behave unexpectedly in such cases. This ambiguity may lead to convergence to a substandard solution, resulting in artifacts. It is also possible for filtered backprojection (FBP)-based reconstruction methods to be affected in such a way.

[0065] The computer system SYS is configured herein to facilitate reducing such image artifacts in reconstructed images that have been caused by changing the contrast agent concentration, or the system SYS can alert the user that such artifacts may manifest in the current reconstructed image. Such facilitation can include alerting the user that the projection data may have been acquired outside of a stationary period and enabling reconstruction artifacts to be expected. The user can then take appropriate action, or the system can suggest such action and, optionally, automatically or when user approval input can be received via appropriate UI interactions, etc., implement the action. Thus, the system can include a user interface (UI) (such as a G (graphical) UI or others). The (G)UI can be configured to alert the user and / or allow the user to control which remedial action to take. Thus, the system SYS can be referred to herein as an (image) artifact reducer or alertor.

[0066] Now referring to Figure 2 the block diagram of Figure 2 illustrates the components of the artifact alertor system SYS. The system SYS can be implemented on a single computing unit PU or can be implemented in a distributed cloud environment where some or all of the components are implemented by different computing units / systems / platforms, etc., preferably interconnected in a suitable wired or wireless communication network. The system can be integrated into an imaging device IA, such as in its operation console OC or in an associated workstation, etc. Preferably, the system SYS is configured for on-site (quasi) real-time use, i.e., during an imaging session when the patient is in an examination room, theater, or bathroom laboratory, etc. Thus, the system SYS can be implemented in a computing system having a high-throughput processor (e.g., a processor with a multi-core design).

[0067] The projection data acquired at the detection module XD is processed by a spectral processor SP into spectrally processed projection data. Due to the multi-spectrum setting, the acquired projection data is multi-energy data, i.e., each pixel in the projection domain is assigned at least two intensity values, each intensity value for a different X-radiation energy range. Spectral processing algorithms (such as material decomposition) are used to resolve the original multi-energy projection data into spectrally resolved projection data ·’, where the contrast is now material-specific, such as for a target material of interest. Such a target material can be a contrast agent (e.g., iodine or others). Thus, the image contrast in the spectral projection data ·’ can exclusively or predominantly correspond to the concentration of the contrast agent. It is possible to calculate only a virtual contrast agent projection image (“VC”), and this is indeed contemplated in an embodiment. Thus, each pixel in the VC projection image measures the concentration (mass per volume element) of a specific contrast agent material, which is known a priori. Such spectrally processed contrast-agent-only projection data ·’ (such as VC) can be understood and referred to herein as a “2D contrast agent map” due to its customized CA-specific imaging contrast. There is one or more such 2D contrast agent maps for each (primary) projection direction (e.g., the direction of the central beam).

[0068] Various spectral decompositions can be used, such as the 1976 method of “Energy-selective reconstructions in X-ray computerized tomography” (Phys Med Biol., Vol. 21, No. 5, p. 733) published by R.E. Alvarez & A. Macovski or its homologs. Such or similar related methods involve solving a system of simultaneous linear or non-linear equations based on the attenuation coefficients of different materials at different energies. The different materials include the target material and at least one other material (or material mixture) known / assumed to be present in the FOV at the time of acquisition.

[0069] Then, such spectrally processed projection data ·’ (such as VC) is passed to a tomographic reconstructor RECON to generate a reconstructed image V in the image domain, such as a volume V or a single cross-sectional slice. Under ideal conditions (no change in contrast agent concentration), the material-specific contrast can enter the image domain. Thus, the reconstructed volume V can be understood as a 3D contrast agent map under ideal conditions, which may still be corrupted by the artifacts mentioned.

[0070] The so-reconstructed volume V based on the spectrally processed projection data ·’ is received at the input port IN.

[0071] A segmentation component SC processes the input volume V to obtain a segmented volume V’.

[0072] The segmenter SC can be configured to process an input volume V of each voxel. The segmented volume V' can be considered a modified version of the input image V. The segmentation operation can be understood as an image domain sanity check for each voxel. Thus, the segmentation component applies a consistency check at the image domain level regarding whether the reconstructed value of each voxel is consistent with the assumption that the corresponding voxel value represents a true contrast agent contribution.

[0073] Accordingly, the segmenter SC can assign a fractional binary or floating / continuous metric, which is configured to represent and vary with the degree to which a voxel indicates a true or real contrast agent contribution. If the segmenter finds that a given voxel is not a true or real contrast agent contribution, or is otherwise corrupted, such as being considered untrue, the voxel is modified, such as by replacement with an estimate, such as a default value or otherwise calculated. For example, its value can be interpolated from neighboring voxels that are not found to be corrupted. "Corruption" in this context includes the assumption that the value at a given ("corrupted") voxel may be due to contrast agent concentration fluctuations as recorded in pixels in the protected domain, which are related to that voxel position under projection. The assignment of such a sanity score and the voxel modification operation can be two separate processing steps, or can in fact be combined into a single operation.

[0074] Taking into account the contrast agent for the corresponding ROI, the consistency check can be based on one or more strategies representing prior assumptions about the way voxel values are expected to vary. Thus, the strategies can be implemented and stored as a corresponding set of conditions, such as in terms of value ranges, topological conditions, etc. Each such strategy can be a function of the contrast agent type and / or the ROI. Such strategies can be stored in memory. Thus, different such sanity strategies can be used depending on the ROI / application / contrast agent type, etc., each strategy representing certain assumptions about voxel values and their distributions based on prior medical knowledge of the case at hand. The user interface UI can allow the user to select which such image domain sanity check strategy the segmenter SC is to apply, in order to better align the operation of the system SYS with the details of the case at hand. For example, the UI can be a GUI, which is arranged to include a list widget of stitching boxes, a dropdown widget, or other menu or selection structure widgets.

[0075] An image value threshold or an allowable range strategy can be used. More specifically, since negative concentrations are not possible, negative values can most certainly be excluded as artifacts. If, for example, small concentration values are not expected in a CA image (e.g., angiography), then another range-based strategy can be used. The strategy can then be configured more aggressively because even all values below a small positive threshold are then set to zero or some other default value. A similar strategy can be run when, for example, high concentration values are not expected, but starting from the other “end”. Thus, these values can be limited at a certain predefined maximum amount. Secondly, if for some reason a uniform CA concentration region is expected, then a topology-based strategy can be used to enforce connectivity and / or uniformity between adjacent voxels, e.g., by processing the image to redistribute values with a level set method, or by image modification driven by L1-TV regularization, etc.

[0076] The forward projector FP is operable to forward project the so-segmented volume V’ (the object in the image domain) into the projection domain and onto the original spectrally processed projection data ·’ (2D contrast agent map) during a forward projection operation.

[0077] It is expected that if the consistency check based on the image domain is correctly performed within a specific error tolerance and if there is no contrast agent concentration change recorded by the projection data ·’, the synthetic projection FP(V’) obtained by forward projecting V’ should again correspond to (be equal to) the spectrally processed input projection data ·’ within a suitable error bound. If it does not so correspond, this can be regarded as an indication that at least some of the projection data are recorded outside the stationary phase and are actually corrupted.

[0078] The quality check module QC implements quality check measures by establishing quality metric data indicating contrast agent concentration changes in the field of view during acquisition. For this purpose, for example, the quality checker is configured to compare the forward projection synthetic projection FP(V’) with the original spectrally processed projection data ·’ to obtain a trust or mismatch metric (also referred to herein as mismatch information). The quality metric data can be based on this mismatch metric.

[0079] The check module QC can use a series of metrics, such as the Euclidean distance, weighted Euclidean distance, or any other metric taken per pixel. The quality checker QC can use any distance based on the norm L p (p>0). The quality checker module QC can be performed at the per-pixel level in the projection domain.

[0080] The mismatch measures can be combined into a single number, such as a binary measure, which indicates whether the reconstructions V, V' can be considered faithful reconstructions, i.e., whether there were no changes in the contrast agent concentration during acquisition of the underlying contrast projection images. In an example, the volumes V, V' are considered faithful if the projection data is acquired during a stationary phase.

[0081] On the other hand, if there are changes in the contrast agent concentration during acquisition, this will likely manifest as image artifacts in the reconstructed image / volume / slice V. These artifacts are expected to be identifiable by determining the mismatch between the forward projected projection image and the spectrally processed projection data. A continuous measure with a range in a bounded interval, such as the unit interval [0,1], can be used to quantify the mismatch, if any. A mismatch less than a certain error tolerance is considered a match, and the conclusion is that there is no corruption, and thus the reconstructed volumes V, V' are considered faithful.

[0082] Very similar to the segmentation performed by the segmentation module, the quality check module QC can use a thresholding strategy. The quality check output indicating the mismatch between the forward projected volume FP(V') and the spectrally processed projection data ·' can alternatively be provided as a map in the projection domain, which indicates the mismatch at each pixel location in the projection domain. This can be used to determine where the mismatch is more prevalent in the projection domain.

[0083] Thus, the mismatch and thus the faithfulness of the reconstructed image can be represented locally as a map. The faithfulness map can be overlaid on one or more frames in the projection data ·', and can be displayed by the visualizer VIZ on the display device DD. Alternatively or additionally, the mismatch map can be backprojected and fused with the reconstructed volume V or V' to highlight in 3D (in the image domain) where the inconsistencies are most prevalent. The reconstructed volumes V, V' with the backprojected mismatch map can be displayed by the visualizer VIZ on the display device DD for visual inspection by the user. The mismatch values can be thresholded and optionally color-coded to quickly inform the user of the spatial distribution of the mismatch and how its severity varies spatially.

[0084] Thus, depending on the domain (in the projection domain or the image domain) in which the mismatch is to be used (displayed, processed, etc.), the mismatch can be globally represented by a single scalar or locally represented as a 2D or 3D map.

[0085] Sometimes, if desired, the mismatch value (scalar) or 2D / 3D mismatch map can be displayed alone instead of being combined with the projection data or the reconstructed volumes V, V'. Instead of or in addition to displaying, the mismatch data (scalar or 2D / 3D map) can be stored in memory or otherwise used, for example for control purposes. Thus, the control interface CL can use the mismatch data to facilitate a number of useful control operations to assist the user in operating the imaging device IA, etc., for example for downstream applications / tasks after imaging, or any other tasks, such as remedial processing, issuing alerts in various forms, etc.

[0086] For example, the transducer TR can be used to convert mismatch information in user-consumable signals such as a flash, an audible alert signal, or a tactile signal. In the latter case, the UI operating element (joystick, etc.) of the imager IA is set to vibrate corresponding to the detected amount of mismatch.

[0087] If it is found that the spectral volume suffers from such artifacts, as described above, then one or more of the following control operations can be included: triggering a re-acquisition of the projection data, adjusting the imager settings (kV, mA, etc.), re-running the reconstruction by the reconstructor RECON preferably using a different reconstruction algorithm, using an image processing artifact reduction algorithm to control the corrector COR to correct the spectral volume V', scheduling a subsequent imaging or other medical session, etc. Any one of such control operations can be used in any combination / sub-combination with one or more other control operations, or used alone as needed.

[0088] It should be understood that the operations of the quality checker QC and / or the segmenter SC can be based on machine learning ("ML") methods. In this embodiment, corresponding machine learning models trained based on training data, such as neural networks (preferably of the convolutional type), are used. The training data can be obtained from a medical image data database. A supervised method can be used, where the training data is labeled by medical experts on a pixel / voxel or patch basis. The medical experts can assign a score label to the pixels indicating the degree of contrast agent contribution, or which mismatch level is a true mismatch or small enough to pass as a match, etc. Once properly trained, segmentation or quality checking can be obtained by applying the trained model to the current reconstructed volume V' or the mismatch data FP(V') pair. In addition to or instead of the machine learning model, the segmenter SC and / or the quality checker QC can use a hydrodynamic analysis model based on physiological insights.

[0089] Now referring to Figure 3 , Figure 3 a flowchart of a computer-implemented method for facilitating contrast agent-assisted tomography imaging is shown.

[0090] At step S310, multi - energy projection data is acquired by an imaging device configured for spectral imaging. As described above, detector - side embodiments or source - side embodiments are contemplated herein. The projection data may be corrupted because it is acquired at least partially outside the stationary phase PP (see Figure 1A above).

[0091] At step S320, the acquired projection data is processed by a spectral processing algorithm such as a material decomposition algorithm. In particular, the thus - obtained spectrally - processed projection data mainly or uniquely represents the contrast originating from the contrast agent. Any other contributions from other materials or from any other intervening structures are reduced or excluded. Thus, a “virtual contrast agent only” (VC) projection image is obtained. It should be understood that if there is corruption in ·, then this corruption is fed into the spectrally - processed projection data ·'.

[0092] At step S330, the thus - spectrally - processed projection data ·' is reconstructed into an input spectral volume V. Material - specific contrast is brought into the reconstruction and thus into the spectral input volume V. The input volume V is received at step S340.

[0093] At step S350, the thus - received reconstructed volume V is then segmented. Segmentation is to be interpreted broadly and includes at the voxel level the assignment of a corresponding quantification regarding the degree to which the corresponding voxel represents a true contribution from the contrast agent. If a voxel is not considered to be so representative, it is replaced based on a sanity - check strategy. The result of this operation is a modified / segmented volume V'. The segmentation may provide a segmentation map associated with the volume V'. Also contemplated is slice - by - slice processing instead of or in combination with voxel - by - voxel processing, and in some cases, it may not be necessary to segment every voxel. Sometimes, some regions may be excluded from consideration.

[0094] The segmented volume V' is forward - projected at step S360 to obtain a set of synthetic projection data FP(V').

[0095] Then, at step S370, the synthetic projection FP(V') is compared with the original spectrally - processed projection data ·' to obtain mismatch data as a quality check. If the projection data is not corrupted (and thus in ·'), a match is expected. If there is a mismatch, the conclusion is that such corruption exists and this fact is flagged. The mismatch data is preferably configured to vary with the amount / severity of the mismatch and represent the amount / severity of the mismatch, and preferably also represents its spatial distribution in the image domain (after back - projection) or in the projection domain.

[0096] The mismatch data obtained in the comparison at step S370 can be calculated by a suitable distance or similarity metric. The mismatch data can be provided as a global scalar value (even, in binary form: "is a mismatch" / "is not a mismatch"), or can be provided in a pixel - level manner as a 2D map indicating the amount of mismatch between two sets of projection data ·' and FP(V'). The 2D map of mismatch in the projection domain can be back - projected into the image domain to provide a 3D mismatch map.

[0097] As required, the mismatch information is provided in step S380 in any form as the output of all means for further processing.

[0098] The mismatch data can be configured to quantify the total amount of contrast agent change, e.g., by summing the absolute mismatches detected on some or all of the 2D contrast agent maps.

[0099] The mismatch data · is an indication of the extent to which the projection images · are acquired at least partially during a stationary period, or an indication of whether the projection images · are acquired at least partially during a descending or ascending phase, outside the stationary phase, where the contrast agent contribution changes. Thus, the mismatch data is a measure of the faithfulness of the reconstruction.

[0100] If the metric indicates low faithfulness at step S420, any one of a number of different remedial processes or other steps can be initiated. For example, as an alternative or supplement to step S420, at step S430, once the mismatch exceeds a certain predefined threshold, an alarm signal of an auditory, tactile, or visual type can be issued.

[0101] Regarding the remedial processing step S420, this can include running a second reconstruction operation. Instead of or in addition to the original spectrally processed set ·', this can be based on the synthetic projection data FP(V'). For example, a computationally more demanding reconstruction algorithm can be used for the second reconstruction, which is designed to account for contrast agent fluctuations in the underlying projection data.

[0102] Thus, the proposed system and method allow for saving valuable CPU time, since such a computationally more intensive reconstruction algorithm is only run when needed. A computationally more demanding reconstruction algorithm can also be run on the original data · or ·'. Similarly, and as an alternative, in the remedial processing step S420, in the case where artifacts are found through a quality inspection step to provide a corrected spectral volume, a more complex image corrector algorithm can be run through a corrector COR that processes the reconstructed spectral volume V'. Such a correction algorithm can also be computationally expensive, since it attempts to remove artifacts caused by contrast agent concentration changes recorded by the underlying projection data ·, ·'. Thus, due to the proposed quality inspection step S370, this computational cost is only incurred when needed (when artifacts are found).

[0103] Additionally or alternatively, in the remedial processing step S420, the imager IA can be controlled to acquire a new projection data set with the newly administered contrast agent, where this time repeated attempts are made to capture the projection data within a stationary period. However, this would expose the patient to a higher dose of contrast agent and radiation, which would normally be avoided due to negative health effects. Therefore, the acquisition of a new set of contrast projections may only be done in abnormal situations such as extreme damage. Then, the proposed method again helps to avoid such re-acquisition, and the associated efficiency cost and health cost: the re-acquisition is only done when needed.

[0104] At step S410, fidelity data or mismatch data is displayed. This can be presented graphically and / or can be displayed numerically. In one embodiment, the image portions related to the mismatch are located in the projection domain or in the image domain.

[0105] A graphical indication can be provided, such as by color or gray-level coding of the mismatch data, to indicate where the mismatch originates, which portions are more affected, and their extent, etc. Thus, it may not be that the entire volume is unfaithful, but only certain portions of it. Thus, "locating" as used herein is the operation of mapping the mismatch data into 2D (projection domain) or 3D (image domain) in order to represent the fidelity distribution.

[0106] The mismatch data provides the user with feedback on the quality / reliability metric for the acquisition. Then, the user can choose to repeat the acquisition. Alternatively, a re-acquisition is automatically triggered. Similarly, the user can decide whether an improved reconstruction (with a more advanced concentration fluctuation compensation reconstruction) should be performed, or such a reconstruction is automatically triggered. In both cases, the automatic trigger is through a suitable control interface CL.

[0107] It will be understood that any component of the system SYS( Figure 2 ) and any processing step described above at each voxel / pixel level( Figure 3 ) can instead be performed for each tile (subset of voxels), which is an optional consideration if saving computational time is of interest.

[0108] The components of the system SYS can be implemented as one or more software modules, running on one or more general-purpose processing units PU (such as a workstation associated with the imager IA), or on a server computer associated with a set of imagers.

[0109] Alternatively, some or all components of system SYS can be arranged in hardware, such as a suitably programmed microcontroller or microprocessor, such as an FPGA (Field Programmable Gate Array), or as a hardwired IC chip, an Application Specific Integrated Circuit (ASIC), integrated into the imaging system SYS. In yet another embodiment, system SYS can be implemented partly in software and partly in hardware both.

[0110] The different components of system SYS can be implemented on a single data processing unit PU. Alternatively, some or more components are implemented on different processing units PU, which may be remotely arranged in a distributed architecture and connectable in a suitable communication network, such as in a cloud setting or a client-server setting, etc.

[0111] One or more features described herein can be configured as or implemented as or have circuitry and / or combinations thereof encoded within a computer-readable medium. The circuitry can include discrete and / or integrated circuits, a system-on-chip (SOC) and combinations thereof, a machine, a computer system, a processor and a memory, a computer program.

[0112] In another exemplary embodiment of the present invention, there is provided a computer program or a computer program unit, characterized in that it is adapted to execute the method steps of the method according to one of the previous embodiments on a suitable system.

[0113] Therefore, the computer program unit can be stored on a computer unit, which can also be part of an embodiment of the present invention. The computing unit can be adapted to execute the steps of the method described above or induce the execution of the steps of the method described above. In addition, it can be adapted to operate the components of the device described above. The computing unit can be adapted to automatically operate and / or execute the commands of a user. The computer program can be loaded into the working memory of a data processor. The data processor can thereby be equipped to execute the method of the present invention.

[0114] This exemplary embodiment of the present invention covers both a computer program that uses the present invention from the start and a computer program that transforms an existing program into a program using the present invention by means of an update.

[0115] Furthermore, the computer program unit can provide all the necessary steps to implement the process of the exemplary embodiment of the method described above.

[0116] According to another exemplary embodiment of the present invention, there is provided a computer-readable medium, such as a CD-ROM, wherein the computer-readable medium has a computer program unit stored thereon, which is described in the previous part.

[0117] A computer program can be stored and / or distributed on a suitable medium (in particular but not necessarily a non-transitory medium), such as an optical storage medium or a solid-state medium provided together with or as part of other hardware, but the computer program can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0118] However, the computer program can also exist on a network such as the World Wide Web and be downloadable into the working memory of a data processor from such a network. According to another exemplary embodiment of the present invention, a medium for making a computer program unit available for download is provided, the computer program unit being arranged to execute the method according to one of the previously described embodiments of the present invention.

[0119] It must be noted that the embodiments of the present invention are described with reference to different subjects. Specifically, some embodiments are described with reference to the claims of the method type, while other embodiments are described with reference to the claims of the device type. However, those skilled in the art will understand from the above and the following description that, unless otherwise indicated, any combination between the features related to different subjects is also considered to be disclosed by this application, in addition to any combination of the features belonging to one type of subject. However, all features can be combined to provide a synergistic effect that exceeds the simple sum of the features.

[0120] Although the present invention has been described in detail and illustrated in the accompanying drawings and the foregoing description, such description and illustration should be considered illustrative or exemplary and not restrictive. The present invention is not limited to the disclosed embodiments. By studying the drawings, the disclosure and the dependent claims, those skilled in the art can understand and implement other variations of the disclosed embodiments when practicing the claimed invention.

[0121] In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality. A single processor or other unit can perform the functions of several items recited in the claims. Although specific measures are recited in mutually different dependent claims, this does not indicate that the combination of these measures cannot be used advantageously. Any reference signs in the claims should not be construed as limiting the scope. Such reference signs can consist of numbers, letters or any alphanumeric combination.

Claims

1. A system (SYS) for image processing, comprising: An input interface (IN) for receiving a volume image when the system is in use, the volume image being reconstructed from spectrally processed projection data acquired by a spectral imaging device (IA) during acquisition while a contrast agent is present in the field of view of the spectral imaging device; A segmenter (SC) component configured to segment the volume image for contrast agent contribution to obtain a segmented volume image; A forward projector (FP) configured to forward project the segmented volume image onto the spectrally processed projection data; A quality checker (QC) configured to establish quality metric data based on a mismatch between the forward projected segmented volume image and the spectrally processed projection data, the quality metric data indicating a change in the concentration of the contrast agent in the field of view during the acquisition, and An output interface (OUT) for providing output data including the quality metric data.

2. The system according to claim 1, comprising a visualizer (VIZ) configured to generate a graphical display for display on a display device (DD) based on the output data, wherein, In the graphical display, if there is such a mismatch, the mismatch is located in at least part of the projection data or in the volume image.

3. The system according to any one of the preceding claims, comprising a control interface (CL) configured to request any one or more of the following based on the output data: i) emitting an alarm signal, ii) additional processing related to the projection data, iii) additional acquisition of new projection data.

4. The system according to claim 3, wherein, The alarm signal indicates whether such additional processing or additional acquisition is necessary, or whether such additional processing or additional acquisition is recommended, and wherein the additional processing or the additional acquisition is requested based on a user input.

5. The system according to claim 3 or 4, wherein, The alarm signal includes any one or more of the following: i) an audio alarm signal, ii) a tactile feedback signal, iii) a visual alarm signal.

6. The system according to any one of claims 3-5, wherein, The additional processing includes any one or more of the following: i) additional reconstruction operations by a reconstructor (RECON) for reconstructing a second volume image based on the new projection data or based on the forward projection obtained by the forward projector (FP); ii) correction of the volume image by a corrector (COR) for the mismatch based on the output data.

7. The system according to claim 1, wherein, The spectrally processed projection data includes a 2D contrast agent (2D CA) map representing the concentration of the contrast agent, and the volume image includes a 3D contrast agent (3D CA) map.

8. The system according to claim 1, wherein, The segmenter (SC) is configured to apply a consistency check to the volume image at the image domain level, the consistency check indicating whether the reconstructed value of each voxel is consistent with the assumption that the corresponding voxel value represents a true contrast agent contribution.

9. The system according to claim 8, wherein, The consistency check includes an image value thresholding strategy and / or a topology-based strategy implementing uniformity and / or connectivity between adjacent voxels.

10. An imaging arrangement device (MIA) comprising a system according to any one of the preceding claims and further comprising one or more of the following: i) the display device (DD), ii) the imaging device.

11. A method of image processing, comprising: receiving (S340) a volume image reconstructed from spectrally processed projection data acquired by a spectral imaging device (IA) during acquisition when a contrast agent is present in the field of view of the spectral imaging device; segmenting (S350) the volume image for contrast agent contribution to obtain a segmented volume image; forward projecting (S360) the segmented volume image onto the spectrally processed projection data; establishing (S370) quality metric data based on a mismatch between the forward projected segmented volume image and the spectrally processed projection data, the quality metric data indicating a change in concentration of the contrast agent in the field of view during the acquisition, and providing (S380) output data comprising the quality metric data.

12. A computer program unit which, when run by at least one processing unit, is adapted to cause the processing unit to perform the method according to claim 11.

13. At least one computer-readable medium having stored thereon the program unit according to claim 12.

Images