Method for determining myocardial extracellular volume fraction, processing system, medical imaging device, computer program, and computer-readable storage medium

Abstract

To provide a method for determining a myocardial extracellular volume fraction.SOLUTION: A method for determining a myocardial extracellular volume fraction 1 comprises: receiving a measurement data set 3, the measurement data set including energy resolved data based on a CT scan; reconstructing a morphology preserving image data set 8 based on a first photon-energy band 9 or a first combination of photon-energy bands described by the measurement data set; segmenting a blood pool 10 within a myocardium 11 in the morphology preserving image data set; reconstructing a contrast agent map 12 based on a second photon-energy band or a second combination 13 of photon-energy bands 14, 15, 16 described in the measurement data set; determining a reference value 17 based on a pixel or voxel 19 of the contrast agent map within the segmented blood pool; and determining each myocardial extracellular volume fraction based on the reference value and a value given for a pixel or voxel outside the segmented blood pool by the contrast agent map.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a computer-implemented method for determining at least one myocardial extracellular volume fraction of a patient. The present invention also relates to a processing system, a medical imaging device, a computer program, and a computer-readable storage medium. [Background technology]

[0002] Interstitial myocardial fibrosis is a significant disease process associated with different types of cardiomyopathies. The myocardial extracellular volume fraction (ECV) is increased in cases of diffuse myocardial fibrosis or infiltration and is therefore considered a distinguishing feature of myocardial pathology for cardiomyopathies.

[0003] Magnetic resonance imaging can determine the myocardial extracellular volume fraction by determining the equilibrium distribution of gadolinium contrast between the myocardium and blood.Iodine-based computed tomography contrast agents also exhibit similar kinetics to gadolinium, allowing the determination of myocardial extracellular volume fraction using computed tomography.

[0004] In principle, conventional single-energy computed tomography (CT) scans could be used for this purpose. However, this requires a baseline original scan without contrast agent, a conventional coronary CT angiography scan with contrast agent, and a so-called late-enhancing scan, recorded several minutes after the injection of contrast agent to depict the equilibrium distribution of contrast agent already in the intercellular volume of the myocardium. Using three separate scans exposes the patient to relatively high x-ray radiation and requires registration of each scan, acquired at different times, which is prone to introducing artifacts and therefore reduces the accuracy of the results.

[0005] Significant improvements can be achieved by using dual-energy computed tomography. In dual-energy tomography, X-ray absorption is determined for different photon-energy bands. This makes it possible to distinguish between different materials by analyzing X-ray absorption based on photon energy. With such dual-energy acquisition, for example, iodine-based contrast agents can be selectively imaged, thus providing a contrast agent map. On the other hand, the same dual-energy scan can also map water, i.e., primarily tissue.

[0006] If the segmentation (or division) of the myocardium (e.g., the left ventricular myocardium) and the blood pool (particularly the intramyocardial blood pool) is known, it is possible to determine the myocardial extracellular volume fraction from such contrast agent maps using the following formula:

[0007]

number

[0008] In the upper number, "ΔHU t" indicates the change in X-ray absorption, expressed in Hounsfield units, based on the presence of contrast agent in the myocardial tissue, which can be determined directly from the contrast agent map. And "ΔHU b " indicates the change in X-ray absorption, expressed in Hounsfield units, based on the presence of contrast in the blood pool, which can also be determined from a contrast map. The variable "h" indicates the hematocrit value for the patient, which can be known from patient data, determined by a blood test, or calculated from computed tomography data, for example.

[0009] The use of dual-energy computed tomography to determine myocardial extracellular volume fraction is known, for example, see Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document 3.

[0010] A method that allows the hematocrit value to be determined directly from computed tomography data is described in Non-Patent Document 4.

[0011] In principle, it is possible to perform an automatic segmentation of, for example, the blood pool in the left ventricle and the myocardium of the left ventricle. An exemplary method for such automatic segmentation is given in [5]. [Prior art documents] [Non-patent literature]

[0012] [Non-Patent Document 1] Lee, H.-J. et al., "Myocardial Extracellular Volume Fractionation Using Dual-Energy Equilibrium Contrast-enhanced Cardiac CT in Nonischemic Cardiomyopathy (NICM): Prospective Comparison with Cardiac MR Imaging," Radiology 280:1 (2016), pp. 49-57. [Non-patent document 2] Abadia, AF et al., "Myocardial extracellular volume fraction for differentiating healthy from cardiomyopathic myocardium using dual-source dual-energy CT," JCCT (J Cardiovasc Comput Tomogr) 14(2) (2020) pp. 162-167 [Non-patent document 3] Oda, S. et al., "Extracellular Volume Quantification and Myocardial Late Iodine Enhancement Using Dual-Layer Spectral Detector Dual-Energy Cardiac CT," Radiology, Cardiothoracic Imaging, 2019 [Non-patent document 4] Treibel, TA et al., "Automated quantification of myocardial extracellular volume by cardiac computed tomography: synthetic ECV by CCT," JCCT, 11(3) (2017), pp. 221-226. [Non-Patent Document 5] Zheng, Y. et al., "Four-chamber cardiac modeling and automatic segmentation for three-dimensional cardiac computed tomography volumes using marginal space learning and steerable features," IEEE Trans Med Imaging 27(11) (2008) pp. 1668-81 Summary of the Invention [Problem to be solved by the invention]

[0013] However, known algorithms for segmenting the myocardium and blood pool have performed poorly when applied to contrast maps. Therefore, the above-described method for determining myocardial extracellular volume fraction based on dual-energy computed tomography requires either manual segmentation of the image data or acquisition of an additional scan for automatic segmentation, preferably within a relatively short time after contrast injection. For example, a full computed tomography angiography sequence can be performed, or at least an additional scan containing contrast essentially only within the vessels can be used. However, this requires registering at least two scans taken at different times. Furthermore, the patient is exposed to more X-ray radiation than with a single scan.

[0014] The present invention therefore provides an improved method for determining at least one myocardial extracellular volume fraction for a patient, essentially obviating the need for manual segmentation, even when using only a single computed tomography scan. [Means for solving the problem]

[0015] The above problem is solved by a computer-implemented method (or a method implemented using a computer) for determining at least one myocardial extracellular volume fraction for a patient, the method comprising the following steps: receiving a measurement dataset including energy resolved data based on a computed tomography scan (or CT scan) of a patient; - reconstructing a morphology-preserving image dataset based on a first photon-energy band or a first combination of photon-energy bands described by said measurement dataset; - segmenting the blood pool within the myocardium in said morphology-preserving image dataset; - reconstructing a contrast agent map based on a second photon-energy band or a second combination of photon-energy bands described by said measurement data set; - determining a reference value based on at least one pixel or voxel of the segmented blood pool contrast agent map; - determining each myocardial extracellular volume fraction based on a value assigned to at least one pixel or voxel outside the blood pool segmented by said contrast agent map and said reference value; Includes.

[0016] When resolving multiple, preferably at least three, photon-energy bands (also called energy bins), different energy bands and / or different combinations of energy bands can be used during reconstruction to generate different image impressions. To combine energy bands, for example, it is possible to add or subtract images reconstructed from different energy bands. Two possible impressions are the aforementioned contrast agent map and tissue or water map. The contrast of such materials is based on the fact that most contrast agents (especially iodine) have X-ray absorption that is strongly dependent on photon energy. Therefore, a contrast agent map can be determined, for example, from the difference in absorption in different photon-energy bands. Once the absorption by the contrast agent is known, the remaining absorption can be attributed to water and tissue.

[0017] Obviously, the difference between different materials can be further improved if the difference between X-ray absorption for more than two photon-energy bands is taken into account.

[0018] By selecting the correct photon-energy band or combination of photon-energy bands, it is possible to generate a morphology-preserving image dataset, which allows the use of conventional segmentation algorithms, typically used to segment image datasets acquired by coronary CT angiography, and thus image datasets recorded after a relatively short waiting period after injection of contrast agent, thus segmenting an image dataset in which essentially all of the contrast agent is still contained within the blood pool and has essentially not migrated to the tissues.

[0019] The first photon-energy band, or first combination of photon-energy bands, is selected to provide a significantly higher contrast-to-noise ratio for the morphological features to be segmented, particularly for the blood pool and / or myocardium, e.g., at least two or three times higher than that in the contrast agent map. As described in more detail below, it has been found that this can be achieved by including only, or primarily, photons with photon energies at the lower end of the available spectrum. The myocardium and blood pool considered may in particular be the left ventricular myocardium and the blood pool therein.

[0020] In classical dual-energy or multi-energy computed tomography acquisitions, the accelerating voltage can be modified or multiple X-ray sources can be used to generate photons of different photon energies. Alternatively or additionally, multiple detectors with different sensitivity bands can be used in a sandwich configuration to directly measure the X-ray intensity in two or more given energy bands. In either case, the detectors measure the intensity for different photon-energy bands.

[0021] In a photon-counting detector, the photon-energy of each detected photon can be determined, at least approximately, because the height of the current spike that results upon detection of each photon depends on the photon-energy of that photon. Thus, for example, individual photon detections within each pixel of the detector can be logged along with their associated photon energy, followed by essentially arbitrary binning of events based on detected energy to provide multiple photon-energy bands.

[0022] As mentioned above, the presence of a contrast agent can be detected particularly if the local absorption varies strongly with photon energy, therefore it is preferred to reconstruct the contrast agent map based on a combination of at least two, preferably more, photon-energy bands described in the measurement data set.

[0023] It may be sufficient to use a single photon-energy band to reconstruct a morphology-preserving image dataset. However, it is also possible for both the morphology-preserving image dataset and the contrast agent map to be based on a single photon-energy band, or for both to be based on a combination of multiple photon-energy bands. Essentially, it is also possible to determine a morphology-preserving image dataset based on a combination of multiple photon-energy bands, while the contrast agent map is based on a single photon-energy band. The different photon-energy bands may not overlap, or at least one pair of the photon-energy bands may overlap.

[0024] The morphology-preserving image datasets and contrast agent maps are preferably three-dimensional image datasets, in particular formed from a three-dimensional array of voxels. Alternatively, they may be two-dimensional image datasets, for example formed from an array of pixels. The measurement dataset may include multiple projection images. For example, it may include a single projection image for each photon energy band and projection angle, or it may include, for example, a respective event list containing all individual photon detection events for each detector pixel and projection angle.

[0025] Image acquisition is preferably completed before the start of the computer-implemented method, rather than as part of the method, or image acquisition can be included as an additional step, with the acquired image data being received directly from, for example, a computed tomography scanner.

[0026] Preferably, all the above-described steps of the computer-implemented method are performed in a fully automated manner. Optionally, a user may be allowed to interact, for example, to modify the results of any step. For example, the selection of the energy band or bands used to reconstruct the morphology-preserving image dataset, the blood pool segmentation, and / or the myocardium segmentation may be user modifiable.

[0027] Preferably, however, the segmentation of the blood pool and / or myocardium, or at least the initial segmentation that can be modified by the user, is performed automatically without user interaction. During the development of the present invention, the automatic segmentation method for three-dimensional cardiac computed tomography volumes, discussed in the article by Y. Zheng cited at the beginning, was primarily adopted. However, in CT angiography, many additional methods exist for automatically segmenting contrast-enhanced images, which can typically be used directly to segment the topology-preserving image datasets of the present invention without further modification.

[0028] Preferably, the measurement dataset is based on a photon-counting computed tomography scan. In non-photon-counting multi-energy computed tomography scans, the weight (or weighting) of lower energy photons is reduced within a given energy band due to the measurement principle used. This effect can be avoided in a photon-counting computed tomography dataset because individual photons are counted and photon energy is detected as another parameter of detection. Lower energy photons are particularly relevant for preserving the depicted cardiac morphology and therefore for robust segmentation of the myocardium and blood pool. Using a photon-counting computed tomography scan as the basis for the input dataset allows for more robust morphology preservation, in particular a higher contrast-to-noise ratio in the morphology-preserving image dataset, which allows for more accurate and more robust automatic segmentation.

[0029] Furthermore, the use of photon-counting computed tomography allows essentially free and dynamic selection of the photon-energy bands used in reconstructing the morphology-preserving image dataset, on the one hand, and the contrast agent map, on the other hand, as needed. The only limitations on the number of photon-energy bands used are sufficient photon counts within each band and sufficient energy resolution of the detector. While classical multi-energy computed tomography is limited to two, or at most three or four, photon-energy bands, which are typically fixed, photon-counting computed tomography allows for a greater number of bands and / or dynamic adjustment of the position and width of different energy bands to provide optimal results. This can be used to further improve the contrast-to-noise ratio in the morphology-preserving image dataset and / or to further improve material selectivity for the contrast agent map.

[0030] Preferably, the myocardium is segmented in the morphology-preserving image dataset, with each myocardial extracellular volume fraction being determined only for pixels or voxels within the segmented myocardium. Additionally or alternatively, an extracellular volume image dataset can be generated that includes a respective myocardial extracellular volume fraction for each pixel or voxel within the segmented myocardium and a value based solely on the morphology-preserving image dataset or a fixed value for each pixel or voxel outside the segmented myocardium. Determining the extracellular volume fraction outside the myocardium is typically not necessary or useful, for example, when attempting to detect myocardial fibrosis. At the same time, generating and displaying such values ​​in regions outside the myocardium can significantly reduce the available contrast, particularly since the above-described mathematical formula for calculating the extracellular volume fraction tends to lead to a maximum value in the blood pool region and, therefore, outside the region of interest.

[0031] Using fixed values ​​for pixels or voxels outside the segmented myocardium can be useful, for example, to generate a three-dimensional dataset with full transparency within those regions. Such a dataset can be useful for facilitating visualization, for example, to enable robust mapping to a two-dimensional coordinate system (e.g., using a two-dimensional polar map) and / or to generate an overlay that can be superimposed on a morphology-preserving image dataset. Such a superimposition can allow a physician or examiner to easily orient themselves within an image, for example, based on the recognizability of morphological features. Such a superimposition can use different color palettes for the projections generated from the morphology-preserving image dataset and the projections generated from the extracellular volumetric image dataset. For example, a color image can be generated from the extracellular volumetric image dataset using a given palette and then superimposed with an image generated from the morphology-preserving image dataset using a grayscale palette.

[0032] As mentioned above, alternatively, a combination of these two data sources in a single two-dimensional image data set or three-dimensional image data set can be used.

[0033] Alternatively or additionally, restricting the determination of myocardial extracellular volume fraction to the volume actually covered by the myocardium following segmentation, or masking the generated values ​​to leave only these values, may be advantageous for statistical analysis of the myocardial extracellular volume fraction, since the analysis can then be restricted to the relevant volume. Simple examples of such statistical analysis include calculating the mean and / or median of the myocardial extracellular volume fraction and / or determining a measure describing the distribution (e.g., standard deviation, etc.).

[0034] The reference value can be determined based on the average or median value of multiple pixels or voxels, but is preferably determined based on the average or median value of all pixels or voxels of the contrast agent map within the segmented blood pool. Instead of using all pixels or voxels within the segmented blood pool, it is possible to discard a given number of pixels or voxels in the outer rows of the multiple pixels or voxels, in which case the impact of potential segmentation errors can be reduced. The use of the average or median value is beneficial because the distribution of contrast agent in blood is expected to be essentially uniform within a given region, and therefore the determination of the average or median value can be used to further reduce errors.

[0035] The measurement dataset can include energy-resolved data based on computed tomography scans of the patient performed at least 2 minutes, or at least 3 minutes, and / or at most 6 minutes or at least 5 minutes after the start of contrast injection. If the waiting time between contrast injection and image acquisition is too short, equilibrium between the concentration of the contrast agent in the extracellular volume of the tissue and its blood has not yet been reached because the contrast agent has insufficient time to diffuse from the blood to the tissue. On the other hand, a very long waiting time will not significantly improve equilibration. Rather, the overall concentration of the contrast agent will typically decrease, which can lead to increased noise.

[0036] The average photon energy within the first photon-energy band or the first combination of photon-energy bands may be less than 70 keV or less than 50 keV. As described above, the contrast-to-noise ratio in the morphology-preserving image dataset can be increased when primarily using photons with relatively low photon energies. For example, a single first photon-energy band centered at 40 keV can be used. The width of the first photon-energy band, or the energy spread covered by the first combination of photon-energy bands, may be, for example, less than 30 keV or even less than 10 keV.

[0037] As mentioned above, when using photon-counting computed tomography, the position and width of the assumed photon-energy band can essentially be freely selected. If necessary, for example, photon detection from several adjacent pixels of the detector can be pooled to increase the overall photon count, thus, for example, allowing the use of narrower photon-energy bands. Since the photon energy of individual photons can be detected during photon-counting computed tomography, it is also possible in principle to continuously vary the weighting coefficients of individual photons within the assumed photon-energy band.

[0038] The second photon-energy band or second combination of photon-energy bands can include at least one photon energy higher than the highest photon energy in the first photon-energy band or first combination of photon-energy bands. Using a relatively wide spread of photon energies to determine the contrast agent map is advantageous because the contrast of the material used to generate the contrast agent map depends primarily on detecting the different absorption behavior of the contrast agent at different photon energies. Therefore, it is advantageous to use absorption in multiple photon-energy bands that are preferably as far apart as possible.

[0039] The contrast agent map can be based on a second combination of photon-energy bands, where the second combination of photon-energy bands includes at least one photon-energy band or at least two photon-energy bands that do not overlap with any photon-energy bands of the first photon-energy band or the first combination of photon-energy bands. The second combination of photon-energy bands can further include at least one photon-energy band of the first photon-energy band or the first combination of photon-energy bands. As mentioned above, it is advantageous to consider at least two, and preferably more, photon-energy bands spread across a wide energy spectrum to determine x-ray absorption at multiple photon energies, thus clearly and robustly distinguishing between different materials, particularly contrast agents, and water or tissue.

[0040] At least one myocardial extracellular volume fraction can be determined from energy decomposition data based on a single computed tomography scan of the patient to form a measurement dataset and the provided hematocrit value, without using other input data such as the measurement dataset and the hematocrit value. Using a single computed tomography scan avoids potential artifacts due to registration when performing multiple scans. In the computer-implemented method according to the present invention, the topography-preserving image dataset and the contrast agent map are inherently registered to each other because they are essentially based on the same received measurement dataset, which corresponds to a single computed tomography scan. This makes the determination more robust and less prone to errors than the above-described method, which requires registration of multiple computed tomography scans.

[0041] A representation of the segmented blood pool, particularly the contours of the segmented blood pool in the morphology-preserving image dataset, and / or a representation of the segmented myocardium, particularly the contours of the segmented myocardium in the morphology-preserving image dataset, can be displayed to a user, who can then edit the segmentation of the blood pool and / or myocardium and subsequently use the edited segmentation to determine a baseline value and / or at least one myocardial extracellular volume fraction. The automated segmentation of the blood pool and myocardium in the above-described method is highly robust, and the outcome of the method can be improved by a physician or other individual reviewing intermediate results and potentially modifying the results, if necessary.

[0042] Alternatively or additionally, a user can interactively change the first photon-energy band or first combination of photon-energy bands used to reconstruct the morphology-preserving image dataset. For example, a user may display at least one projection image of the initially generated morphology-preserving image dataset and use a fader or other control to slightly shift the first photon-energy band used, for example, until optimal contrast is reached. Alternatively, such optimization can be performed automatically.

[0043] In addition to the computer-implemented methods described above, the present invention also relates to a processing system, which includes: a first interface configured to receive a measurement dataset including energy-resolved data based on a computed tomography scan of a patient; a second interface configured to provide at least one myocardial extracellular volume fraction for the patient; and a computing unit configured to perform any of the computer-implemented methods described above; and Includes.

[0044] The processing system may, for example, be part of a computed tomography scanner or part of a workstation used to control the scanner, or it may be a separate workstation not used to control the scanner, or it may be implemented as a cloud-based solution.

[0045] The present invention also relates to a medical imaging device including an energy-resolved computed tomography scanner and a processing system according to the present invention. As mentioned above, preferably the computed tomography scanner is a photon-counting computed tomography scanner. Alternatively, a non-photon-counting multi-energy computed tomography scanner may be used.

[0046] Furthermore, the present invention may relate to a computer program product, which comprises instructions (or directions) for causing a processing unit (or processing device) to perform a computer-implemented method according to the present invention.

[0047] Furthermore, the invention may relate to a computer readable storage medium, which comprises stored therein a computer program according to the invention.

[0048] In particular, features and advantages described with respect to a computer-implemented method according to the invention may also be configured as corresponding subunits of a device (or apparatus) according to the invention or as corresponding subunits of a computer program according to the invention, and conversely, features and advantages described with respect to a device according to the invention or features and advantages described with respect to a computer program according to the invention may also be configured as corresponding processing steps of a method according to the invention.

[0049] Other objects and features of the present invention will become apparent from the following description, which is given with reference to the accompanying drawings. However, these drawings are merely basic sketches constructed solely for the purpose of illustrating the present invention, and are not constructed for the purpose of limiting the present invention. In these drawings, the following are illustrated: [Brief explanation of the drawings]

[0050] [Figure 1] FIG. 1 is a flowchart diagram of an exemplary embodiment of a computer-implemented method according to this invention. [Figure 2] FIG. 2 is a diagram of relevant data structures used in an exemplary embodiment of a computer-implemented method according to this invention. [Figure 3] FIG. 3 is a diagram illustrating the segmentation of the myocardium and blood pool within the heart, shown from a different perspective than FIG. [Figure 4] FIG. 4 is a diagram illustrating the segmentation of the myocardium and blood pool within the heart, shown from a different perspective than FIG. [Figure 5] FIG. 5 is a diagram of an exemplary embodiment of a medical imaging device in accordance with the present invention, including an exemplary embodiment of a processing system in accordance with the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0051] A flowchart of a computer-implemented method for determining at least one myocardial extracellular volume fraction (ECV) for a patient is illustrated in Figure 1. The method is further described with reference to Figure 2, which illustrates associated data structures used during the determination process.

[0052] The basic idea of ​​the method, which will be explained in more detail below, is to use different photon-energy bands 9, 14, 15, 16 and / or combinations 13 of (multiple) photon-energy bands 14, 15, 16 to generate, on the one hand, a morphology-preserving image data set 8 and, on the other hand, a contrast agent map 12 from the same measurement data set 3, which includes energy-resolved data based on a computed tomography scan of the patient. The contrast agent map 12 allows to determine the change in X-ray absorption in each pixel (or picture element) or voxel (or volume element) 19, expressed in particular as a change in Hounsfield units, due to the presence of a contrast agent in the region of the patient depicted by each voxel 19. The morphology-preserving image dataset 8 is further used to enable a robust automatic segmentation (or partitioning) of the patient's myocardium, preferably the myocardium 11 of the left ventricle of the patient's heart, and the blood pool 10 therein. It is therefore possible to determine the changes in X-ray absorption in the blood pool 10 and the myocardium 11, and based on these values, the extracellular volume fraction 1 of the myocardium can be calculated.

[0053] To determine the myocardial extracellular volume fraction 1, first, in step S1, a measurement dataset 3 is received. This measurement dataset 3 includes energy-resolved data based on a computed tomography scan of the patient. For example, this measurement dataset 3 is acquired using a photon-counting computed tomography scan. This measurement dataset 3 may be provided directly by an energy-resolved computed tomography scanner, or the processing described below may be performed at a later time, and the measurement dataset 3 may be retrieved, for example, from a database.

[0054] The computed tomography scan to acquire the data set is preferably performed when there is approximate equilibrium between the contrast agent in the blood vessels and the contrast agent in the extracellular volume fraction of the tissue. Such equilibrium is typically reached approximately 3 to 5 minutes after the start of contrast agent injection. Therefore, the scan is preferably performed within this time interval.

[0055] By using a photon-counting detector, for each pixel 5 of the detector and each projection 4 acquired, the measurement dataset 3 contains a list of detection events 6 for each pixel 5 and projection 4. Since a photon-counting tomography scan allows for the acquisition of the photon-energy for each detected photon, each detection event 6 also describes a measurement of the photon-energy 7 of the detected photon.

[0056] Alternatively, it is also possible to provide a measurement data set 3 in which the binning of photon energies 7 into each photon-energy band has already been performed. Such a measurement data set 3 can, for example, provide a photon count for each energy band, pixel 5 and projection 4. Instead of a photon-counting computed tomography scan, it is also possible to use a conventional (classical) multi-energy acquisition which provides the detected intensity only for each given projection 4, pixel 5 and energy band.

[0057] In step S2, a morphology-preserving image dataset 8 is reconstructed. For example, this reconstruction is based only on photons detected within a given photon-energy band 9. This may be achieved, for example, by considering only those detection events 6 in the measurement dataset 3, where a given photon-energy 7 is present within the given photon-energy band 9. Other detection events 6 are ignored, so that they are essentially discarded during the reconstruction of the morphology-preserving image dataset.

[0058] The count (or enumeration) of the remaining detection events 6 for a given pixel 5 and projection 4 may then be considered directly to describe the X-ray intensity in a given photon-energy band 9, or further processing (e.g., energy weighting) may be applied. The resulting number of projection images considering only photon-energies 7 within a given photon-energy band 9 may then be used to reconstruct a morphology-preserving image dataset 8 using well-known image reconstruction techniques (e.g., filtered backprojection or iterative reconstruction approaches).

[0059] In other embodiments, it is also possible to consider multiple photon-energy bands when reconstructing the morphology-preserving image dataset 8. To this end, for example, one may calculate counts or intensities for each pixel 5, each projection 4, and each energy band, and then, for example, calculate a weighted sum (or weighted sum) of the counts or intensities during reconstruction.

[0060] If individual detection events 6 are not available (e.g., due to pre-binning of events based on photon-energy 7 or due to direct detection of intensities within several energy bands), the counts or intensities provided in the measurement dataset 3 for the relevant photon-energy bands 9 may be used directly to reconstruct a morphology-preserving image dataset 8.

[0061] In step S3, a contrast agent map 12 is reconstructed based on a combination 13 of multiple photon-energy bands 14, 15, 16. As described above for the photon-energy bands 9, the projections 4, pixels 5, and photon counts or intensities for each of the energy bands 14, 15, 16 may be determined or provided from the measurement dataset 3. Because the material contrast of contrast agents (e.g., iodine) is typically based on the fact that contrast agents have X-ray absorption that varies strongly with photon energy, the combination of the counts and / or intensities in the various photon-energy bands 14, 15, 16 may advantageously be more complex than a weighted sum and may, for example, include determining the difference between the intensities or counts for the different photon-energy bands 14, 15, 16 and / or evaluating the relative intensities or counts determined for the different photon-energy bands 14, 15, 16. Determining the material contrast of contrast agents (particularly iodine) based on intensity or photon counting for multiple photon-energy bands is known in the art, and there are many different ways to determine such material contrast that are used in various existing products from various manufacturers. Therefore, the determination of values ​​describing material contrast, and in particular the change in Hounsfield unit values ​​due to X-ray absorption by the contrast agent, will not be detailed.

[0062] Considering the different photon-energy bands 9, 14, 15, 16 used in the example, photon-energy band 9 preferably contains only photons with relatively low photon-energy 7. Photon-energy band 9 may be centered, for example, at 40 keV or 50 keV and can have a width, for example, from 10 keV to 30 keV.

[0063] Preferably, photon-energy bands 14, 15, 16 span the available spectrum, for example from 40 keV to 120 keV. For example, photon-energy band 14 may be identical to photon-energy band 9 or may partially overlap with photon-energy band 9. Preferably, photon-energy bands 15, 16 include detected photons having photon-energy 7 above photon-energy band 9, and preferably are spaced apart to describe the absorption of contrast agents in different parts of the spectrum.

[0064] In steps S4 and S5, automatic segmentation is applied to the morphology-preserving image dataset 8. In step S4, the blood pool 10 is segmented, and in step S5, the myocardium 11 is segmented. For ease of explanation, these segmentations are shown as separate steps, but advantageously both segmentations may be performed simultaneously using a common segmentation algorithm.

[0065] In this example, the segmentation algorithm described in the paper by Y. Zheng cited at the beginning is used. Essentially, any segmentation algorithm can be used to segment the cardiac computed tomography volume. An example of such a segmentation is illustrated schematically in FIGS. 3 and 4. FIG. 3 illustrates a short-axis view of the heart, and FIG. 4 illustrates a long-axis view of the heart. The segmentation algorithm used allows for the automatic segmentation of five cardiac regions within the volume depicted in the morphology-preserving image dataset 8. These regions are shaded in FIGS. 3 and 4. These regions correspond to the left ventricular myocardium 11, the left ventricular blood pool 10, the right ventricle 21, and the left and right atria 22, 23. Other exemplary structures are shown in FIGS. 3 and 4 but are not segmented.

[0066] Segmentation of the right ventricle 21 and the left and right atria 22, 23 is not used in the embodiment described in this method: segmentation of these regions may be skipped or the segmentation of these regions may be discarded.

[0067] The morphology-preserving image dataset 8 and the contrast agent map 12 are determined from the same measurement dataset, i.e., from a single computed tomography scan of the patient, and therefore the contrast agent map 12 is inherently registered with the morphology-preserving image dataset 8. Thus, each voxel 19 of the contrast agent map 12 corresponds to a corresponding voxel 19 in the morphology-preserving image dataset 8 at the same location and describes the same region of the patient's heart. Therefore, the segmentation of the blood pool 10 and myocardium 11 in the morphology-preserving image dataset 8 also describes the segmentation of the blood pool 10 and myocardium 11 in the contrast agent map 12.

[0068] Thus, using the segmentation of the blood pool 10, it is possible to select voxels 19 of the contrast agent map 12 that are located in this blood pool 10. The image data of each voxel is indicative of the change in X-ray absorption due to the contrast agent, and in particular the change in Hounsfield units due to the contrast agent. Therefore, by averaging the image data from these voxels 19, a reference value 17 can be determined, which in step S6 describes the change in X-ray absorption in the blood pool due to the presence of the contrast agent. If this reference value 17 describes the change in Hounsfield units, then ΔHU b It may be labeled as

[0069] In step S7, voxels 19 in the contrast agent map 12 that depict the left ventricular myocardium 11 are selected using the segmentation of the myocardium 11. The image data of these voxels 19 describes the change in X-ray absorption at each location in the tissue of the myocardium 11, and therefore ΔHU tWhen these changes are given in Hounsfield units, the extracellular volume fraction (EVC) can be calculated for the area of ​​myocardial tissue depicted within each voxel 19 according to the formula given above. CT ) has already been described.

[0070]

number

[0071] In step S8, the hematocrit value 20, labeled h in the given formula, is provided, e.g., from patient data records. Therefore, all terms on the right hand side of the given formula are known, allowing the myocardial extracellular volume fraction 1 to be calculated in step S9 for a given voxel 19 in the segmented myocardium 11.

[0072] To more easily visualize the determined, locally resolved myocardial extracellular volume fraction 1, an extracellular volume image dataset 18 may be generated, which includes the myocardial extracellular volume fraction 1 for each voxel 19 within the segmented myocardium 11. Voxels outside the segmented myocardium 11 may be set to a fixed value, e.g., to be fully transparent. The extracellular volume image dataset 18 may, for example, be used to generate projections superimposed on corresponding projections of the morphology-preserving image dataset 8, in particular using a different color palette as described above in the general description. The extracellular volume image dataset 18 is automatically registered to the morphology-preserving image dataset 8, so no additional registration is required.

[0073] Additionally or alternatively, statistical analysis of the determined myocardial extracellular volume fractions 1 can be performed, for example, by calculating the mean value or selecting the median value from these determined fractions, determining the parameters of the distribution, or analyzing the regional variance of this fraction.

[0074] Although the above-described method employs a fully automated approach to determining the myocardial extracellular volume fraction 1, in some cases it may be advantageous for a user (or administrator) to be able to modify the results of any step. For example, segmentations of the blood pool 10 and myocardium 11 may be output to the user (e.g., using the exemplary diagrams in Figures 3 and 4) so ​​that the segmentations can be modified by the user. The modified segmentations may then be used in further steps.

[0075] 5 illustrates a medical imaging device (or imaging apparatus) 28 including an energy-resolved computed tomography scanner 29 (preferably a photon-counting computed tomography scanner) and a processing system 30. The processing system 30 comprises a first interface 24 configured to receive the measurement dataset 3 and a second interface 25 configured to provide the myocardial extracellular volume fraction 1 (particularly in the form of an extracellular volume image dataset 18).

[0076] In this embodiment, the measurement dataset 3 is provided by the energy-resolved computed tomography scanner 29 immediately after scanning the patient 2. Alternatively, it is also possible to store the measurement dataset 3 intermediately in the scanner 29 itself or, for example, in a patient database. The determined myocardial extracellular volume fraction 1 is stored in the exemplary patient database 27, in particular in the form of an extracellular volume image dataset 18. Alternatively or additionally, information about the myocardial extracellular volume fraction 1 can be output directly to the user, for example by means of a screen of a processing device 26 implementing the processing system 30.

[0077] Although the present invention has been described in detail with reference to preferred embodiments, the present invention is not limited to the disclosed embodiments, and those skilled in the art will be able to derive other variations from the disclosed embodiments without departing from the scope of the present invention.

Claims

1. 1. A computer-implemented method for determining at least one myocardial extracellular volume fraction (1) for a patient (2), comprising: - receiving a measurement dataset (3) comprising energy resolved data based on a one-time dual-energy computed tomography scan of a patient (2); - reconstructing a morphology-preserving image dataset (8) based on a first photon-energy band (9) or a first combination of photon-energy bands described by said measurement dataset (3); - segmenting the blood pool (10) in the myocardium (11) within said morphology-preserving image dataset (8); - reconstructing a contrast agent map (12) based on a second photon-energy band or a second combination (13) of photon-energy bands (14, 15, 16) described by said measurement data set (3); - determining a reference value (17) based on at least one pixel or voxel (19) of said contrast agent map (12) within said segmented blood pool (10); - determining a respective myocardial extracellular volume fraction (1) based on a value assigned to at least one pixel or voxel (19) outside the blood pool (10) segmented by said contrast agent map (12) and on said reference value (17); A method including each of the steps.

2. 2. The computer-implemented method of claim 1, wherein the measurement data set (3) is based on a photon-counting computed tomography scan.

3. The myocardium (11) is segmented in the morphology-preserving image dataset (8), wherein: determining a respective myocardial extracellular volume fraction (1) only for pixels or voxels (19) within the segmented myocardium (11); and / or 2. The computer-implemented method of claim 1, wherein an extracellular volume image dataset (18) is generated, the extracellular volume image dataset (18) including a respective myocardial extracellular volume fraction (1) for each pixel or voxel (19) within the segmented myocardium and a value or fixed value based solely on the morphology-preserving image dataset for each pixel or voxel (19) outside the segmented myocardium (11).

4. 2. The computer-implemented method of claim 1, wherein the reference value (17) is determined based on an average or median value of a plurality of pixels or voxels (19).

5. 2. The computer-implemented method of claim 1, wherein the measurement dataset (3) comprises energy-resolved data based on a computed tomography scan of the patient (2) performed at least 2 minutes and at most 6 minutes after the start of injection of a contrast agent.

6. 2. The computer-implemented method of claim 1, wherein the average photon energy in the first photon-energy band or first combination of photon-energy bands is less than 70 keV.

7. 2. The computer-implemented method of claim 1, wherein the second photon-energy band or second combination (13) of photon-energy bands (14, 15, 16) includes at least one photon energy (7) higher than the highest photon energy (7) in the first photon-energy band (9) or first combination of photon-energy bands.

8. 2. The computer-implemented method of claim 1, wherein the contrast agent map (12) is based on a second combination (13) of a plurality of photon-energy bands (14, 15, 16), wherein the second combination (13) of the plurality of photon-energy bands (14, 15, 16) includes at least one photon-energy band (15, 16) or at least two photon-energy bands (15, 16) that do not overlap with the first photon-energy band (9) or any photon-energy band of the first combination of the plurality of photon-energy bands.

9. 2. The computer-implemented method of claim 1, wherein the at least one myocardial extracellular volume fraction (1) is determined from energy-resolved data based on a single computed tomography scan of the patient (2) to form the measurement data set (3) and a provided hematocrit value (20), without using other input data from the measurement data set (3) and the hematocrit value (20).

10. - displaying the segmented blood pool (10), in particular the contour of the segmented blood pool (10) in the morphology-preserving image dataset (8); and / or a representation of the segmented myocardium (11), in particular a representation of the contours of the segmented myocardium (11) in the morphology-preserving image dataset (8), is displayed to a user; 2. The computer-implemented method of claim 1, wherein a user can edit the segmentation of the blood pool (10) and / or the myocardium (11), and then each of the edited segmentations is used to determine the reference value (17) and / or the at least one myocardial extracellular volume fraction (1).

11. 1. A processing system comprising: a first interface (24) configured to receive a measurement dataset (3) including energy resolved data based on a computed tomography scan of a patient (2); a second interface (25) configured to provide at least one myocardial extracellular volume fraction (1) for a patient (2); a computing unit (26) configured to perform the computer-implemented method according to any one of claims 1 to 10; a processing system including:

12. 1. A medical imaging device, comprising: an energy-resolved computed tomography scanner (29); A processing system (30) according to claim 11; 1. A medical imaging device comprising:

13. A computer program comprising: A computer program comprising instructions for causing a processing device to perform the computer-implemented method of any one of claims 1 to 10.

14. 1. A computer-readable storage medium, comprising: A computer readable storage medium having stored thereon a computer program (12) according to claim 13.

Images