Technique for spectral mapping based on image generation from energy-resolved medical imaging

Abstract

The invention relates to a technique for generating a medical image from a spectral map produced from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique. A computer-implemented method (100), performed by a computing device (200), comprises receiving (S102) multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique. An analysis scope of the received (S102) multispectral raw medical imaging data is determined (S106). At least one spectral map to be generated from the received (S102) multispectral raw medical imaging data is selected (S108). The selection (S108) is based on the determined (S106) analysis scope. The at least one selected (S108) spectral map is generated (S110).The received (S102) multispectral medical raw imaging data are processed (S112) to generate (S114) at least one medical image based on the at least one generated (S110) spectral map.

Description

[0001] The present invention relates to a technique for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, in particular comprising a method, a computing device, a system comprising the computing device, and a computer program product.

[0002] Photon-counting computed tomography (CT) is a form of CT in which X-rays are detected using a photon-counting detector that records the interactions of individual photons. By tracking the energy emitted in each interaction, the detector pixels record an approximate energy spectrum, making it a spectral or energy-resolved CT technique.

[0003] In clinical applications, photon counting can offer significant advantages. Compared to images from conventional CT scanners, clinical images generated using photon-counting detectors are characterized by high resolution and improved contrast. The technology also paves the way for substantial further reductions in radiation dose during CT scans. Imaging with photon-counting detectors is therefore less taxing on patients and offers physicians a genuine alternative for screening programs and follow-up examinations, for example, in cancer treatment. Furthermore, multispectral CT data, due to their inherently energy-resolved nature, offer an additional dimension. They redefine clinical decision-making by providing all relevant CT findings in a single scan.

[0004] However, in everyday clinical practice, i.e., during the routine evaluation of medical images, fully exploiting the potential of multispectral images presents a significant challenge. This is because the additional dimension in the data also adds another dimension to the clinician's decision-making process when evaluating multispectral medical images. Traditionally, clinicians must select appropriate spectral ranges and apply the correct image processing tools to draw the right conclusions. Missing something here can conventionally mean missing clinically relevant results, with potentially serious consequences for the patient.

[0005] It is therefore an objective of the present invention to provide a solution for improving image generation based on multispectral medical raw imaging data, particularly with regard to higher resolution and / or improved contrast. Alternatively or additionally, an objective of the present invention is to provide the clinician with reliable support through automatic processing of multispectral image data in order to provide improved medical diagnosis and / or to enable improved decision-making regarding a possible treatment plan. Alternatively or additionally, an objective of the present invention is to improve medical image acquisition (e.g., with regard to reduced acquisition time and / or the need for only a single scan compared to conventional methods, e.g.,to enable reductions in radiation doses and / or reductions in the administration of contrast agents (paired scans), resulting in a reduced health risk and / or reduced burden for the patient.

[0006] This objective is achieved by a method for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using energy-resolved medical imaging technology, by a computing device, by a system, by a computer program (and / or a computer program product), and by a computer-readable storage medium according to the attached independent claims. Advantageous aspects, features, and embodiments are described in the dependent claims and in the following description together with advantages.

[0007] The solution according to the invention is described below with reference to the claimed method for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, and with reference to the claimed computing device. Features, advantages, or alternative embodiments herein may be assigned to the other claimed subject matter (e.g., the system, the computer program, or a computer program product) and vice versa. In other words, claims for the computing device and / or the system may be enhanced with features described or claimed in the context of the method. In this case, the functional features of the method are implemented by structural units of the system, and vice versa.

[0008] Regarding a first aspect, a (particularly computer-implemented) method for generating a medical image from a spectral map is provided. This map is generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique. The method is performed by a computing device. The method comprises a step for receiving multispectral raw medical imaging data. The multispectral raw medical imaging data was acquired using an energy-resolved medical imaging technique. The method further comprises a step for determining the scope of analysis of the received multispectral raw medical imaging data. The method also comprises a step for selecting at least one spectral map to be generated from the received multispectral raw medical imaging data.The selection of the at least one spectral map to be generated is based on the specified scope of analysis. The method further comprises a step for generating the at least one selected spectral map. The method also includes a step for processing the received multispectral raw medical imaging data to generate at least one medical image based on the at least one generated spectral map. Alternatively or additionally, at least one medical image can be generated from the at least one generated spectral map (e.g., by processing the at least one generated spectral map).

[0009] The inventive technique enables the automatic processing (and / or processing by a computer implementation) of multispectral raw medical imaging data acquired using spectral and / or energy-resolved medical imaging techniques. The resulting medical images exhibit higher resolution and / or improved contrast compared to conventional medical images acquired without energy resolution. Acquiring multispectral raw medical imaging data offers the advantage that all relevant data can be acquired in a single scan and / or with a reduced need for radiation, e.g., with a total X-ray dose that is lower than conventionally obtained.

[0010] The technology according to the invention enables the selection of a suitable spectral range for diagnosis and / or decision-making and / or the selection of suitable image processing tools, in particular automatically and / or in a computer-implemented manner.

[0011] Energy-resolved medical imaging techniques may include (or be) computed tomography (CT), in particular photon-counting CT. Alternatively or additionally, energy-resolved medical imaging techniques may include (or be) magnetic resonance imaging (MRI), such as MRI with two or more different B0 fields or MRI with variable acquisition sequences using different magnetic field strengths. Furthermore, alternatively or additionally, energy-resolved medical imaging techniques may include (or be) energy-resolved neutron imaging (e.g., using the RADEN energy-resolved neutron imaging system, as described in [8], which is hereby incorporated by reference).Alternatively or additionally, energy-resolved medical imaging techniques may include (or be) dual-energy X-ray imaging, photon-counting X-ray imaging, phase-contrast X-ray imaging, and / or energy-resolved X-ray imaging (as described, for example, in [9], which is hereby incorporated by reference). X-ray imaging may also be referred to as (e.g., digital) radiography and / or may include (especially two-dimensional, 2D) projection imaging. Any X-ray imaging technique may be similar to (and / or use the same technique as) CT, but for conventional radiography, enabling the generation of some contrasts with only one image acquisition.

[0012] Energy-resolved medical imaging technology can alternatively or additionally include (or be) photoacoustic imaging (also called optoacoustic imaging), which can be energy-resolved in the sense that different wavelengths, e.g., energy levels of visible and / or near-infrared light (e.g., provided by non-ionizing laser pulses), are used to acquire the medical imaging data.

[0013] Energy-resolved medical imaging techniques can include a combination of two different energy-resolved medical imaging techniques, e.g., photon-counting CT and (especially energy-resolved) MRI.

[0014] The multispectral medical raw imaging data may relate to a patient and / or an individual (especially a living one), a human and / or an animal.

[0015] Multispectral medical raw imaging data can include, for example, the amount of energy emitted per unit area, as in single-photon CT. The energy emitted can correspond to the energy of a single photon (and / or neutron), which is proportional to its wavelength (and / or spectral color). The amount of energy emitted can be measured precisely or per predetermined energy bin. The unit area can be a pixel of the (e.g., X-ray) detector.

[0016] Multispectral medical raw imaging data can be acquired by using multispectral imaging acquisition data at different wavelengths, each of which interacts differently with biological tissues.

[0017] For example, different wavelengths can penetrate tissue to varying depths, revealing details that might not be visible at other wavelengths.

[0018] The multispectral medical raw imaging data can include, for example, k-space data acquired using an MRI scanner.

[0019] The spectral data allows tissue types to be distinguished based on their specific energy absorption characteristics (especially for CT) and / or based on their specific Larmor frequency (especially for MRI).

[0020] The scope of analysis of the received multispectral medical raw imaging data may include (or be contained within) a dataset that specifies a medical use case and / or a reason why the medical imaging was performed, such as a dataset that specifies an anatomical structure to be imaged and / or a region of interest (RoI), for example, with regard to a possible diagnosis and / or with regard to symptoms described by a patient.

[0021] At least one spectral map (or simply map) can include a predetermined functional dependence of the resolved energy and / or the spectrum per unit area or unit volume (e.g., per pixel or per voxel).

[0022] For multispectral medical raw imaging data using CT, a spectral map may include, for example, a virtual monoenergetic image with a predefined monoenergy (keV) level, an iodine map, an effective atomic number map, a virtual non-enhanced map and / or a virtual non-contrast map, as further described in [1], [2], [3], [4], all of which are hereby incorporated by reference.

[0023] For multispectral medical raw imaging data using MRI, a spectral map may include, for example, a proton density-fat fraction (FF) map (e.g., on a magnitude or complex basis), a B0 map, an R2* map (e.g., from a short TE train or from all echoes with FF compensation), an unwrapped phase map, a susceptibility-weighted imaging (SWI) map, a local frequency shift (LFS) map, a quantitative susceptibility mapping (SQM) map, a contrast map, a T1 relaxation time map, a T2 relaxation time map, and / or a T2p map, as further described in [5], [6], [7], all of which are hereby incorporated by reference.

[0024] The selection of at least one spectral map may depend on a medical application (e.g., the scope of the analysis). For example, an iodine map of multispectral CT raw data may be suitable for imaging myocardial perfusion, pulmonary embolism, and / or pulmonary veins. As another example, a lesion in a pancreatic body may be advantageously visualized in a monoenergetic 50 keV image, an iodine concentration overlay map, and / or an effective atomic number map. As yet another example, a kidney stone may be characterized by an effective atomic number map.

[0025] Generating at least one spectral map may involve applying the predetermined functional dependence of the resolved energy and / or spectrum per unit area or unit volume (e.g., per pixel or per voxel) to the multispectral medical raw imaging data.

[0026] The processing may include converting at least one spectral map into at least one medical image (or simply image). The at least one medical image may be a volumetric image (also called a three-dimensional, 3D image) or a planar image (also called a two-dimensional, 2D image). The planar image may, for example, be associated with a slice of a volumetric medical imaging dataset comprising the multispectral raw medical imaging data.

[0027] The generated medical image can be displayed, for example, on a screen, a head-mounted display (HMD), and / or an extended reality (XR) headset. The screen can be a display (e.g., a touchscreen) on a computing device such as a PC, mobile device, and / or tablet.

[0028] The multispectral medical raw imaging data can comprise a time series. In this case, the medical image can also comprise a time series (e.g., a video sequence).

[0029] Alternatively or additionally, various CT methods exist for acquiring multispectral imaging data, such as dual-energy CT and / or photon-counting CT.

[0030] There are various MRI methods for acquiring multispectral imaging data, such as using two different magnets to successively perform measurements with two different B0 field values ​​and / or applying specially designed excitation pulse sequences.

[0031] Each of the energy-resolved medical imaging techniques can provide improved data, for example, because the patient naturally remains in the same position, as opposed to the conventional practice of taking two separate measurements. Furthermore, less contrast agent can be administered, and the radiation dose to which the patient is exposed can be reduced, thus benefiting the patient's health.

[0032] The method can further include a step to determine a spectral range, a body part, and / or at least one region of interest (ROI) of the received multispectral medical raw imaging data. Optionally, this determination can be based on metadata of the received multispectral medical raw imaging data. Alternatively or additionally, the determination can be based on voxel or pixel data of the received multispectral medical raw imaging data. In particular, the determination can be based on a machine-learned function.

[0033] Alternatively or additionally, the procedure may include a step to receive user input specifying a spectral range, body part and / or at least a range of interest (ROI) of the received multispectral medical raw imaging data.

[0034] Determining the scope of analysis can be based on the spectral range, the imaged body part, and / or at least one region of interest (ROI). The relevant data can be determined automatically and / or by a computing device (e.g., the computing device performing the procedure). Such determination can be based on metadata from the multispectral raw medical imaging data (e.g., a DICOM header) and / or on an analysis of the semantic content, in particular by applying a machine-learned function to the multispectral raw medical imaging data. This allows the procedure to be (e.g., completely or at least primarily) computer-implemented.

[0035] Determining the body part and / or the region of interest (ROI) can be performed using an anatomically aware algorithm. An anatomically aware algorithm can be one designed with a deep understanding of (especially human) anatomy, enabling it to analyze, process, or predict medical data in a way that is consistent with anatomical structures and relationships.

[0036] At least one ROI can be voxel-based or pixel-based. Alternatively or additionally, at least one ROI can have a predetermined geometric shape, e.g., according to a bounding box, a rectangle, a circle, an ellipse, a cube, and / or a sphere.

[0037] The scope of the analysis can be based on user input. For example, the body part and / or at least one region of interest (ROI) to which the multispectral medical raw imaging data pertain can be provided via user input. This can allow the user (e.g., a medical practitioner) to provide a scope that differs from the content provided by the metadata.

[0038] User input can be received via a user interface (UI), such as a graphical user interface (GUI). The UI (and / or GUI) can include a computer mouse, joystick, touchpad, keyboard, touchscreen, and / or a microphone (especially for receiving audio user input and / or verbal instructions to modify the rendering).

[0039] Determining the scope of analysis may involve analyzing metadata related to the multispectral medical raw imaging data, particularly from a worklist, scheduling system, scan protocol, electronic health record (EHR) database, demographic data, and / or from (and / or related to) a medical or patient history, especially of the patient (and / or individual) to whom the multispectral medical raw imaging data pertain. For example, information about the clinical indication (and / or reason for the examination; also: reason for acquiring the multispectral medical raw imaging data) or findings reported in previous examinations may be used.

[0040] Alternatively or additionally, determining the scope of analysis may include using the specified spectral range, the received spectral range, the specified body part, the received body part, the specified at least one RoI and / or the received at least one RoI.

[0041] Alternatively or additionally, determining the scope of analysis can involve running a Large Language Model (LLM).

[0042] Determining the scope of analysis can be based on DICOM headers or any other type of metadata stored in relation to the multispectral medical raw imaging data.

[0043] The metadata can be analyzed semantically using the LLM. This can improve the computer implementation of the procedure.

[0044] The procedure may further include a step to provide the generated at least one medical image for display.

[0045] The presentation can be done using a screen, such as a computer monitor in a doctor's office or a screen in an operating room. Alternatively or additionally, the presentation can be done using an XR headset or HMD. The screen can be a display (e.g., a touchscreen) on a computing device, such as a PC, mobile device, and / or tablet.

[0046] The rendering can include cinematic rendering. Cinematic rendering can encompass an image processing technique used in medical diagnostics, particularly to generate three-dimensional (3D) photorealistic images from cross-sectional data, such as CT or MRI scans.

[0047] Based on a volumetric Monte Carlo path tracing algorithm, cinematic rendering can trace hundreds to thousands of light paths per voxel or per pixel through the data generated by a virtual camera. The light input can be averaged along these paths and transported back to the virtual camera sensor from the high dynamic range images. Scattering, absorption, and emission can then be simulated along the optical paths by means of the interaction between the light and the volumetric data, resulting in vivid, realistic anatomical images with image quality similar to computer-generated imagery (CGI) sequences used in the film industry.

[0048] With its approval for use in the medical field, cinematic rendering can be applied to a range of different areas, including radiology (to supplement available cross-sectional images), surgery (to plan preoperative procedures, such as oral and maxillofacial surgery, trauma surgery, and orthopedics), as well as cardiovascular surgery and interventional radiology. Cinematic rendering can also be used across disciplines, for example, to train postgraduate medical staff and to support patient education and interdisciplinary discussions (such as tumor boards).

[0049] The method may include a step for receiving user input specifying a request to modify the rendering. The method may further include a step for changing at least one rendering parameter based on the received user input. The method may also include a step for providing the at least one medical image with at least one modified rendering parameter for further rendering.

[0050] The rendering can be optimized for the scope of analysis based on the user input received, especially interactively.

[0051] User input can include hovering (e.g., over a row of interest and / or a collection of pixels), zooming in or out, scrolling, performing a virtual reality (VR) rotation, and / or performing a (particularly three-dimensional, 3D) flip. Alternatively or additionally, user input can specify the introduction (and / or modification) of a section plane and / or the introduction (and / or modification) of a division plane.

[0052] Providing the medical image for rendering may include determining a hanging protocol and / or a report template.

[0053] Furthermore, the hanging protocol and / or report template can facilitate an efficient workflow for the user (the medical practitioner), e.g., by selecting a hanging protocol and / or report template that is appropriate for the scope of analysis.

[0054] The selection of the at least one spectral map to be generated can be based on a predefined, configurable set of spectral maps. Alternatively or additionally, the selection of the at least one spectral map to be generated can be rule-based and / or based on a learned function.

[0055] Selecting the at least one spectral map to be generated may involve selecting two or more spectral maps. Processing and / or generating the at least one medical image (also called a medical view) may involve generating, for example, a medical image associated with each spectral map, and / or generating a combination of the two or more spectral maps. Alternatively or additionally, processing and / or generating the at least one medical image may be based on the combination of the two or more spectral maps. Alternatively or additionally, at least one combination image may be generated, for example, with one part based on a first spectral map and another part based on a second map.

[0056] Selecting the at least one spectral map to be generated may include performing a segmentation on at least one of the at least one selected spectral map (e.g., on a first selected spectral map among two or more spectral maps).

[0057] A set of spectral maps can be predefined, from which one or more spectral maps can be selected for the scope of the analysis.

[0058] The selection process can follow predefined rules and / or can be computer-implemented based on a learned function that can be trained on historical data.

[0059] Two or more spectral maps can be advantageously used to differentiate between various diagnostic results. For example, a lesion can be clearly identified by the selected combination of spectral maps, with the associated area being conspicuous in the medical picture.

[0060] The rendering can be based on selecting at least one rendering parameter from a predefined, configurable set of rendering parameters. Optionally, the selection of the at least one rendering parameter can be based on the specific scope of the analysis.

[0061] The rendering parameter can include a camera view, a clipping plane, a clipping box, one or more sampling parameters, one or more interpolation parameters, one or more value-to-color and / or alpha mapping transfer functions, color palettes, a material, lighting, depth value, opacity and / or texture coordinates, e.g., per voxel or per pixel.

[0062] The processing can involve selecting one or more tools (e.g., preprocessing, processing, and / or postprocessing) from a predetermined set of tools (e.g., preprocessing, processing, and / or postprocessing). The tools may be capable of operating fully autonomously and / or automatically.

[0063] The tool (e.g., preprocessing, processing and / or postprocessing) can be, for example, a segmentation tool, a measurement tool, a reconstruction tool and / or a visualization-relevant tool such as masking through segmentation.

[0064] The method may further include a step to provide a list of one or more tools (e.g., preprocessing, processing, and / or postprocessing). Alternatively or additionally, the method may also include a step to receive user input specifying a selection of one or more tools (e.g., preprocessing, processing, and / or postprocessing).

[0065] In one embodiment, the selection of the tools (e.g., preprocessing, processing and / or postprocessing) can be performed by a user (e.g., a medical practitioner) based on the scope of the analysis.

[0066] In other embodiments, the selection of the tools (e.g. pre-processing, processing and / or post-processing) can be carried out by the computing device.

[0067] The method may further include a step for receiving previous image data (hereinafter referred to as: previous images). The method may also include a step for registering the previous image data with the generated at least one spectral map and / or registering the previous image data with the generated at least one medical image. Optionally, the registration may include performing an optimization process. In particular, the registration may include selecting the at least one rendering parameter with the lowest uncertainty.

[0068] Using the registered previous image data, a patient's health history can be calculated and verified.

[0069] Previous image data can be received before processing the multispectral raw medical imaging data (and / or before processing the at least one generated spectral map), e.g., on request, if the computing device (especially automatically) and / or the user deems it necessary. Previous image data or previous results (e.g., extracted from these images) may have been stored alternatively or additionally for precisely this purpose.

[0070] In CT and / or MRI, (especially image) voxels can be the part of interest and / or the relevant fundamental unit of (e.g., multispectral raw) medical imaging data (e.g., for a range of interest and / or in comparison to pixels). CT slices, for example, can contain voxels because they represent a "true slice" of a given thickness of the body (simply called slice thickness). Pixels are used, for example, in CT to describe the signals measured at the detector (specifically, essentially in individual projections). Pixels can also be relevant for, e.g., X-ray imaging (especially as the fundamental unit of, e.g., multispectral raw medical imaging data).

[0071] With regard to one aspect of the device, a computing device is provided for generating a medical image from a spectral map created from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique. The computing device includes a receiving interface for multispectral raw medical imaging data, designed to receive such data. The computing device further includes a determination module designed to determine the scope of analysis of the received multispectral raw medical imaging data. The computing device also includes a selection module designed to select at least one spectral map to be generated from the received multispectral raw medical imaging data.The selection is based on the specified scope of analysis. The computing device further comprises a generation module designed to generate the at least one selected spectral map. The computing device also comprises a processing module designed to process the received multispectral raw medical imaging data to generate at least one medical image based on the at least one generated spectral map. Alternatively or additionally, the processing module can be designed to generate this at least one medical image from the at least one generated spectral map (e.g., by processing the at least one generated spectral map).

[0072] The generation of at least one medical image based on the at least one generated spectral map can be performed by a medical image generation module. Alternatively or additionally, the medical image generation module can be a submodule of the processing module.

[0073] The computing device may be designed to perform any one step and / or may include any of the features disclosed within the context of the procedure according to the procedure aspect.

[0074] With regard to a system aspect, a system for generating a medical image from a spectral map is provided. This map is generated from multispectral raw medical imaging data acquired using energy-resolved medical imaging technology. The system comprises at least one medical scanner designed to acquire multispectral raw medical imaging data. The system further comprises a computing device as defined in the device aspect. The computing device's receiving interface for multispectral raw medical imaging data is designed to receive the multispectral raw medical imaging data from the at least one medical scanner. The system further comprises a display device (e.g., a computer screen) designed to display the generated at least one medical image.

[0075] With regard to another aspect, a computer program product is provided, comprising program elements that enable a computing device to execute the steps of the procedure for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired by means of an energy-resolved medical imaging technique, according to the procedure aspect, when the program elements are loaded into a memory of the computing device.

[0076] With regard to yet another aspect, a computer-readable medium is provided on which program elements are stored that can be read and executed by a computing device in order to carry out steps of the procedure for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired by means of an energy-resolved medical imaging technique, according to the procedure aspect, when the program elements are executed by the computing device.

[0077] The properties, features and advantages of this invention described above, as well as the manner in which they are achieved, will become clearer and more understandable in light of the following description and embodiments, which are described in more detail in the context of the drawings.

[0078] The following description does not limit the invention to the embodiments contained herein. Identical components or parts may be designated by the same reference numerals in different figures. Generally, the figures are not drawn to scale.

[0079] It is understood that a preferred embodiment of the present invention may also be any combination of the dependent claims or of the above embodiments with the respective independent claim.

[0080] These and other aspects of the invention are evident from the embodiments described below and are explained with reference to them. Fig. Figure 1 is a flowchart of a method for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, according to a preferred embodiment of the present invention; and Fig. Figure 2 is an overview of the structure and architecture of a computing device for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, according to a preferred embodiment of the present invention.

[0081] Any reference numerals in the claims should not be interpreted as limiting the scope of protection.

[0082] Fig. Figure 1 schematically illustrates an exemplary flowchart for a computer-implemented method 100 for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique.

[0083] Procedure 100 includes a step S102 for receiving multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique.

[0084] Procedure 100 further includes a step S106 for determining the scope of analysis of the received S102 multispectral raw medical imaging data. The scope of analysis can be calculated from metadata of or associated with the multispectral raw medical imaging data, which is contained, for example, in a DICOM header.

[0085] Method 100 further comprises a step S108 for selecting at least one spectral map to be generated from the received S102 multispectral raw medical imaging data. The selection S108 is based on the determined S106 analysis scope.

[0086] Method 100 further includes a step S110 for generating at least one selected S108 spectral map.

[0087] Method 100 further comprises a step S112 for processing the received S102 multispectral raw medical imaging data to generate S114 at least one medical image based on the at least one generated S110 spectral map. In an alternative embodiment, the at least one medical image can be generated by processing the generated S110 at least one spectral map S114.

[0088] Optionally, procedure 100 can include a step S104-A to determine a spectral range, body part, and / or region of interest (ROI) of the received S102 multispectral medical raw imaging data. Optionally, the determination of S104-A can be based on metadata of the received S102 multispectral medical raw imaging data. Alternatively or additionally, the determination of S104-A can be based on voxel or pixel data of the received S102 multispectral medical raw imaging data. In particular, the determination of S104-A can be based on a machine-learned function.

[0089] Alternatively or additionally, the procedure 100 may include a step S104-B to receive a user input specifying a spectral range, body part and / or RoI of the received S102 multispectral medical raw imaging data.

[0090] Method 100 may include a step S111-A for providing a list of one or more processing tools (and / or preprocessing and / or postprocessing tools). Alternatively or additionally, Method 100 may include a step S111-B for receiving user input specifying a selection of one or more processing tools (and / or preprocessing and / or postprocessing tools).

[0091] The procedure 100 can include a step S116 to provide the generated S114 at least one medical image for display.

[0092] Method 100 may include a step S118 for receiving user input specifying a request to modify the rendering. Method 100 may further include a step S120 for changing at least one rendering parameter based on the received user input S118. Method 100 may also include a step S122 for providing the at least one medical image with at least one rendering parameter that has been modified S120 for further rendering.

[0093] Procedure 100 can produce (in Fig. (1 not shown) step to receive previous image data. Method 100 may further include a (in Fig. One step (also not shown) to register the previous image data with the generated S110 includes at least one spectral map and / or register the previous image data with the generated S114 includes at least one medical image. Optionally, the registration may include performing an optimization process. In particular, the registration may include selecting the at least one rendering parameter for the lowest uncertainty.

[0094] The previous image data can be received before step S112 for processing the received S102 multispectral raw medical imaging data. Alternatively or additionally, the previous image data can be received no later than between processing step 112 and step 114 for generating at least one medical image. In a first embodiment, the previous image data can, for example, be retrieved last, if needed, before step 114. In a second embodiment, the previous image data can alternatively or additionally be stored and therefore be available at any time during the execution of method 100.

[0095] Method 100 can be performed by the computing device 200 of the following Fig. 2 will be carried out.

[0096] Fig.Figure 2 schematically illustrates an exemplary architecture of a computing device 200 for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique.

[0097] The computing device 200 includes a receiving interface 202 for multispectral medical raw imaging data, which is designed to receive multispectral medical raw imaging data. The multispectral medical raw imaging data was acquired using an energy-resolved medical imaging technique.

[0098] The computing device 200 also includes a determination module 206, which is designed to determine the scope of analysis of the received multispectral raw medical imaging data.

[0099] The computing device 200 further comprises a selection module 208, which is designed to select at least one spectral map to be generated from the received multispectral medical raw imaging data, the selection being based on the specified scope of analysis.

[0100] The computing device 200 further comprises a generation module 210, which is designed to generate at least one selected spectral map.

[0101] The computing device 200 further comprises a processing module 212, which is designed to process the received multispectral raw medical imaging data to generate at least one medical image based on the at least one generated spectral map. In an alternative embodiment, the at least one medical image can be generated by the processing module 212 by processing the generated at least one spectral map.

[0102] The generation of at least one medical image based on the at least one generated spectral map can be performed by a medical image generation module 214. Alternatively or additionally, the medical image generation module 214 can be a submodule of the processing module 212.

[0103] Optionally, the computing device 200 can include a spectral range determination module 204-A, which is designed to determine a spectral range, a body part, and / or a region of interest (ROI) of the received multispectral raw medical imaging data. Optionally, the determination can be based on metadata of the received multispectral raw medical imaging data, such as that contained in a DICOM header. Alternatively or additionally, the determination can be based on voxel or pixel data of the received multispectral raw medical imaging data. In particular, the determination can be based on a machine-learned function.

[0104] Alternatively or additionally, the computing device 200 can include a first user input receiving interface 204-B designed to receive user input specifying a spectral range, body part and / or a RoI of the received multispectral medical raw imaging data.

[0105] The computing device 200 can include a processing tool provisioning module 211-A, which is designed to provide a list of one or more processing tools (and / or pre-processing and / or post-processing tools).

[0106] Alternatively or additionally, the computing device 200 may include a second user input receiving interface 211-B designed to receive user input specifying a selection of one or more processing tools (and / or preprocessing and / or postprocessing tools).

[0107] The computing device 200 can include a medical image provisioning interface 216, which is designed to provide the generated at least one medical image for display.

[0108] The computing device 200 can include a third user input receiving interface 218 designed to receive user input specifying a request to modify the rendering.

[0109] The computing device 200 can include a rendering parameter modification module 220, which is designed to modify at least one rendering parameter based on the user input received.

[0110] The medical image delivery interface 216 can also be designed to provide the at least one medical image with at least one rendering parameter that has been changed, for further rendering.

[0111] The computing device 200 can include an input-output interface 224. The input-output interface 224 can implement the following: the multispectral medical raw imaging data receive interface 202, the optional first user input-receive interface 204-B, the optional second user input-receive interface 211-B, the optional medical image delivery interface 216, and / or the optional third user input-receive interface 218.

[0112] The computing device 200 can include a processor 226. The optional spectral range determination module 204-A, the determination module 206, the selection module 208, the generation module 210, the optional processing tool provisioning module 211-A, the processing module 212, the optional medical image generation module (or sub-module) 214 and / or the optional rendering parameter modification module 220 can be implemented by the processor 226.

[0113] The computing device 200 can include a memory 228. Program elements for carrying out the steps of the procedure 100 can be stored within the memory 228.

[0114] The computing device 200 can be designed to carry out the procedure 100.

[0115] A system can comprise the computing device 200 and at least one medical scanner designed to acquire multispectral raw medical imaging data. The system can further comprise a display device designed to display the generated at least one medical image. The system can be designed to perform the method 100.

[0116] The inventive technique (e.g., comprising the method 100, the computing device 200, and / or the system) can, according to a first embodiment, include a first step S102 for receiving a multispectral CT raw data set. Optionally, spectral regions and one or more body parts to be displayed (also: imaged) can be determined from a DICOM header file (hereinafter: DICOM header) and / or directly from pixel data (e.g., by applying a machine-learned function).

[0117] A reason for the examination (and / or the reason for acquiring the multispectral medical raw imaging data) can be determined according to step S106. This may involve querying relevant data from a worklist, planning system, and / or electronic health record (EHR) database and optionally using the spectral ranges or body part known for the CT dataset. The data may include additional patient-related information, such as demographic data and / or medical history. Step S106 may be performed by a lecturer / librarian (LLM). Alternatively or additionally, one or more anatomy-aware algorithms may be used to determine the body part and / or a more specific region (e.g., at least one region of interest) of the anatomical structure.

[0118] A pixel-based ROI can be more accurate than data retrieved from medical imaging metadata or from an EPA database and / or EGA data sources. Data available in the selected and used CT protocol for the scan may provide further reasons or patterns for the scan's category.

[0119] Furthermore, this process (particularly determining the reason for the examination and / or the scope of analysis) can be applied beforehand (e.g., as a preparatory step), possibly before any user interaction. Therefore, increasing speed by attempting to predict which medical images will be most useful and preparing the predicted medical images in advance so they are ready to be displayed can improve a medical (e.g., diagnostic and / or therapeutic) workflow.

[0120] One or more spectral maps are selected according to the reason for the investigation in step S108. This may also include selecting a combination of (e.g., two or more) spectral maps.

[0121] The multispectral CT raw data are processed in step S112, so that images are generated according to the selected S108 spectral maps S114.

[0122] A visualization can be generated based on the images. This can involve creating composite images, with one part based on a first spectral map and another part based on a second map. This can include applying a segmentation tool to one image (representing, for example, the first map) to obtain segmented image data and using that segmented data when generating the visualization of the other image (representing, for example, the second map). The segmentation can be selected based on the reason for the investigation (and / or the scope of the analysis). A visualization can include cinematic rendering.

[0123] An assignment of the information collected in processing step S112 to a predefined configurable set of spectral maps and rendering options can be used (such as to automatically generate the most probable visualizations).

[0124] Multiple representations can be interactively displayed simultaneously. In a first embodiment, multiple representations can be displayed by hovering over any region (e.g., ROI) and / or any pixel, with this view being enabled ad-hoc, e.g., on a shortcut or in a magnifying glass-like tool. In a second embodiment, scrolling and / or performing a VR (also: VRT) rotation and / or displaying measurements in all representations can be enabled. In a third embodiment, one or more 3D flips (e.g., on Apple and / or Windows operating systems) are enabled to display multiple representations simultaneously. In a fourth embodiment, one or more orthogonal 3D section planes are used, each displaying a different representation. In a fifth embodiment, a split-plane 3D visualization (e.g.,such as 3D Neuro), which displays different representations in each half-space. All of the above embodiments can be combined.

[0125] Optionally, one or more image processing tools (and / or image preprocessing and / or postprocessing tools) can be selected and applied to each medical image generated (e.g., in the future) (e.g., in step S112). Processing tools (and / or preprocessing and / or postprocessing tools) can be generally designed to detect medical findings in the medical imaging data. Image processing tools (and / or image preprocessing and / or postprocessing tools) can be specific to a particular spectral map and / or may perform best with a particular spectral map. Therefore, the step of selecting and applying one or more processing tools (and / or preprocessing and / or postprocessing tools) can also include feeding the most suitable spectral map into the selected tool (e.g., rule-based and / or with a machine-learned function).

[0126] Optionally, a hanging protocol and / or one or more report templates can be determined based on the spectral maps selected (e.g. in step S108) (e.g. in step S116).

[0127] Longitudinal data can be used optionally. Previous image data can be registered with the multispectral CT dataset for subsequent reading. The different spectral maps can be registered inherently. This allows them to be used separately for registration with a previous study (which, for example, does not have to be multispectral or even a CT dataset). In particular, different representations can be used for registration with the previous data, and the best one (e.g., with the lowest uncertainty) can be used for the final registration of the multispectral CT data with the previous image data. Alternatively or additionally, the individual registrations can be aggregated in an optimization process.

[0128] According to a second embodiment, a multispectral CT raw dataset is received in step S102. Optionally, spectral regions and one or more body parts to be displayed (and / or imaged) are determined from a DICOM header file (and / or a DICOM header) and / or directly from pixel data (e.g., by applying a machine-learned function).

[0129] An image processing task is received from the user (e.g., a radiologist or other healthcare provider). Receiving the image processing task may involve a command, such as "show me the liver and liver lesions." The command can be entered implicitly (e.g., by opening a liver case) and / or explicitly, such as by activating a tool and / or typing natural language.

[0130] One or more spectral maps can be selected according to the task (e.g., in step S108). This selection S108 can be rule-based (e.g., "Liver analysis => iodine map and low kV image") and / or using more complex learned functions.

[0131] One or more spectral images can be generated according to the Selected S108 spectral maps S114.

[0132] One or more image processing tools can be selected based on the task and the spectral maps and / or images (e.g. for processing step S112). The image processing tools (and / or preprocessing and / or postprocessing tools) can be applied to one or more medical images (e.g., in processing step S112) and / or can be suggested to a clinical user for use (e.g., by displaying only this selection tool in a tool menu).

[0133] Different tools (e.g., preprocessing, processing, and / or postprocessing) can be applied to different (especially multispectral raw) medical imaging data and / or to different medical images (especially those to be generated). Generally, a combination of tools (e.g., preprocessing, processing, and / or postprocessing) may be required to successfully complete a task. For example, a segmentation mask can be defined in (or for) a medical image whose spectrum is best suited for this purpose, and the segmentation mask can be transferred to another medical image whose spectrum is best suited for identifying lesions within the mask. The task can be completed based on the image processing results (and / or preprocessing and / or postprocessing results).

[0134] Regardless of the grammatical use of a term, individuals (e.g., patients and / or people) with male, female, or other gender identities are included in the term.

[0135] Unless expressly stated otherwise, individual embodiments or their individual aspects and features described with reference to the drawings may be combined or interchanged without limiting or extending the scope of protection of the described invention, wherever such combination or interchangeability is sensible and in line with the purpose of this invention. Advantages described with respect to a particular embodiment of the present invention or with respect to a particular figure may also be advantages of other embodiments of the present invention.

Claims

[1] Computer-implemented method (100) performed by a computing device (200) for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired by means of an energy-resolved medical imaging technique, comprising the following steps: - Received (S102) multispectral medical raw imaging data acquired using an energy-resolved medical imaging technique; - Determine (S106) the scope of analysis of the received (S102) multispectral raw medical imaging data; - Selecting (S108) at least one spectral map to be generated from the received (S102) multispectral raw medical imaging data, the selection (S108) being based on the specified (S106) scope of analysis; - Generating (S110) at least one selected (S108) spectral map; and - Processing (S112) the received (S102) multispectral raw medical imaging data to generate (S114) at least one medical image based on the at least one generated (S110) spectral map. [2] Method (100) according to claim 1, wherein the energy-resolved medical imaging technique is selected from the group consisting of: - Computed tomography, CT, especially photon-counting CT; - Magnetic resonance imaging (MRI); - photoacoustic medical imaging; - Digital dual-energy X-ray imaging; - photon-counting X-ray imaging; and - Energy-resolved neutron imaging. [3] Method (100) according to any one of the preceding claims, further comprising at least one of the following steps: - Determine (S104-A) a spectral range, body part and / or region of interest, RoI, of the received (S102) multispectral medical raw imaging data, optional ◯ where the determination (S104-A) is based on metadata of the received (S102) multispectral medical raw imaging data, and / or optionally ◯ wherein the determination (S104-A) is based on voxel data or pixel data of the received (S102) multispectral medical raw imaging data, wherein the determination (S104-A) is in particular based on a machine-learned function; - Receiving (S104-B) a user input specifying a spectral range, body part and / or RoI of the received (S102) multispectral medical raw imaging data. [4] Method (100) according to any of the preceding claims, wherein determining (S106) the scope of analysis comprises: - Analyzing metadata relating to multispectral medical raw imaging data, in particular from a worklist, a planning system, a scan protocol, an electronic patient record or EPR database, demographic data and / or in relation to a medical history; - Using the specific (S104-A) and / or received (S104-B) spectral range, body part and / or RoI; and / or - Running a Large Language Model (LLM), especially on metadata of multispectral medical raw imaging data. [5] Method (100) according to any one of the preceding claims, further comprising the following step: - Providing (S116) of the generated (S114) at least one medical image for display. [6] Method (100) according to the immediately preceding claim, wherein the rendering comprises cinematic rendering. [7] Method (100) according to claim 5 or 6, further comprising the following steps: - Receiving (S118) user input indicating a request to modify the rendering; - Modify (S120) at least one rendering parameter based on the received (S118) user input; and - Providing (S122) the at least one medical image with at least one rendering parameter that has been modified (S120) for further rendering. [8] Method (100) according to any one of claims 5 to 7, wherein providing (S116) the medical image for rendering comprises determining a hanging protocol and / or a report template. [9] Method (100) according to one of the preceding claims, wherein the selection (S108) of the at least one spectral map to be generated: - based on a predefined, configurable set of spectral maps; - is rule-based and / or based on a learned function; - Selecting two or more spectral maps includes, and wherein the processing (S112) and / or generating (S114) of the at least one medical image includes generating a combination of the two or more spectral maps, and / or wherein the processing (S112) and / or generating (S114) of the at least one medical image, which, for example, includes a combination image, is based on a combination of the two or more spectral maps; and / or - Performing a segmentation on at least one of the at least one selected (S108) spectral map. [10] Method (100) according to any of the preceding claims, wherein the rendering (S116) is based on selecting at least one rendering parameter from a predefined configurable set of rendering parameters, optionally where the selection of at least one rendering parameter is based on the specific (S106) scope of analysis. [11] Method (100) according to one of the preceding claims, wherein the processing (S112) comprises selecting one or more, in particular pre-processing, processing and / or post-processing, tools from a predetermined set of, in particular pre-processing, processing and / or post-processing, tools. [12] Method (100) according to any one of the preceding claims, further comprising at least one of the following steps: - Providing (S111-A) a list of one or more tools, in particular pre-processing, processing and / or post-processing tools; and - Receiving (S111-B) user input specifying a selection of one or more tools, in particular preprocessing, processing and / or postprocessing tools. [13] Method (100) according to any one of the preceding claims, further comprising the following steps: - Receiving previous image data; and - Registering the previous image data with the generated (S110) at least one spectral map and / or with the generated (S114) at least one medical image, optional where registration includes performing an optimization process, and in particular, registration includes selecting at least one rendering parameter for the lowest possible uncertainty. [14] Computing device (200) for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, wherein the computing device (200) comprises: - a receiving interface (202) for multispectral medical raw imaging data, designed to receive multispectral medical raw imaging data acquired using an energy-resolved medical imaging technique; - a determination module (206) designed to determine the scope of analysis of the received multispectral raw medical imaging data; - a selection module (208) designed to select at least one spectral map to be generated from the received multispectral medical raw imaging data, the selection being based on the specified scope of analysis; - a generation module (210) designed to generate at least one selected spectral map; and - a processing module (212) designed to process the received multispectral raw medical imaging data to produce at least one medical image based on the at least one generated spectral map. [15] Computing device (200) according to the immediately preceding claim, which is further designed to perform any one of the steps according to any one of the method claims 2 to 13 and / or includes any one of the features thereof. [16] System for generating a medical image from a spectral map generated from multispectral raw medical imaging data acquired using an energy-resolved medical imaging technique, the system comprising: - at least one medical scanner designed to capture multispectral medical raw imaging data; - a computing device (200) according to claim 14 or 15, wherein the receiving interface (202) for multispectral medical raw imaging data is designed to receive the multispectral medical raw imaging data from the at least one medical scanner; and - a display device designed to show the generated at least one medical image.

Images