Deforming functional image data to match associated anatomical image data

Abstract

A system (102) includes a spatial mismatch correction module (168) configured to receive functional transmission data, anatomical image data, and functional image data that is reconstructed based on the functional transmission data and attenuated based on the anatomical image data. The system also includes a dataset provider (502) configured to provide the first dataset and the second dataset that are spatially mismatched. The system also includes a voxel-of-interest identifier (512) configured to identify voxels or regions of reconstruction inconsistency due to a spatial mismatch between a real attenuation value and an attenuation value derived from the anatomical image data based on a relationship between the first data set and the second data set. The system also includes an image data generator (614) configured to deform the functional image data based on the identified voxels or regions and generate corrected functional image data independent of the functional-anatomical correlation while maintaining image quality of the functional image data.

Description

Technical Field

[0001] The following text relates generally to functional imaging and more specifically to deforming functional image data to match associated anatomical image data, and is particularly applicable to deforming positron emission tomography (PET) imaging data to match computed tomography (CT) image data and / or magnetic resonance (MR) image data used for attenuation correction of PET image data. Background Technology

[0002] Multimodal image spatial matching is used in conjunction with positron emission tomography (PET)-computed tomography (CT), PET-magnetic resonance (MR), single-photon emission computed tomography (SPECT)-CT, and SPECT-MRI medical imaging to match functional image data with associated anatomical image data. Using practical clinical protocols, it is difficult to acquire PET emission and CT transmissions at the same patient spatial location, and spatial mismatches will exist between functional and anatomical image data, for example, because PET emission is acquired continuously during free breathing, while CT transmissions are acquired under arbitrary breath-holding conditions (typically during inspiration). Respiratory motion is just one example of motion that can lead to such spatial mismatches between functional and anatomical image data. Other motions that can cause spatial mismatches include cardiac motion, occasional patient movement, etc.

[0003] Existing PET reconstruction and multimodal image registration techniques spatially match functional image data with associated anatomical (CT, MR, etc.) image data for attenuation correction or clinical diagnostic review workflows. Such PET reconstruction or image registration techniques typically require functional-anatomical correlation. However, due to the different characteristics of functional and anatomical image information, this correlation is unreliable in clinical practice. Unfortunately, existing methods for addressing spatial mismatch cannot achieve accurate functional-anatomical image data matching for clinical diagnostic workflows while maintaining the original image quality of the functional image data.

[0004] Given at least the foregoing, there remains an unresolved need for improved spatial matching methods in multimodal imaging. Summary of the Invention

[0005] The aspects described herein address the aforementioned and other problems. This invention summary introduces concepts described in more detail in the detailed description. It should not be used to identify the essential features of the claimed subject matter, nor should it be used to limit the scope of the claimed subject matter.

[0006] In one aspect, a system includes a spatial mismatch correction module configured to receive functional emission data, anatomical image data, and functional image data reconstructed based on the functional emission data and attenuated based on the anatomical image data. The system also includes a dataset provider configured to provide a first dataset and a second dataset, wherein the first and second datasets include spatial mismatches. The system further includes a voxel identifier configured to identify voxels or regions of reconstruction inconsistency caused by spatial mismatch between true attenuation values ​​and attenuation values ​​derived from the anatomical image data, based on a relationship between the first and second datasets. The system also includes an image data generator configured to deform the functional image data based on the identified voxels or regions, independently of functional-anatomical correlations, while maintaining image quality of the functional image data, and generate corrected functional image data.

[0007] In another aspect, a computer-implemented method includes receiving functional emission data, anatomical image data, and functional image data reconstructed based on the functional emission data and attenuated based on the anatomical image data. The computer-implemented method further includes providing a first dataset based at least on the anatomical image data and a second dataset based at least on the functional emission data or modified anatomical image data. The computer-implemented method also includes identifying voxels or regions of reconstruction inconsistency caused by spatial mismatch between true attenuation values ​​and attenuation values ​​derived from the anatomical image data, based on the relationship between the first and second datasets. The computer-implemented method further includes deforming the functional image data based on the identified voxels or regions while maintaining the image quality of the functional image data, thereby generating deformed functional image data.

[0008] In another aspect, a computer system includes: a computer-readable storage medium having instructions for correcting motion in data; and a processor for receiving functional emission data, anatomical image data, and functional image data reconstructed based on the functional emission data and attenuated based on the anatomical image data. The instructions further cause the processor to provide a first dataset based at least on the anatomical image data and a second dataset based at least on the functional emission data or modified anatomical image data. The instructions further cause the processor to identify voxels or regions of reconstruction inconsistency caused by spatial mismatch between true attenuation values ​​and attenuation values ​​derived from the anatomical image data, based on the relationship between the first and second datasets. The instructions further cause the processor to deform the functional image data based on the identified voxels or regions while maintaining the image quality of the functional image data, thereby generating deformed functional image data.

[0009] Other aspects of this application will be recognized by those skilled in the art upon reading and understanding the accompanying specification. Attached Figure Description

[0010] This application is illustrated by way of example and is not limited to the figures in the accompanying drawings, in which the same reference numerals indicate similar elements.

[0011] Figure 1 A non-limiting example cross-sectional side view of a multimodal imaging system including a spatial mismatch correction module according to an embodiment of the present invention is schematically illustrated.

[0012] Figure 2 The implementation scheme according to this document is illustrated schematically. Figure 1 A front view of a non-limiting example of an imaging system subsystem of a multimodal imaging system.

[0013] Figure 3 The implementation scheme according to this document is illustrated schematically. Figure 1 A front view of a non-limiting example of another imaging system subsystem of a multimodal imaging system.

[0014] Figure 4 An example of a spatial mismatch correction module for a multimodal imaging system according to an embodiment of this paper is illustrated schematically.

[0015] Figure 5 An example spatial mismatch identifier of a spatial mismatch correction module according to an embodiment of this paper is illustrated schematically.

[0016] Figure 6 An example PET image data generator is illustrated schematically, representing a spatial mismatch correction module according to an embodiment of this paper.

[0017] Figure 7 Example first image data of two image datasets is illustrated schematically according to the spatial mismatch correction method of the embodiment of this paper.

[0018] Figure 8 Example second image data of two image datasets is illustrated schematically according to the spatial mismatch correction method of the embodiment of this paper.

[0019] Figure 9 An example spatial mask illustrating the spatial mismatch correction method according to the implementation scheme of this paper is shown schematically.

[0020] Figure 10 An example mask with a main direction for spatial mismatch correction method superimposed on anatomical image data is schematically illustrated according to the embodiment of this paper.

[0021] Figure 11An example mask having line segments superimposed on anatomical image data for a spatial mismatch correction method is schematically illustrated according to an embodiment of the present invention.

[0022] Figure 12 An example variation of the spatial mismatch correction method according to the implementation scheme of this paper is illustrated schematically.

[0023] Figure 13 A non-limiting example of a region of a spatial mask superimposed on anatomical image data used for a spatial mismatch correction method, according to an embodiment of the present invention, is illustrated graphically.

[0024] Figure 14 The diagram illustrates, graphically, the region after applying volumetric filtering and thresholding according to the embodiment described herein, leaving an internal region with a high value relative to the rest of the region.

[0025] Figure 15 The main directions shown at multiple points in the region according to the implementation scheme of this paper are illustrated graphically.

[0026] Figure 16 The main directions shown at multiple points of the spatial mask according to the implementation scheme of this paper are illustrated graphically.

[0027] Figure 17 The anatomical distribution of the line segments above the mask region and the segments on the opposite sides of the mask prior to deformation is depicted graphically according to the embodiment described herein.

[0028] Figure 18 The diagrams depict the distribution of the anatomical structures of the line segments above the mask region and the segments on the opposite sides of the mask after deformation, according to the embodiment described herein.

[0029] Figure 19 An example geometry illustrating the image data deformation process according to the embodiment described herein is shown schematically.

[0030] Figure 20 An example of an algorithm for generating a dataset for functional image data deformation schemes, according to an embodiment of this paper, is illustrated schematically.

[0031] Figure 21 An example functional image data with mismatch artifact input is depicted for a data deformation scheme according to the embodiment described herein.

[0032] Figure 22 Example error image data for a data deformation scheme according to the implementation scheme described herein is depicted.

[0033] Figure 23An example outlined anatomical image of a data deformation scheme according to the embodiment described herein is depicted.

[0034] Figure 24 An example mask for a data deformation scheme according to the implementation of this paper is depicted.

[0035] Figure 25 An example mask is depicted according to the embodiment described herein, which has a region with a high mismatch probability in the attenuated data determined for the data deformation scheme.

[0036] Figure 26 Error image data with a mask superimposed thereon for a data deformation scheme, according to an embodiment of the present invention, is depicted, the mask having areas with a high probability of mismatch.

[0037] Figure 27 The region segmented for potential mismatch in the data deformation scheme according to the implementation scheme described herein is depicted.

[0038] Figure 28 The region segmented following a morphological operation, according to the implementation scheme described herein, is depicted. This morphological operation is used to filter out misidentified voxels from clustering for a data transformation scheme.

[0039] Figure 29 The present invention describes input functional image data with mismatch artifacts caused by anatomical image data used for data deformation schemes, according to the embodiments described herein.

[0040] Figure 30 The input functional image data for mismatch artifact correction based on corrected anatomical image data for data deformation schemes, according to the embodiments described herein, are depicted.

[0041] Figure 31 A limiting example of a flowchart illustrating a computer-implemented method, according to an embodiment of this paper, for deforming functional image data to match corresponding anatomical image data independently of function-anatomical correlation.

[0042] Figure 32 A limited example of a flowchart illustrating a computer-implemented method for generating a mask for deforming functional image data to match corresponding anatomical image data, independent of function-anatomical correlation, according to an embodiment of this paper.

[0043] Figure 33 A limiting example of a flowchart illustrating a computer-implemented method for generating error image data for deforming functional image data to match corresponding anatomical image data, according to an embodiment of this paper.

[0044] Figure 34 A limiting example of a flowchart illustrating a computer-implemented method for generating error image data for deforming functional image data to match corresponding anatomical image data, according to an embodiment of this paper.

[0045] Figure 35 Another non-limiting example of a flowchart illustrating a computer-implemented method for identifying inconsistencies in a method for deforming functional image data to match corresponding anatomical image data, according to an embodiment of this paper. Detailed Implementation

[0046] Embodiments of this disclosure will now be described by way of example with reference to the accompanying drawings, wherein systems, methods, and / or computer-readable media include instructions for matching functional image distortions of multimodal medical imaging (positron emission tomography (PET)-computed tomography (CT), PET-magnetic resonance (MR), single-photon emission computed tomography (SPECT)-CT, and SPECT-MR, etc.) with associated anatomical image data independent of functional-anatomical correlation.

[0047] As discussed in this paper, existing multimodal medical imaging methods spatially match functional and anatomical image data for attenuation correction, requiring functional-anatomical correlations. These correlations are often unreliable in clinical imaging due to motions such as respiration, cardiac activity, and occasional patient movements (which generate functional and anatomical image data at different spatial locations). Existing methods for addressing this spatial mismatch cannot achieve accurate image data matching for clinical diagnostic workflows while maintaining the original image quality of the functional image data.

[0048] As described in more detail below, the method presented in this paper utilizes local functional image deformation based on reconstruction inconsistencies occurring in regions with spatial mismatches between the true attenuation values ​​and those derived from anatomical image data. In one case, the method described in this paper allows for anatomical image data attenuation correction of functional image data independent of function-anatomical correlations, while maintaining the diagnostic quality and reliability of the original (i.e., before deformation) functional image data.

[0049] First refer to Figure 1A cross-sectional side view of an imaging system 102 configured for multimodal imaging is schematically illustrated. Examples of such systems include hybrid PET-CT, PET-MRI, SPECT-CT, SPECT-MRI, and other imaging systems that include functional imaging subsystems and anatomical imaging subsystems integrated together in a single system, and / or separate functional imaging and anatomical imaging systems (e.g., separate PET scanners and CT scanners, etc.). For purposes of explanation and brevity, this method is described below in conjunction with a hybrid PET-CT imaging system. Imaging system 102 includes a PET imaging subsystem 104 and a CT imaging subsystem 106 integrated into a single imaging system.

[0050] Temporary turn Figure 2 A sample front view of the PET imaging subsystem 104 is schematically illustrated. (Reference) Figure 1 and Figure 2 The PET imaging subsystem 104 includes a PET gantry 108. The PET gantry 108 includes a radiation-sensitive detector array 110 arranged in a generally annular ring around a PET examination area 112. The radiation-sensitive detector array 110 includes multiple detectors (photosensors) in optical communication with a scintillator material (scintillator crystal), the scintillator material being disposed between the multiple detectors and the PET examination area 112.

[0051] The scintillator material will respond to the patient 118 placed therein ( Figure 2 The positron annihilation event 116 that occurred in the inspection area 112 of ) Figure 2 The resulting 511 keV gamma radiation 114 ( Figure 2 The photons are converted into visible photons, and the multiple detectors convert the photons into electrical signals. The multiple detectors include one or more photoelectric sensors, such as avalanche photodiodes, photomultiplier tubes, silicon photomultiplier tubes, and / or another type of photoelectric sensor.

[0052] The PET imaging subsystem 104 also includes a PET data acquisition system (DAS) 120. The PET data acquisition system 120 receives data from the radiation-sensitive detector array 110 and generates PET emission data, which includes a list of events detected by the multiple radiation-sensitive detectors 110. The PET DAS 120 identifies overlapping gamma pairs by identifying events detected in time as coincidentally (or nearly simultaneously) along a response line (LOR) that serves as a straight line connecting the two detectors that detected the events, and generates a list of pattern data and / or a histogram (sine graph) indicating it.

[0053] Coincidence can be determined by several factors, including event time stamps and the Level of Response (LOR), which must fall within a predetermined time period to indicate coincidence. Unpairable events can be used to estimate and correct for random coincidence, but not directly for the reconstructed data. Pairable events are located and recorded as coincident event pairs. PET emission data provides information about the LOR for each event, such as the lateral and longitudinal location of the LOR, as well as the lateral and azimuth angles. Additionally or alternatively, PET emission data is re-segmented into one or more sinogram or projection modules.

[0054] When the PET imaging subsystem 104 is configured for time-of-flight (TOF), the PET emission data can also include TOF information that allows estimation of the location of an event along the LOR. For example, when a positron annihilation event occurs closer to the first detector crystal than the second detector crystal, an annihilation photon may arrive at the first detector crystal before another annihilation photon arrives at the second detector crystal (e.g., nanoseconds or picoseconds). The TOF difference can be used to constrain the location of the positron annihilation event along the LOR.

[0055] Temporary turn Figure 3 A sample front view of the CT imaging subsystem 106 is schematically illustrated. (Reference) Figure 1 and Figure 3 The CT imaging subsystem 106 includes a CT gantry 124. The CT gantry 124 includes a radiation-sensitive detector array 126 arranged in a ring around a CT examination area 128. The CT gantry 124 further includes a radiation source 130, such as an X-ray tube or source, rotating around the CT examination area 128. The radiation-sensitive detectors 126 detect radiation emitted by the radiation source 130 that has passed through the examination area 128 and the subject 118 therein. Figure 3 ) radiation 132 ( Figure 3 ).

[0056] The radiation source 130 and the radiation-sensitive detector array 126 are positioned opposite each other across the CT examination area 128 on the rotating frame 134. Figure 3 The rotating frame 134 rotates the X-ray source 130 in coordination with the X-ray radiation detector array 126. The X-ray source 130 emits X-ray radiation 132 that traverses the examination area 128 and the subject 118 positioned therein, and the X-ray radiation detector array 126 detects the X-ray radiation impacting it. For each arc segment, the X-ray radiation detector array 126 generates a projected view. The CT data acquisition system (DAS) 136 processes the signals from the CT detector 126 to generate projected data indicating radiation attenuation along multiple lines or rays passing through the examination area 128.

[0057] refer to Figure 1 The patient support 140 includes a table 142 movably coupled to a frame / base 144. In one case, the table 142 is slidably coupled to the frame / base 144 via bearings or the like, and a drive system (not visible) including a controller, motor, lead screw, and nut (or other drive system) causes the table 142 to translate along the frame / base 144 into and out of the examination areas 128 and / or 112. The table 142 is configured to support an object or patient in the examination areas 128 and / or 112 for loading, scanning, and / or unloading the patient or object. The examination areas 112 and 128 are arranged along a common longitudinal axis or z-axis (Z). In the case where the PET subsystem 104 and the CT subsystem 106 are separate imaging systems, each may have its own patient support.

[0058] Controller 146 is configured to control components such as rack 124 ( Figure 3 The controller 146 controls the rotation of the X-ray source 130, the operation of the detector arrays 126 and / or 110, the operation of the patient support 140, etc. For example, in one embodiment, the controller 146 includes a patient support controller configured to control the movement and / or height of the patient support 140 for loading, scanning, and / or unloading of the patient or object. In cases where the PET subsystem 104 and the CT subsystem 106 are separate imaging systems, each may have its own controller.

[0059] The CT reconstructor 148 uses a reconstruction algorithm to reconstruct CT projection data to generate volumetric image data (i.e., CT image data) that indicates the radiation attenuation of the subject or object. Suitable reconstruction algorithms include algebraic reconstruction techniques (ART), analytical image reconstruction algorithms such as filtered back projection (FBP), iterative reconstruction algorithms such as advanced statistical iterative reconstruction (ASIR), maximum likelihood maximum expectation (MLEM), another algorithm, and / or combinations thereof.

[0060] Attenuation corrector 150 generates attenuation correction data (e.g., attenuation correction (μ-) map, etc.) to correct PET emission data for attenuation (i.e., photon loss) in the subject or object as 511 keV coincident photons travel along the LOR to detector array 110. In this example, the attenuation correction data is generated based on CT image data reconstructed by CT reconstructor 148 (e.g., by scaling the number of CTs in the CT image data from the average CT energy to the PET photon energy of 511 keV). PET emission data can be processed before and / or after energy scaling (e.g., resolution matching). In one case, attenuation corrector 150 utilizes a bilinear function that will be expressed as the reciprocal of the centimeter (cm²). -1 A unique 511-keV linear attenuation value is mapped to each measurement in Henle units (HU) in the CT image data. Typically, attenuation correction adds the count back to areas with more attenuation and / or subtracts the count from areas with less attenuation than other tissues.

[0061] PET reconstructor 152 uses known iterative or other techniques to reconstruct attenuation-corrected PET emission data to generate volumetric image data (i.e., PET image data) indicating the distribution of radionuclides in the scanned object. Suitable reconstruction algorithms include ART techniques, analytical image reconstruction algorithms such as FBP, iterative image reconstruction algorithms such as Ordered Subset Expectation Maximization (OSEM), Block Order Regularized Expectation Maximization (BSREM), and other algorithms and / or combinations thereof.

[0062] The imaging system 102 also includes an operator console 156. The operator console 156 includes a computing system, such as a computer, workstation, server, etc. The operator console 156 includes input devices 158 such as a keyboard, mouse, touchscreen, microphone, etc., and output devices 160 such as human-readable devices such as a display monitor. The operator console 156 also includes input / output (I / O) 162, which is configured to send and / or receive signals and / or data, for example, via input devices 158, output devices 160, wireless technology, portable devices, etc.

[0063] The operator console 156 also includes a processor 164, such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor unit (μPU), etc. The operator console 156 also includes a computer-readable storage medium 166 (“memory”), which includes non-transitory media (e.g., storage cells, devices, etc.) and does not include transient media (i.e., signals, carrier waves, etc.). In the illustrated example, the operator console 156 receives one or more of the following: CT projection data, CT image data, CT attenuation maps, PET emission data, PET projections, PET list pattern data, PET LOR, PET sine curves, etc.

[0064] The memory 166 is encoded with computer-executable instructions. In the illustrated example, the computer-executable instructions include a spatial mismatch correction module 168, which is configured to perform spatial matching of functional image data and anatomical image data when there is no functional-anatomical correlation between the matched functional image data and the anatomical image data, and while maintaining the image quality of the functional image data. (To be continued...) Figure 4 Examples of spatial mismatch correction modules 168 include a spatial mismatch identifier 402 and a PET image data generator 404.

[0065] Spatial mismatch identifier 402 is configured to identify spatial voxels that meet predetermined reconstruction inconsistency criteria from spatial mismatches between tissue voxel attenuation values ​​and tissue-specific voxel attenuation values ​​derived from anatomical image data and used for attenuation correction during functional image reconstruction due to different tissue motions during functional and anatomical image acquisition. As described in more detail below, the predetermined criteria are based on relationships between the two image datasets (e.g., differences, ratios, combinations thereof, etc.) (or based on the results of combined reconstruction), and these relationships are used to generate a spatial mask, wherein the two image datasets may include reconstructed image data, projections, response lines (LORs), sine waves, etc.

[0066] PET image data generator 404 is configured to generate PET image data with improved spatial conformity to anatomical image data relative to the initial PET image data, based on a spatial mask generated by spatial mismatch identifier 402. As described in more detail below, PET image generator 404 uses the spatial mask along with one or more principal motion directions associated with the anatomical image data or motion model, one or more line segments of interest for correction, and one or more deformation algorithms to deform the original PET image data to generate new PET image data. In one case, the generated PET image data maintains the diagnostic quality level of the deformed original PET image data. The generated PET image data can be displayed, realized, captured, processed, and visually presented together with the original anatomical image data, etc.

[0067] Return to Figure 1 System 102 also includes remote resources 170. In one instance, remote resources 170 include a Radiology Information System (RIS), a Hospital Information System (HIS), an Electronic Medical Records (EMR), a Picture Archiving and Communication System (PACS), one or more other separate and / or hybrid imaging systems, servers, databases, cloud-based resources (including shared remote data storage and / or computing power, including processing resources distributed across multiple locations / data centers), etc. Imaging system 102 is in electrical communication with remote resources 170 and is configured to send and / or receive image data via medical digital imaging and communication (DICOM), etc., and to send and / or receive other data via health level 7 (HL7), etc.

[0068] Go to Figure 5 An example of a spatial mismatch identifier 402 is illustrated schematically. The spatial mismatch identifier 402 includes a dataset provider 502. The dataset provider 502 is configured to acquire, determine, generate, etc., two image datasets for determining relationships and generating spatial masks. In the illustrated example, the dataset provider 502 employs one or more algorithms from a dataset algorithm 504. In the example, the dataset algorithm 504 includes at least a first algorithm 506, a second algorithm 508, a third algorithm 510, ...

[0069] Using the first algorithm 506, one of the two image datasets includes an original full-volume PET reconstruction with attenuation correction using the original CT image data, and the other of the two image datasets includes a full-volume PET reconstruction with modified attenuation correction using modeled reconstructed CT image data.

[0070] Using the second algorithm 508, the first image dataset is the original CT image data (or the attenuation map itself) used for attenuation correction, and the second image dataset is the modeled reconstructed CT image data (or the modified attenuation map itself) used for modified attenuation correction. Typically, the second algorithm 508 is similar to the first algorithm 506, but employs different reconstruction steps using the same data generated by the first algorithm 506.

[0071] Using the third algorithm 510, the first image dataset consists of PET projections, LOR, or sinograms of the raw emission data, which are affected only by the true physical attenuation values ​​and not by the modeled attenuation correction. The second image dataset consists of corresponding (comparable) PET projections, LOR, or sinograms of the updated reconstruction step using the raw CT image data with modeled attenuation correction. The following section combines... Figure 20 A more detailed example of the third algorithm 510.

[0072] Voxel identifier 512 is configured to identify one or more spatial voxels that satisfy a predetermined criterion 514 for reconstruction inconsistency. In one case, the inconsistency significance level criterion or threshold is predetermined based on, for example, a functional image value scale relative to the median background in soft tissue or based on an absolute PET SUV scale. Mask generator 516 is configured to generate a spatial mask based on the identified one or more spatial voxels. Typically, the spatial mask labels all these regions. Local symbols of the differences between two image sets may also be recorded. The process may include spatial filtering or other image processing steps for the desired mask characteristics. The final mask may include binary values ​​(e.g., 0 or 1) or continuous relative weights (e.g., in the range between 0 and 1). Continuous weights can more precisely influence subsequent mask-based algorithmic steps.

[0073] Figure 6 An example of a PET image data generator 404 is illustrated schematically. The PET image data generator 404 includes a principal direction determiner 602, a principal direction standard 604, a line segment determiner 606, a local region of interest identifier 608, a deformation scheme identifier 610, one or more deformation algorithms 612 (which include schemes for merging data deformed using different deformation algorithms), and an image data generator 614.

[0074] The main direction determiner 602 receives the information from the spatial mismatch identifier module 402. Figure 4 and Figure 5The generated spatial mask is used as input. The principal orientation determiner 602 is configured to determine the principal orientation (in 2-D and / or 3-D) for each voxel or set of voxels in the spatial mask based on a pre-determined principal orientation criterion 604. Examples of pre-determined principal orientation criteria include, but are not limited to, proximity to pre-determined vicinity and / or proximity to associated anatomical image data on a natural patient motion model.

[0075] The line segment determiner 606 is configured to determine a set of line segments (in 2-D and / or 3-D) related to the width and / or shape of a local region in a spatial mask and the margins from both sides of the local region, each line segment along a corresponding principal direction. In the illustrated example, the principal direction and / or the set of line segments are processed. Examples of such processing include smoothing, adjusting, and regularizing the determined principal direction and / or the determined set of line segments, etc.

[0076] The local region of interest (ROI) identifier 608 is configured to evaluate regions of the functional image for each defined line segment along a corresponding principal direction, and to identify voxels to be deformed and those not to be deformed. For example, in one case, the ROI identifier 608 evaluates local functional image structures or regions that have specific characteristics of morphology, localization, and relative intensity that should be preserved to maintain sufficient clinical diagnostic image quality, and evaluates local functional image regions that contain background or blurred structures and can therefore be deformed without degrading the quality of the diagnostic image.

[0077] The deformation scheme identifier 610 is configured to identify functional image deformation schemes from one or more deformation algorithms 612 for different classifications of structural or region types within the determined line segment, and to determine a method for the continuous merging of deformation schemes. The image data generator 614 is configured to apply the identified functional image deformation schemes and merging methods to the functional image data based on analyzed local characteristics, so as to deform the original functional image data and generate deformed (new) functional image data.

[0078] Next, combine Figure 7 , Figure 8 , Figure 9 , Figure 10 , Figure 11 and Figure 12 Describe an example variation of the spatial mismatch identifier 402. Figure 7 and Figure 8 Two images provided by dataset provider 502 are illustrated schematically. Figure 7 A first image 702 is schematically illustrated. In this example, the first image is a first reconstructed PET image having a first relationship with a modeled attenuation value. Figure 8A second image 802 is schematically illustrated; in this example, it is a second reconstructed PET image having a second relationship with modeled attenuation values. In this example, these attenuation values ​​are derived from anatomical CT image data with arbitrary function-anatomical mismatches at several locations. The first image 702 and the second image 802 can be different due to the different modeled attenuation values ​​depending on the function-anatomical spatial mismatch being analyzed.

[0079] Figure 7 and Figure 8 Both showed liver dome 704 and lung lesion 706. Figure 7 The liver dome 704, which terminates at the first positioning 708 relative to reference frame 710, is shown, and Figure 8 The liver dome 704 is shown terminating at different second positions 804 relative to reference frame 710. Additionally, Figure 7 A lung lesion 706 is shown at a first distance 712 from the liver dome 704, and... Figure 8 A lung lesion 706 is shown at a second distance 806 from the liver dome 704. In this example, the second location 804 of the liver dome 704 is more accurate than the first location 708 of the liver dome 704, and the second distance 806 is more accurate than the first distance 712.

[0080] As discussed herein, spatial mismatch identifier 402 determines the differences, ratios, and / or other relationships between image data 702 and image data 802 to analyze functional-anatomical spatial mismatches of interest to generate a spatial mask. Figure 9 An example spatial mask 902 is schematically illustrated, generated based on image data 702 and 802 using difference relationships. Spatial mask 902 includes regions 904, 906, 908, and 910, which represent differences in images 702 and 802 that meet predetermined criteria. In some cases, spatial mismatch identifier 402 additionally utilizes anatomical image data, including corrected anatomical image data, to create mask 902.

[0081] Figure 10A spatial mask 902 and a reference frame 710 are schematically illustrated overlaid on anatomical image data 1002. A principal orientation determiner 602 determines at least one principal orientation for each of regions 904, 906, 908, and 910 in the spatial mask 902 based on analysis relative to proximal anatomical structures in the anatomical image data 1002. For example, the principal orientation determiner 602 determines principal orientations 1004 and 1006 for region 904, principal orientation 1008 for region 906, principal orientation 1010 for region 908, and principal orientations 1012 and 1014 for region 910. In this example, the spatial mask 902 is independent of the voxel values ​​of the absolute functional image data and only involves differences due to inconsistencies in attenuation correction. Figures 13 to 16 The document describes a more detailed example for determining the principal direction.

[0082] Figure 11 An enlarged view 1102 illustrates a portion of the mask 902 superimposed on functional image data 802. A line segment determiner 606 determines at least one line segment for each of regions 904, 906, 908, and 910 based on the widths of regions 904, 906, 908, and 910 and the main directions 1104, 1106, 1108, 1110, 1012, and 1104. For example, the line segment determiner 606 determines a first line segment 1104 and a second line segment 1106 for region 904. The line segment determiner 606 identifies a set of segments to be deformed for each of line segments 1104 and 1106. For example, the line segment determiner 606 determines a first segment 1106 on the width of region 904, a second segment 1108 on a first side of the first region 904, and a third segment 1110 on the opposite second side of the first region 904 for line segment 1104.

[0083] Figure 12 An example of deformed functional image data 1202 is shown. Image data generator 614 applies one or more deformation schemes (hereinafter combined) along each of the line segments in regions 904, 906, 908, and 910 and based on the main directions 1004, 1006, 1008, 1010, 1012, and 1014. Figure 17 and Figure 18 (To be described in more detail). The image data generator 614 then combines the deformed segments to generate a continuous volume of functional image data.

[0084] refer to Figure 7 , Figure 8 and Figure 12 In contrast Figure 8 The end positioning 802 of the liver dome 704 in the deformed functional image data more closely matches the end position 1204 of the liver dome 704 relative to the reference frame 710. Figure 7 The end position 710 of the liver dome 704 in the image. Additionally, relative to... Figure 7 The distance of 712 in the deformed functional image data is more closely matched to the distance of 1206 in the deformed functional image data. Figure 8 The distance is 804. As discussed above, in this example, ending position 708 and distance 804 more accurately represent the true position and distance than ending position 802 and distance 712. Therefore, the deformation is corrected for the initial spatial mismatch.

[0085] Figure 13 , Figure 14 , Figure 15 and Figure 16 A graphical example of a method for determining the principal direction is illustrated. Figure 13 Region 1302 of the spatial mask superimposed on the anatomical image data is illustrated graphically. Figure 14 In the process, a volumetric filter and thresholding are applied to region 1302, leaving a first inner sub-region 1402 and a second inner sub-region 1404, both with high values ​​relative to the rest of the spatial mask 902. Local maxima are identified in the filtered mask. For each identified local maxima, growth spheres 1406 and 1408 are used to identify the nearest soft tissue distribution or another tissue type of interest. Tissue types of interest can be, for example, lung parenchyma, muscle tissue, or a specific visceral organ. Identification of tissue types of interest can be aided by anatomical segmentation methods or a predetermined range of anatomical image values. Within these spheres 1406 and 1408, the centroid is calculated relative to the center point to determine the principal orientation. In one case, various techniques are applied to focus on the most relevant soft tissue components, such as filtering out narrow vessels that may not be relevant to the assessment of the mismatch direction.

[0086] Figure 15 The diagram graphically illustrates the main directions 1502 and 1504 for two points 1506 and 1508. Additional points can be added if the filtered mask value of an additional point equals the maximum value. Figure 16 In this process, for each point in the original region 1302, the nearest point on the filtered region is detected and assigned the same principal direction. For example, point S1 is closer to point P2 than point P1, and therefore has the same principal direction as P2. In this example, the segment passing through point S1 will deform towards the spleen region rather than towards the adjacent rib (visible as saturated white on the CT image). In another case, a weighted or average principal direction is assigned to voxels among several maximum points based on their relative distances to adjacent points. For example, S2 could have a weighted average of the directions of P1 and P2, 1602.

[0087] Figure 17 and Figure 18The example variation scheme is illustrated graphically. Figure 17 The diagram graphically depicts segments having line segments above the mask region before deformation and segments on opposite sides of the mask (e.g., Figure 11 The distribution of the anatomical structure of line segment 1104 and segments 1106, 1108 and 1110 is shown in P1. Figure 18 A distribution P2 graphically depicts the anatomical structure of the area above the mask and the segments on opposite sides of the mask after deformation. Distributions P1 and P2 are 1-D distributions (i.e., values ​​along the lines). However, 2-D or 3-D methods are also anticipated in this paper. In other examples, deformation schemes may include different options, more complex image processing and computer vision algorithm steps, additional pre-processing or post-processing steps including 3-D structural analysis, etc.

[0088] First refer to Figure 17 Distribution P1 represents the distribution of the original functional image data. The line segment is divided into a first segment a1, a second segment b1, and a third segment c1. The first segment a1 runs along line segment 1104 ( Figure 11 The width of segment a1 matches the width of segment 1106 of spatial mask 902, the width of segment b1 matches the width of region 1110 on the first side of spatial mask 902, and the width of segment c1 matches the width of region 1108 on the opposite second side of spatial mask 902. In this example, Wa is the width of the first segment a1, Wb is the width of region b1, and Wc is the width of region c1. In this example, Wb = X * Wa and Wc = Y * Wa, where X and Y are values ​​greater than zero and are either the same or different.

[0089] Go to Figure 18 Distribution P2 represents the distribution of deformed functional image data along a line segment using a deformation scheme that maintains diagnostic image quality. In this example, the deformation involves rigidly shifting segment a1 by a distance d to generate segment a2. Local rigid shifting can preserve the accurate original image pattern of the edges of important organs. In one case, the distance d is a predetermined constant portion of Wa. For example, d can be between 0.5 and 1.0 of Wa. In the illustrated example, d is approximately 0.75 of Wa. The deformation also includes shrinking all voxels in segment b1 (e.g., via interpolation, including linear, nonlinear, etc.) to create segment b2. The deformation also includes conditionally expanding some voxels in segment c1 (e.g., via interpolation, including linear, nonlinear, etc.) to create segment c2, where the local condition is related to a predetermined relative value threshold T.

[0090] For example, T can be set as the median of the values ​​in segment c2, another percentile, etc. Dominant image structures with values ​​above the threshold T are preserved without expansion or shifting, while values ​​between them are interpolated along the new segment range. In this way, the probability of maintaining the correct location, size, and structure of potentially significant lesions is increased. Background values ​​are interpolated, but their exact structure and location are less important for clinical diagnostic purposes. L1, L2, and L3 represent structures. Figure 18 In this process, structures L1 and L3 are preserved, while structure L2 is reduced in size through interpolation. This may be reasonable in practice if segment b2 primarily contains characteristic elastic organs such as the liver, spleen, stomach, or heart.

[0091] In another case, the algorithm determines that within segment c2, if parts of image structure L3 with the dominant value are relatively close to a2, they will be rigidly shifted (by a constant d) to maintain the same distance from segment a2 (e.g., equal to the distance in P1), and if other structural parts are relatively far from a2, they will remain in place (e.g., as illustrated). Figure 11 , Figure 17 and Figure 18 In the first case, deformation is performed along parallel line segments. However, in another case, the line segments are not parallel to each other and are not aligned with the image voxel grid. In this case, the mapping of data to the new voxel can be calculated using interpolation of a set of nearby points around the target voxel.

[0092] Figure 19 An example geometry illustrating the image data deformation process is shown. The image data is distributed within a voxel grid. M is part of a spatial mask representing significant differences in a certain region. Each voxel has an assigned principal direction, which may not be parallel to the voxel grid and may be equal to or different from its neighboring voxels. In this example, voxels in columns V1 and V2 have the same principal direction with angle a1, and voxels in columns V3 and V4 have the same principal direction with angle a2, where a2 is different from a1.

[0093] Segment s1 is transformed into segment s2, and segment s3 is transformed into segment s4. The set of points p1 and p3 along segments s1 and s3 does not coincide with the image voxel grid. Their values ​​are calculated by interpolation of neighboring voxels (e.g., using methods such as nearest neighbor, linear, cubic, etc.). New transformed data is first calculated on the set of points p2 and p4 along segments s2 and s4. These points also do not coincide with the image voxel grid. New image data is constructed by interpolating the set of points p2 and p4 into the coordinates of the image voxel grid. For example, the new value of voxel t will be the weighted average (or another interpolation technique) from the nearest points p2 and p4.

[0094] As mentioned above Figure 5 As described, the spatial mismatch identifier 402's dataset provider 502 provides two image datasets for determining relationships and generating a spatial mask 902 based on at least one of the dataset algorithms 504. Combined Figure 20 An example of the third algorithm 510 is described. Again, for this algorithm, the first image dataset is a PET projection, LOR, or sine plot of the raw emission data that is affected only by the true physical attenuation value and not by the modeled attenuation correction, and the second image dataset is a corresponding (comparable) PET projection, LOR, or sine plot of the updated reconstruction step using the raw CT image data with modeled attenuation correction.

[0095] In this example, the dataset provider 502 receives PET emission data and corresponding anatomical image data as input. The dataset provider 502 evaluates and corrects common artifacts in PET data caused by misregistration between PET and anatomical data by locating regions of artifacts based on error image data from the reconstructed PET image data and deforming (estimated motion) the CT image data using localization and air-tissue boundaries from the CT image data to generate artifact-free PET image data. The dataset provider 502 includes a reconstructor 2002 (or... Figure 1 The PET reconstructor 152), error image data generator 2004, mismatch identifier 2006, inconsistency identifier 2008, and anatomical image data corrector 2010.

[0096] Reconstructor 2002 reconstructs the emission data to generate estimated functional image data. In one case, reconstructor 2002 is configured to perform a standard reconstruction. In another case, reconstructor 2002 is configured to perform fewer iterations than a standard reconstruction. Alternatively, the estimation can be completed during the “standard” reconstruction and trigger a correction process. Error image data generator 2004 generates error image data based on the estimated functional image data. In one case, this includes forward projection of the estimated functional image data, optionally applying corrections (e.g., attenuation correction, scattering correction, normalization, etc.) to the forward projection, determining errors based on the corrected projection and emission data, and back-projecting the error sine curve to generate the error image data. The error image data reveals how well the estimated functional image data interprets the acquired emission data. Regions in the error image data that are higher than other regions (typically, in terms of their absolute values, as they can be positive or negative in direction) indicate inconsistencies in the data.

[0097] The mismatch identifier 2006 is configured to process anatomical image data and create a mask that identifies regions with a high probability of mismatch in attenuation data that may manifest as artifacts in functional image data. Typically, the mismatch identifier 2006 finds and thickens the contour lines of air and tissue boundaries. This step assumes that major artifacts typically occur when mismatches are present in the data, where there is a significant difference in attenuation between the anatomical image and the emission data, often due to changes in lung volume and shape during the respiratory cycle.

[0098] Inconsistency Identifier 2008 processes cross-axis slices of the error image and identifies high-inconsistency regions (i.e., voxels with large errors compared to their neighboring voxels). To do this, Inconsistency Identifier 2008 first applies a mask to the error image and then, for each cross-axis slice, normalizes the voxel values ​​to identify which voxels are outliers in the distribution. Alternatively, the processing can be applied directly to the 3D volume of the error image using sufficient 3D operators. Due to the mismatch between the emission data and the anatomical image, the identified voxels are suspected of suffering from artifacts. Inconsistency Identifier 2008 applies morphological operations to filter out small clusters of misidentified voxels.

[0099] The anatomical image data corrector 2010 modifies anatomical image data to account for motion based on local error voxels. In one case, this involves segmenting or clustering image voxels or regions into different types of tissues or organs based on anatomical image values ​​or anatomical structure models, and modifying the anatomical values ​​corresponding to identified high-inconsistency regions to values ​​corrected for the tissue type. The value modification is based on a pre-determined anatomical image value correction scheme corresponding to the type of tissue or organ. For example, this may also include calculating a histogram of the anatomical image data and quantizing the histogram into multiple bins, including air bins, lung bins, soft tissue bins, and bone bins. The anatomical image data corrector 2010 determines whether the corresponding voxel in the attenuated image is associated with the lung cluster for each voxel in the segmented mismatch mask. The anatomical image corrector 2010 changes the voxel value to the average of the soft tissue bins of the voxels associated with the lung bin. Otherwise, the voxel value remains unchanged.

[0100] Reconstructor 2002 uses corrected anatomical image data for attenuation correction to reconstruct the input emission data, thereby generating attenuation-corrected functional image data. In this example, dataset provider 502 ( Figure 5 The attenuation-corrected functional image data is provided to the voxel identifier of interest 512. Figure 5 ) for further processing, for example, in conjunction with a spatial mismatch correction module ( Figure 4 The processing described.

[0101] Next, combine Figure 21 , Figure 22 , Figure 23 , Figure 24 , Figure 25 , Figure 26 , Figure 27 , Figure 28 , Figure 29 and Figure 30 describe Figure 20 Examples of methods. Figure 21 Depicting features created by Reconstructor 2002 ( Figure 20 Reconstructed and generated by Error Image Data Generator 2004 Figure 20 Example functional image data used to generate mismatch artifacts in error image data. Figure 22 The image data generator 2004 (with errors) is described. Figure 20 Example error image data generated. Figure 23 Depicting the mismatch identifier 2006 ( Figure 20 Outlined example anatomical image data. Figure 24 Depicting the mismatch identifier 2006 ( Figure 20 Example mask generated from contoured anatomical image data. Figure 25 Example masks depicting regions with a high probability of mismatch in attenuation data, for example, with predetermined margins (±1, 3, ... cm) regarding the boundary contour.

[0102] Figure 26 The error image data depicted has a mask superimposed on it. Figure 22 The mask has a region with a high probability of mismatch. Figure 25 ). Figure 27 Depicting the Inconsistency Identifier 2008 ( Figure 20 The potentially mismatched region of the segment. Figure 28 Depicting the Inconsistency Identifier 2008 ( Figure 20 Morphological operations are applied to filter out misidentified voxels from the clustered segmentation regions. Figure 29 The original functional image data with mismatch artifacts 2902 and 2904 caused by anatomical image data were depicted. Figure 30 Depicting based on anatomical image data corrector 2010 ( Figure 20 The generated anatomical image data is functional image data for mismatch artifact correction.

[0103] In one case, the third algorithm 510 does not require list data and works directly on the sine wave and the image in the reconstruction loop. In a variant, the algorithm can also be applied to reconstructions directly based on list data. Additionally or alternatively, the third algorithm 510 can use short initial reconstructions where the computational cost of segmentation and editing of the anatomical image is negligible. Additionally or alternatively, the third algorithm 510 makes minimal and reasonable assumptions about regions where artifacts are found. Additionally or alternatively, the third algorithm 510 makes no assumptions about the nature of the motion model that causes unregistered regions. Additionally or alternatively, the third algorithm 510 works well and is compatible with attenuation corrections (such as the enhanced AC feature of US2024 / 0046535 A1), for example, to increase the confidence level of segmentation of artifact regions.

[0104] While reconstruction inconsistency analysis is used to determine the areas for deformation, additional information from functional or anatomical image data derived from organ detection and segmentation can also be utilized. For example, such information can be used to identify specific areas that should not be deformed, or areas that should be deformed using only rigid transformations, such as solid bones. In this case, a process can be determined to combine adjacent deformable and non-deformable areas in a continuous or smooth manner.

[0105] Regarding the sign of the dominant direction for a specific location (e.g., in the region between the liver and lungs, if the direction is toward the liver or toward the lungs), the direction sign can be determined by the sign of the difference between the two input image data. For example, if the PET value in the second image is greater than the value in the first image (i.e., in the respiratory mismatch artifact region), the deformation direction is toward the liver. However, if the PET value in the second image is less than the value in the first image, the deformation direction is toward the lungs (i.e., the mismatch is due to the difference in motion in opposite directions, as may occur in a CT scan during full expiration). In this case, the spatial mask area will primarily reside on the anatomical soft tissue area, and a search for proximal tissues can be determined for the "air" region (or the low HU value range of the lungs).

[0106] For a given set of main directions and associated segments, an anatomical organ motion model can be used, which can improve overall accuracy, or it can be used to regularize the shape and distribution of a given set of deformed segments.

[0107] Although the method described above deforms PET image data to better match the original CT image data, the same or similar methods can be used to deform CT image data to better match the original, unaltered PET image data.

[0108] For example, via a dedicated application, deformed PET image data can be linked or registered to the original PET image data, where the pointer to the position in the deformed data will automatically indicate the corresponding position in the original (undeformed) PET image data.

[0109] Reconstruction inconsistency analysis can also be used (independent of the deformation process) to correct attenuation maps used for PET reconstruction.

[0110] Depending on which algorithm is used—namely, algorithm 506, algorithm 508, algorithm 510, another algorithm, or a combination thereof—the spatial mask can be similar. Therefore, a joint mask can be generated by a weighted combination of multiple masks.

[0111] Figure 31 A non-limiting example flowchart is provided for a computer-implemented method of deforming functional image data to match corresponding anatomical image data independently of function-anatomical correlation. It should be understood that the order of actions in this method is not restrictive. Therefore, other orders are contemplated herein. Furthermore, one or more actions may be omitted, and / or one or more additional actions may be included.

[0112] At 3102, functional image data is obtained as described herein and / or otherwise. At 3104, corresponding anatomical image data is obtained as described herein and / or otherwise. At 3106, spatial voxels or regions of functional image data with reconstruction inconsistencies caused by spatial mismatch between the true attenuation values ​​and the attenuation values ​​derived from the anatomical image data are identified as described herein and / or otherwise. At 3108, a spatial mask is created using the identified voxels or regions as described herein and / or otherwise.

[0113] At point 3110, as described herein and / or otherwise, while maintaining the diagnostic image quality of the original functional image data, the functional image data is deformed based on a spatial mask so that the volumetric spatial conformation of the functional image data better matches the volumetric spatial conformation of the anatomical image data. The deformed functional image data may be displayed, captured, archived, etc., with and / or without the anatomical image data. As discussed herein, the deformed functional image data maintains the original image quality of the original functional image data.

[0114] Figure 32 Examples are shown for generating Figure 31 The flowchart below is a non-limiting example of a computer-implemented method for a spatial mask of action 3108. It should be understood that the order of actions in this method is not restrictive. Therefore, other orders are contemplated herein. Additionally, one or more actions may be omitted, and / or one or more additional actions may be included.

[0115] At 3202, as described herein and / or otherwise, for each voxel or group of voxels in the mask generated in action 3108, and based on proximity to associated anatomical image data in its vicinity or based on a natural patient motion model, a principal direction is determined for the deformation. At 3204, as described herein and / or otherwise, a set of line segments along the principal direction is determined for the deformation, wherein the line segments are related to the local mask width or shape and margins from both sides of the mask. At 3206, as described herein and / or otherwise, the principal direction and / or line segments are smoothed, adjusted, and / or regularized. In another case, action 3206 is omitted.

[0116] At 3208, as described herein and / or otherwise, each line segment along the principal direction is evaluated to identify local functional image structures or regions with specific characteristics of morphology, location, and relative intensity, which should be preserved to maintain the quality of the clinical diagnostic image. At 3210, as described herein and / or otherwise, each line segment along the principal direction is evaluated to identify local functional image regions containing uptake values ​​of background or blurred structures and which can be deformed without degrading the quality of the diagnostic image.

[0117] At 3212, as described herein and / or otherwise, functional image deformation schemes for different classification structures or region types within the determined line segment and techniques for merging them are determined. At 3214, as described herein and / or otherwise, local characteristics are used to deform the functional image data based on the deformation scheme. The deformed functional image data may be displayed, photographed, archived, etc., with and / or without anatomical image data. As discussed herein, the deformed functional image data maintains the original image quality of the original functional image data.

[0118] Figure 33 Examples are given for use by Figure 20 The flowchart below is a non-limiting example of a computer-implemented method for generating error image data using an error image data generator 2004. It should be understood that the order of actions in this method is not restrictive. Therefore, other orders are contemplated herein. Additionally, one or more actions may be omitted, and / or one or more additional actions may be included.

[0119] At 3302, as described herein and / or otherwise, unregistered anatomical image data is used to reconstruct functional emission data to generate estimated image data. At 3304, as described herein and / or otherwise, error image data is generated based on emission data and estimated image data. At 3306, as described herein and / or otherwise, regions of attenuation mismatch between attenuated image data and emission data are identified.

[0120] At 3308, as described herein and / or otherwise, error image data is used to identify inconsistency areas. At 3310, as described herein and / or otherwise, the anatomical image is corrected based on local mismatch areas. At 3312, the corrected anatomical image data is used to reconstruct emission data for attenuation correction.

[0121] Figure 34 Examples are given for Figure 33 The flowchart of a computer-implemented method for generating error image data (action 3304) is a non-limiting example. It should be understood that the order of the actions in this method is not restrictive. Therefore, other orders are contemplated herein. Additionally, one or more actions may be omitted, and / or one or more additional actions may be included.

[0122] At 3402, the estimated functional image data is forward-projected as described herein and / or otherwise. At 3404, the forward projection is corrected as described herein and / or otherwise. In another example, action 3404 is omitted. At 3406, the error projection is determined based on the corrected forward projection and the measured emission data as described herein and / or otherwise. At 3408, the error projection is back-projected as described herein and / or otherwise to generate an error image.

[0123] Figure 35 Examples are shown for identifying targets Figure 33 The flowchart below is a non-limiting example of a computer-implemented method for the inconsistency region of action 3308. It should be understood that the order of actions in this method is not restrictive. Therefore, other orders are contemplated herein. Additionally, one or more actions may be omitted, and / or one or more additional actions may be included.

[0124] At 3502, a histogram is generated for the anatomical image data, as described herein and / or otherwise. At 3504, the histogram is quantized into four bins, including an air bin, a lung bin, a soft tissue bin, and a bone bin, as described herein and / or otherwise. At 3506, each voxel is evaluated to determine whether it is in the lung bin, as described herein and / or otherwise. At 3508, for each voxel determined to be in the lung bin, its value is changed to the average value of the soft tissue bin, as described herein and / or otherwise. At 3510, for each voxel determined not to be in the lung bin, its value remains unchanged, as described herein and / or otherwise.

[0125] The above method can be implemented by computer-readable instructions encoded or embedded on a computer-readable storage medium, which, when executed by a computer processor, cause the processor to perform the described action or function. Additionally or alternatively, at least one of the computer-readable instructions may be executed by a signal, a carrier wave, or other transient medium that is not a computer-readable storage medium.

[0126] As used herein, elements or steps listed in the singular and beginning with the word "a" or "an" should be understood to not exclude multiple said elements or steps unless such exclusion is explicitly stated. Furthermore, references to "an embodiment" of the invention are not intended to be construed as excluding the existence of additional embodiments that also include the referenced features. Moreover, unless explicitly stated to the contrary, embodiments that "comprise," "include," or "have" one or more elements having a particular attribute may include such additional elements that do not have that attribute. The terms "comprise" and "in" are used as concise linguistic equivalents to the corresponding terms "comprising" and "wherein". Furthermore, the terms "first," "second," and "third," etc., are used merely as notations and are not intended to impose numerical requirements or a particular order of position on their objects.

[0127] Various implementations and / or components (e.g., modules, components therein, and controllers) may also be implemented as part of one or more computers or processors. A computer or processor may include computing devices, input devices, display units, and interfaces, such as for accessing the Internet. A computer or processor may include a microprocessor. The microprocessor may be connected to a communication bus. A computer or processor may also include memory. Memory may include random access memory (RAM) and read-only memory (ROM). A computer or processor may also include storage devices, which may be hard disk drives or removable storage drives, such as floppy disk drives, optical disk drives, etc. Storage devices may also be other similar means for loading computer programs or other instructions into the computer or processor.

[0128] As used herein, the terms "computer" or "module" can include any processor-based or microprocessor-based system, including systems using microcontrollers, reduced instruction set computers (RISCs), application-specific integrated circuits (ASICs), logic circuits, and any other circuitry or processor capable of performing the functions described herein. The examples above are merely illustrative and are therefore not intended to limit the definition and / or meaning of the term "computer" in any way. A computer or processor executes a set of instructions stored in one or more storage elements to process input data. Storage elements may also store data or other information as desired or required. Storage elements may take the form of an information source within a processor or a physical memory element.

[0129] An instruction set may include various commands that instruct a computer or processor to perform specific operations (such as the methods and processes of various embodiments of the present invention) as a processing machine. The instruction set may be in the form of a software program. Software may take various forms, such as system software or application software. Furthermore, software may take the form of a collection of separate programs or modules, a program module within a larger program, or a portion of a program module. Software may also include modular programming in the form of object-oriented programming. The processor's processing of input data may be in response to operator commands, the results of previous processing, or a request from another processor.

[0130] As used herein, the terms “software” and “firmware” are interchangeable and include any computer program stored in memory for execution by a computer, including RAM memory, ROM memory, EPROM memory, EEPROM memory, and non-volatile RAM (NVRAM) memory. The memory types described above are merely exemplary and therefore do not limit the types of memory that can be used to store computer programs.

[0131] It should be understood that the above description is intended to be illustrative and not restrictive. For example, the embodiments (and / or aspects thereof) described above can be used in combination with each other. Furthermore, many modifications can be made to adapt particular situations or materials to the teachings of various embodiments of the invention without departing from the scope of the invention. While the dimensions and types of materials described herein are intended to define parameters of various embodiments of the invention, these embodiments are by no means restrictive but rather exemplary. Many other embodiments will be apparent to those skilled in the art upon review of the above description.

[0132] This written description uses examples to disclose various embodiments of the invention, including the best mode, and also enables those skilled in the art to practice various embodiments of the invention, including making and using any device or system and performing any included methods. The patent scope of the various embodiments of the invention is defined by the claims, and may include other examples that would occur to those skilled in the art. Such other examples are intended to fall within the scope of the claims if they have structural elements that are not indistinguishable from the literal language of the claims, or if the examples include equivalent structural elements that differ only slightly from the literal language of the claims.

[0133] The embodiments illustrated in the accompanying drawings and described above are merely illustrative embodiments and are not intended to limit the scope of the appended claims, including any equivalents included within the scope of the claims. Various modifications are possible and will be apparent to those skilled in the art. Any combination of the non-mutually exclusive features described herein is intended to be within the scope of this disclosure. That is, features of the described embodiments may be combined with any suitable aspect described above, and optional features of any aspect may be combined with any other suitable aspect. Similarly, features listed in dependent claims may be combined with non-mutually exclusive features of other dependent claims, particularly where the dependent claims are subordinate to the same independent claim. In some jurisdictions that claim single-claim dependents, such dependents may have been used in practice, but this should not be construed as meaning that the features in the dependent claims are mutually exclusive.

Claims

1. A system (102), the system comprising: A spatial mismatch correction module (168) is configured to receive functional transmission data, anatomical image data, and functional image data reconstructed based on the functional transmission data and attenuated based on the anatomical image data; A dataset provider (502) is configured to provide a first dataset and a second dataset, wherein the first dataset and the second dataset include spatial mismatch; Voxel identifier (512), the voxel identifier being configured to identify voxels or regions of reconstruction inconsistency caused by spatial mismatch between the true attenuation value and the attenuation value derived from the anatomical image data, based on the relationship between the first dataset and the second dataset; and An image data generator (614) is configured to deform the functional image data and generate corrected functional image data based on identified voxels or regions, while maintaining the image quality of the functional image data and independent of function-anatomical correlation.

2. The system of claim 1, wherein the first dataset and the second dataset comprise one of the following: Reconstructed functional image data attenuated and corrected using the anatomical image data, and reconstructed functional image data attenuated and corrected using the corrected anatomical image data; The anatomical image data and the corrected anatomical image data; and The functional emission data and the reconstructed functional image data attenuated and corrected using the anatomical image data.

3. The system according to claim 1, wherein the image data generator is further configured to: Generate a spatial mask based on the identified voxels or regions; The main direction is determined based on the spatial mask; The set of line segments is determined based on the spatial mask; A first set of voxels with values ​​to be retained to maintain the image quality of the functional image data and a second set of voxels with values ​​to be deformed without degrading the image quality of the functional image data are identified based on the main direction and the set of line segments. as well as The second set of voxels is deformed.

4. The system of claim 3, wherein the image data generator is configured to determine the principal orientation based on the spatial mask by: Identify the local maxima in the spatial mask; For each maximum value, identify the closest organization type of interest; and For each voxel of the mask, a main direction is assigned based on the local maximum value and the nearest soft tissue.

5. The system of claim 3, wherein the set of line segments includes a first segment overlapping the mask, a second segment on one side of the first segment, and a third segment on the opposite side of the first segment.

6. The system of claim 5, wherein the image data generator uses rigid translation to deform the first segment, uses rigid translation, compression, expansion, or a combination thereof to deform the second segment, and uses rigid translation, compression, expansion, or a combination thereof to deform the third segment.

7. The system of claim 1, wherein the image data generator uses a voxel grid to deform the functional image data.

8. The system of claim 1, wherein the dataset provider is configured to determine the second dataset by: Unregistered anatomical images are used to reconstruct the estimated functional image data; Error image data is generated based on the estimated functional emission data and the functional emission data; Identify the mismatched areas in the anatomical images; Inconsistency areas are identified based on the mismatch region and the error image data; The anatomical image data is corrected based on the mismatch region and the inconsistency region; as well as The functional emission data is reconstructed using corrected anatomical image data to generate the second dataset.

9. The system of claim 8, wherein the dataset provider is configured to generate the error image data by: Forward projection is performed on the estimated functional image data; The error projection is determined based on the estimated forward projection and the functional emission data; as well as The error projection is then reverse-projected.

10. The system of claim 8, wherein the dataset provider is configured to correct the anatomical image data by: The image voxels or regions in the anatomical image data are segmented or clustered into tissue or organ types; Determine an anatomical image value correction scheme corresponding to the type of the said tissue or organ; and The anatomical image data values ​​corresponding to the identified high inconsistency areas are modified based on the determined anatomical image value correction scheme.

11. A computer-implemented method, the computer-implemented method comprising: Receive functional transmission data, anatomical image data, and functional image data reconstructed based on the functional transmission data and attenuated based on the anatomical image data; Provide a first dataset based at least on the anatomical image data and a second dataset based at least on the functional emission data or modified anatomical image data; Based on the relationship between the first dataset and the second dataset, voxels or regions with reconstruction inconsistencies caused by spatial mismatch between the actual attenuation value and the attenuation value derived from the anatomical image data are identified. as well as While maintaining the image quality of the functional image data, the functional image data is deformed based on the identified voxels or regions to generate deformed functional image data.

12. The computer-implemented method according to claim 11, further comprising: Generate a spatial mask based on the identified voxels or regions; The main direction is determined based on the spatial mask; The set of line segments is determined based on the spatial mask; A first set of voxels with values ​​to be retained to maintain the image quality of the functional image data and a second set of voxels with values ​​to be deformed without degrading the image quality of the functional image data are identified based on the main direction and the set of line segments. as well as The second set of voxels is deformed.

13. The computer-implemented method according to claim 12, further comprising: Each line segment is drawn as a first segment overlapping the mask, a second segment on one side of the first segment, and a third segment on the opposite side of the first segment; The first segment is rigidly translated; as well as The second and third segments are deformed by using rigid translation, compression, expansion, or a combination thereof.

14. The computer-implemented method according to claim 11, further comprising: The second dataset is determined by the following steps: Unregistered anatomical images are used to reconstruct the estimated functional image data; Error image data is generated based on the estimated functional emission data and the functional emission data; Identify the mismatched areas in the anatomical images; Inconsistency areas are identified based on the mismatch region and the error image data; The anatomical image data is corrected based on the mismatch region and the inconsistency region; as well as The functional emission data is reconstructed using corrected anatomical image data to generate the second dataset.

15. A computer-readable storage medium (166) having computer-executable instructions encoded thereon, which, when executed by a processor (164), cause the processor to: Receive functional transmission data, anatomical image data, and functional image data reconstructed based on the functional transmission data and attenuated based on the anatomical image data; Provide a first dataset based at least on the anatomical image data and a second dataset based at least on the functional emission data or modified anatomical image data; Based on the relationship between the first dataset and the second dataset, voxels or regions with reconstruction inconsistencies caused by spatial mismatch between the actual attenuation value and the attenuation value derived from the anatomical image data are identified. as well as While maintaining the image quality of the functional image data, the functional image data is deformed based on the identified voxels or regions to generate deformed functional image data.

Images