Multispectral zoned reconstruction for SPECT

Abstract

For SPECT reconstruction, regional reconstruction is provided within the modality. Rather than using CT for the structure, SPECT data from one energy may be used to provide structure information or regions for another energy; or in addition to using CT for a structure, SPECT data from one energy may be used to provide structure information or regions for another energy. Regional reconstruction may be combined with model-based multi-energy image formation. Multi-spectral, regional reconstruction is used as an intra-modal imaging method.

Description

Background Technology

[0001] This embodiment relates to single-photon computed tomography (SPECT). Reconstruction of SPECT images is often challenging because the data is characterized by low signal rates and low signal-to-noise ratios. For SPECT imaging, the count rate is limited by the amount of radionuclide (i.e., radioactive material or radiotracer) that can be administered without harming the patient. Some radionuclides (typically alpha and beta emission isotopes and their progeny (e.g., Tb-161, Ho-166, Lu-177, I-131, Ac-225, Pb208, Pb212, At211, I-123, Ga-67, and Y-90)) have complex gamma spectra, including discrete or continuous spectra due to bremsstrahlung. Typically, the most prominent emission peak with the highest per-decay photon yield is used in acquisition and subsequent reconstruction to optimize image formation. As an example, PET is designed for 511 keV gamma rays from positron annihilation, and in SPECT, different collimators are used for different emission energy ranges. For radionuclides with complex spectra, this limits the efficiency and applicability of imaging, as emissions from other energies suitable for imaging are essentially discarded.

[0002] Multi-energy reconstruction can benefit from complex energy spectra. Image blurring can still occur due to resolution degradation during image formation. SPECT images do not necessarily provide structural information. Therefore, SPECT images are often evaluated with the aid of adjacent structural images. Computed tomography (CT) can be used for structural images. In multimodal imaging, CT data can be used as part of SPECT reconstruction, such as by reconstructing different types of tissues or regions separately. Regional reconstruction can improve accuracy or resolution. When regional reconstruction is used for radionuclides with complex energy spectra from alpha and beta emission isotopes and their progeny (e.g., I-123 and Lu-177), enhanced resolution does not necessarily lead to improved quantitative accuracy due to inaccurate image formation models. In fact, enhanced resolution may decrease quantitative accuracy.

[0003] Regional reconstruction using CT with a combination of multiple energies has been provided (see U.S. Patent No. 10,395,353). This requires separate CT imaging to identify the regions. Summary of the Invention

[0004] By way of introduction, the preferred embodiments described below include methods, systems, instructions, and computer-readable storage media for SPECT reconstruction. Intramodal regional reconstruction is provided. Regional reconstruction can be combined with model-based multi-energy image formation. Instead of using CT for structures, SPECT data from one energy can be used to provide structural information or regions for another energy; or, in addition to using CT for structures, SPECT data from one energy can be used to provide structural information or regions for another energy. Multispectral regional reconstruction is used as an intramodal imaging method. Hybrid energy detectors can be used to detect emissions at multiple energies, thereby allowing image formation to be addressed across multiple energy ranges, with higher-energy formation assisting lower-energy image formation.

[0005] In a first aspect, a method for SPECT reconstruction is provided. The SPECT emission of a patient at a first energy and a second energy is detected. An image object is reconstructed regionally from the detected SPECT emission. Regions for regional reconstruction at the second energy are determined from the SPECT emission at the first energy. An image is generated from the image object.

[0006] In one approach, the radionuclide is a progeny of Lu-177, I-131, Tl-201, Tb161, or Ac225, but other radionuclides may also be used. In another approach, a multi-camera SPECT system can be used to detect emissions from such radionuclides, wherein the first camera of the multi-camera SPECT system is configured to detect at a lower energy than the second camera of the multi-camera SPECT system. In another approach, physical collimation is used to detect SPECT emissions at a first energy, and Compton scattering detection is used to detect SPECT emissions at a second energy.

[0007] In another approach, the regional reconstruction includes generating a region map comprising a first reconstruction from SPECT emission at a first energy, forward projecting SPECT emission at a second energy for each region, weighted summing of the forward projections, obtaining an update from a comparison of the weighted summation with the object model, backward projecting the update, and updating the estimate from the update of the backward projection.

[0008] As another approach, region reconstruction involves generating a region map from SPECT emissions at a first energy level, which is lower than a second energy level. For example, SPECT emissions at the first energy level spatially stabilize a region reconstruction from SPECT emissions at a second energy level, where the second energy is higher than the first energy level.

[0009] According to another approach, region-based reconstruction involves modeling the scattering of a second energy, which is higher than the first energy. The SPECT emission at the second energy is segmented according to the region defined by the first energy. This region is defined with a higher spatial resolution than the image object at the second energy's SPECT emission.

[0010] In another method, an initial reconstruction of the SPECT emission at the first energy is performed. The region is then segmented based on the initial reconstruction for sub-regional reconstruction of the SPECT emission at the second energy.

[0011] As another approach, region-based reconstruction includes iterative reconstruction, in which SPECT emissions at a first energy and at a second energy are reconstructed sequentially in each iteration. For example, iterative reconstruction involves a cyclical ordering from the first energy to the second energy in each iteration. The first energy is lower than the second energy, and each iteration has an objective function that includes both the SPECT emissions at the first and second energies.

[0012] Various types of information can be used to generate images. For example, an image can be generated from an image object of SPECT emission at a second energy level. As another example, an image can be generated by combining an image object reconstructed from regions based on detected SPECT emission at a second energy level with another image object reconstructed from detected SPECT emission at a first energy level. In yet another example, an image can be generated from an image object reconstructed from regions based on detected SPECT emission at a second energy level, combined with an image object reconstructed from regions based on detected SPECT emission at a first energy level.

[0013] In a second aspect, a method for SPECT reconstruction is provided. SPECT emissions from a patient are detected. SPECT emissions originate from radionuclides with multiple energies. The image object is reconstructed using multispectral region-based reconstruction. Multispectral region-based reconstruction includes multiple energies and regions derived from a first energy among the multiple energies. An image is generated from the image object.

[0014] In one approach, reconstructing an image object involves iterative reconstruction, where each iteration includes a successive projection operation from low energy to high energy.

[0015] As an alternative approach, the region is derived from a first energy level, which is lower than the second energy level. Multispectral sub-regional reconstruction uses the region derived from the first energy level to reconstruct the second energy level.

[0016] In a third aspect, a medical imaging system for regional reconstruction within a SPECT modality is provided. A detector arrangement is configured to detect emissions from a patient having different energies from a radiotracer. A processor is configured to reconstruct an object representing the patient from emissions detected for at least two energy windows of distributed energies. The reconstruction of the lower energy window within the at least two energy windows is segmented, and the reconstruction of the higher energy window within the at least two energy windows is regionalized based on the segmentation from the reconstruction of the lower energy window within the at least two energy windows. A display is configured to display an image of the reconstructed object.

[0017] According to one method, the detector arrangement is a detector configured to detect using at least two energy windows. In another method, the detector arrangement is a physical collimator and a Compton scattering detector.

[0018] This invention is defined by the following claims, and nothing in this section should be construed as limiting those claims. Further aspects and advantages of the invention are discussed below in conjunction with preferred embodiments and may be claimed thereafter, independently or in combination. Attached Figure Description

[0019] Components and diagrams are not necessarily drawn to scale; instead, the focus is on illustrating the principles of the invention. Furthermore, similar reference numerals are used throughout the figures to designate corresponding parts in different views.

[0020] Figure 1 This is a flowchart of an embodiment of a method for SPECT reconstruction; Figure 2 The illustration shows examples of different detector arrangements used to detect emissions at different energies; Figure 3 The illustration shows iterative region reconstruction according to one embodiment; Figure 4 The illustration depicts a sequential loop of multispectral reconstruction according to one embodiment; and Figure 5 This is a block diagram of an embodiment of a system for SPECT reconstruction of radionuclides with complex or multiple peak spectra. Detailed Implementation

[0021] Intramodal SPECT regional reconstruction is provided. The use of multi-energy or multispectral and regional techniques together improves both the quantitative accuracy and quality of SPECT images. Extending the SPECT procedure to include multi-energy tracers instead of single-energy tracers may not be straightforward. Instead of requiring additional model information, it allows regional information to be obtained from SPECT emission at one energy for use in regional reconstruction at another energy.

[0022] Image formation is specifically designed for gamma emission ranges or specific energies (e.g., 511 keV - PET). A combination of multispectral and regional techniques can be provided by deploying a hybrid image formation spectral detector (see Figure 11 of US Patent Application US 2022-0330909). This fills a gap in multimodal approaches to EMI using CT / MR, as the energy is used as a modality distinguisher rather than CT / MR (e.g., Y90 emits 511 keV, and bremsstrahlung and lutetium-177 compounds may be contaminated with Lu176 and also emit 511 keV, so the system can image 511 keV (i.e., PET) as well as Lu177 or Y90 at very low count rates). Typically, not only is an isotope with a complex emission spectrum not generated, but 100% pure isotopes are also not generated, which additionally creates more complex spectra because the spectral fingerprint depends on the generation mechanism. This combination of multi-energy and region approaches attempts to understand the data by synthesizing and solving image formation equations optimized from energy ranges for multiple image formations, and using higher-energy image formations to assist lower-energy image formations, which typically do not have good yield spatial resolution.

[0023] The combination of regional reconstruction from multiple energy emissions and multispectral reconstruction simultaneously addresses three important issues. One is the inaccuracy of the modeling process in iterative reconstruction, especially for radionuclides with complex energy spectra (e.g., Lu-177, I-123, Ga-67, and Y-90). Another is image blurring due to resolution degradation during image formation. By combining regional reconstruction and a model-based multi-energy image formation model, SPECT image reconstruction is improved. Regional reconstruction provides more accurate segmentation of critical organs, and the improved quantitative accuracy in critical organs provides more accurate dosimetry. The third is the use of intramodal emissions at a single energy level (such as lower energies with greater spatial resolution) to provide these regions. This eliminates the need for CT scans. Resolution and quantitative accuracy are improved without requiring CT or additional modal information for spatial structures. Lower resolution from collimation (e.g., electrical) at higher energies can be improved.

[0024] In reconstructions targeting multiple energy windows, a separate projection with model-based scattering correction is provided for each of the different regions. This combination is processing-intensive for iterative reconstruction. To reduce processing, the forward projection for model-based scattering correction can be handled on the overall image rather than by region. Region-based scattering correction can be used in earlier iterations, while scattering correction based on overall image objects can be used in later iterations. This different strategy of region-specific forward projection in different iterations achieves both high accuracy and fast reconstruction. Alternative methods utilize other multi-energy reconstruction techniques.

[0025] Single-photon image reconstruction is provided using a combination of image forming methods operating at different emission energies. Multiple image forming models are combined in a joint reconstruction as an extension of both additional modal imaging and multi-emission imaging (for regions, the extension to additional modal imaging uses intramodal imaging). For example, this combination enables Compton reconstruction with lower-energy emission regions to achieve better quality than might not use, for example, electron kinematics.

[0026] Figure 1 An embodiment of a method for SPECT reconstruction is illustrated. Multi-energy image formation (e.g., model-based multi-energy image formation) is combined with region-specific reconstruction. Alternatively, multiple energies are used, one for segmentation and another for region-specific reconstruction. Region segmentation relies on SPECT data at one or more energies, rather than additional modal information (e.g., rather than CT). Resolution enhancement and improved quantitative accuracy can be achieved simultaneously without requiring CT imaging.

[0027] use Figure 5 To achieve this, a system, processor, computer, SPECT imager, and / or another device. Figure 1 The method. For example, the SPECT imager performs action 100. A computer (such as a server, workstation, or processor) performs actions 110-120, such as the SPECT imager.

[0028] The method is performed in the order shown (in numerical terms or from top to bottom), but other orders may also be used. Actions 112-116 are performed concurrently as part of action 110, or in any order.

[0029] Additional, different, or fewer actions can be provided. For example, actions are provided for acquiring CT data for mutagrams or attenuation. As another example, actions are provided for modeling attenuation and scattering, or rather, not scattering, in regional reconstruction. In another example, motion correction is performed. As yet another example, instead of performing image generation as in action 120, the image is stored in memory and / or transmitted via a computer network.

[0030] In action 100, SPECT data is acquired. A SPECT scan is performed on the patient. In alternative or additional embodiments, other functional imaging, such as PET and / or Compton scattering, is performed. SPECT data are measurements of single-photon emissions from the patient.

[0031] SPECT data is obtained from scanning, data transfer, or memory. SPECT systems provide SPECT data directly through scanning, or indirectly through transfer or loading.

[0032] As part of the reconstruction via the SPECT system, the active concentration of the radioactive tracer in a patient who has received one or more radioactive tracers can be determined. After the radioactive tracer or tracer is taken up or injected into the patient, the patient is positioned relative to the SPECT detector, and / or the SPECT detector is positioned relative to the patient. Over time, the emission of the radioactive tracer or tracer from within the patient is detected. The lateral position of the line or cone relative to the detector can be determined. The SPECT detector can be rotated or moved relative to the patient, allowing the detection of emission from different angles and / or locations within the patient.

[0033] The emission originates from radionuclides with multiple energies. The emission occurs at different energies. Energy is detected at two, three, or more levels or windows. The energy is used for a selected range, whether from a continuous energy spectrum, from different main peaks, and / or from different minor peaks. In one embodiment, the emission is generated by two or more radiotracers, such as in the presence of impurities or by design. Each radiotracer induces emission at different energies, such as using Tc-99m MIBI and I-123 MIBG for cardiac imaging. Any combination of two or more radiotracers can be used for a given scan of a patient (i.e., simultaneously). In another embodiment, radionuclides with different emission energies are used. For example, I-123, Lu-177, or In-111 are used. Lu-177 emits at peak energies of 113 kV and 208 kV. Other peaks may not be included, or may be included within an energy range set around the peak used. In yet another embodiment, Y-90 is used. The energy spectrum of Y-90 is generally continuous rather than having specific peaks. A broad spectrum can be approximated as a quasi-emission line. Any two or more portions of the spectrum can be used for multi-energy reconstruction. In other methods, progeny of I-131, Y90, Sm153, Re186, W188 / Re188, Ho166, Lu177, Cu67, I125, Tl-201, Tb161, or Ac225, Ar211, Ra223, Tb149… are used as radionuclides with multiple energy emissions. Other isotopes or combinations of isotopes (radionuclides or combinations thereof) can be used.

[0034] Raw SPECT data or preprocessed data is provided for reconstruction. Reconstruction can use system matrices or projection operators to describe the properties of the SPECT imaging system, iteratively improving the data model representing the image objects in the SPECT data. The image objects can then be displayed using volumetric rendering or other imaging techniques.

[0035] An image object defined in object space is a reconstruction of SPECT data measured in data space. Object space is the space in which the result of the image reconstruction is defined and corresponds to, for example, the scanned 3D volume (i.e., the field of view or "FOV").

[0036] SPECT detectors or gamma cameras may be able to detect different energies. For example, the detector may have an operating energy range that includes multiple energy peaks of radionuclides. For instance, cadmium zinc telluride (CZT) detectors have sufficient thickness to detect within a range of energies. For example, CZT or other types of detectors can detect in the ranges of 40–1000, 40–3000, or 30–3000 keV. Other ranges may be possible.

[0037] In other methods, detector arrangements provide detection at different energies. There are broad categories of emission tomography that separate according to primary emission energy. Figure 2 Four examples are shown. Below or in the range of approximately 400 keV, 511 keV, or another level of keV, detector 200 operates with physical collimation from collimator 202 (such as a parallel-aperture collimator). "Approximately" is used for + / - 10%. Other values ​​are possible based on materials and existing technology. Detector 200 detects individual emissions that travel through collimator 202 in a given direction, rather than relying on any form of coincidence. Collimator 202 in front of detector 200 limits the direction of photons detected by SPECT detector 200, so each detected emission is associated with the energy and line or cone shape from the possible location where the emission occurred. For approximately 400 keV or 500 keV to approximately 600 keV or 700 keV, an object 204 with a known edge or pattern relative to detector 200 forms an coded aperture by moving object 204 relative to detector 200. For other methods in the range of approximately 500-800 keV, electronic collimation is used. Matching is provided, as in positron emission tomography. For example, detector ring 206 detects a pair of emissions that match each other. For approximately 800 keV, 900 keV, or higher, Compton scattering detection can be used. In one example, matching is provided by trap detector 210, which detects scattering from an event detected by scattering detector 208. Compton scattering detection can alternatively use electron tracking in a solid-state detector to perform both scattering and trap detection.

[0038] In other alternatives, a hybrid detector arrangement is used. For example, a multi-camera SPECT system is employed. One camera is configured to detect at one energy level (e.g., a lower energy (LE) window), while another camera is configured to detect at a different energy level (e.g., a higher energy window). For instance, one head or camera is optimized for, for example, LE SPECT, while another head or camera is optimized for medium-energy (ME) SPECT. Gamma from the higher peak becomes a problem of scattering-only correction, while the LE SPECT image is converted into spatial or structural information to improve the ME SPECT.

[0039] As another example, a physical collimator 202 is used for one or more lower energy windows, and a Compton scattering detector is used for one or more higher energy windows. Individual detectors 200, 208, and 210 can be used. Alternatively, scattering and / or trapping detectors 208 and 210 can be used in conjunction with the collimator to detect lower energy events. Inherent single-layer Compton imaging or multi-layer detectors can also detect lower energy events.

[0040] In other methods, a three-dimensional tileable gamma-ray detector, such as that disclosed in US 2022 / 0354443A, is used to detect emissions across multiple energy ranges. Sensor layouts for direct converter detectors, such as those disclosed in US 2022 / 0342091A1, can be used. Multimodal Compton and single-photon emission computed tomography medical imaging systems, such as those disclosed in US 2022 / 0330909A1, can be used to detect emissions in different energy windows.

[0041] In action 110, the image processor reconstructs the image object from the detected emission regions. Any reconstruction used in SPECT targeting different anatomical or spatial structures can be used. Examples include additional modal region reconstructions disclosed in U.S. Patent Nos. 8,577,103; 8,675,936; or 9,171,353. SPECT data (detected emissions) at one or more energies are used for segmentation instead of using CT or other “additional” (non-SPECT) data to identify regions. For example, lower-energy emissions may have greater spatial resolution, so reconstructions of detected SPECT emissions at lower-energy windows are used instead of CT to identify regions.

[0042] In one implementation, region reconstruction uses emission at one energy level (i.e., an energy window or range) to reconstruct image objects from emission at different energies. Region reconstruction uses detected emission at both energies; in this sense, it is multi-energy or multispectral reconstruction, but the emission at that energy level is not used for identifying regions in the projection or optimization. For example, detected emission at a 511 keV emission window is used to generate a region map of the tumor to improve Lu177 imaging with region reconstruction. The energy used for the region map, apart from the region itself, is not used for reconstructing the image objects. In cases where an initial reconstruction is provided for segmentation, a separate reconstruction is used.

[0043] In action 112, as in other methods, SPECT data at different energies are used together to match the resulting image object with SPECT data at different energies. Emissions at the energy level used to identify regions are also used in the reconstruction of the image object as part of the projection and optimization of the sub-regional reconstruction. Examples of multispectral or multi-energy reconstruction are found in U.S. Patents 10,126,439 (Reconstruction with Multiple Peaks in Quantitative SPECT) or 10,395,353 (Multimodal Multi-Energy SPECT Reconstruction). Sub-regional reconstruction is used in combination with multi-energy or spectral reconstruction, but in which the detected photon emissions are used to define regions instead of defining another mode, or to define regions in addition to defining another mode. The image processor uses multispectral sub-regional reconstruction to reconstruct the image object. Multispectral sub-regional reconstruction includes multiple energies and regions from one or more of these energies.

[0044] In action 114, for region reconstruction, the image processor determines regions from SPECT data at one energy level. Region maps of these regions are determined from emissions at lower energy levels, such as lower energy windows of radionuclides. The SPECT data at that energy level used to determine the regions spatially stabilizes the region reconstruction from SPECT data at one or more energies, such as higher energy windows. For example, lower energy photons of an isotope are used to create an image to “stabilize” higher energies of the same isotope. The low-energy peak is imaged, and the resulting image is segmented to identify regions. Higher-energy photon emissions are treated as a scattering correction problem for reconstruction. This reconstruction of the higher-energy gamma uses regions or segments from lower peak energies at higher resolution, thereby improving the spatial resolution of the higher-energy collimation (physical or electrical) of the higher-energy emissions.

[0045] These regions represent locations within the object being examined and are derived from SPECT data at one or more energies. Some energies may have higher spatial resolution than others, allowing for better spatial resolution to be used for region-based reconstruction at other energies. By incorporating region-based information into the reconstruction, region-based reconstruction can improve image quality and / or reduce acquisition time during the SPECT imaging process.

[0046] SPECT emission at one or more energies is reconstructed into an image object. This image object is then segmented using techniques such as thresholding, pattern matching, random walk, model fitting, and / or artificial intelligence. Segmentation separates regions or areas, which can then be used in subregional reconstruction. Each region is a three-dimensional (3D) area resembling an anatomical structure. For example, segmenting bone tissue from non-bone tissue. The reconstructed image object can be derived from a complete or final reconstructed image object, or it can be an image object generated during iterations during reconstruction.

[0047] In multi-energy regional reconstruction, when approximating SPECT images based on acquired SPECT data, the SPECT image of the examined object is reconstructed by considering the object's spatial or space-time structure. The object's structure allows it to be separated into multiple regions. Each organ or tissue type is assigned to a separate region. In reconstruction, volumes within each of those regions are treated separately and equally. This regional independence allows for different, rather than equal, treatment of different regions. Different amounts of signal are assigned to each region based on its contribution to the observed functional characteristics.

[0048] Region-based reconstruction can impose separations on the anatomical regions of the reconstructed image objects, but these regions do not modify the evaluation function of the applied reconstruction algorithm. Figure 3 An example is shown. Different regions 330 of region map 300 are forward projected 320 respectively based on estimates 310 and then renormalized as part of iterative reconstruction. The results of multimodal reconstruction can also improve resolution compared to reconstruction without functional information about region 330 (or even with attenuation correction).

[0049] Reconstruction involves using multiple energy windows. For example, one of the multi-energy models for reconstruction disclosed in U.S. Patent Application Publication No. 2017 / 0086757 is used. Because various image degradation effects (e.g., scattering, attenuation, and / or collimator-detector response functions) differ for different energy ranges, the forward projector 320 for the photon image formation process is modeled separately for each energy range. In one embodiment, scattering, attenuation, and collimator-response functions are modeled separately for each of the different emission energies, emission energy ranges, and / or acquisition energy windows. In action 116, multi-energy or multispectral regional reconstruction uses a scattering model to correct for scattering. Similarly, attenuation and / or collimator-detector response functions are modeled. A model treats scattering, attenuation, and / or collimator-response functions differently for different energies, thus providing a separate model for each energy window.

[0050] Any type of scattering model can be used. Model-based scattering estimates are provided by modeling the physical properties of scattering within the patient's body. Monte Carlo simulations or other simulations can be used. Other physical properties of scattering or different modeling types can be used. Scattering can be modeled differently for different energies. Photons with different energies can scatter differently.

[0051] Any type of attenuation model can be used. For example, the attenuation coefficient, as a function of the patient's three-dimensional location within the body, can be estimated based on anatomical information provided by computed tomography (CT). The measured attenuation coefficient can be used to model the attenuation as emitted photons travel through the patient's tissues. Different energies attenuate differently, which can be modeled as different attenuation coefficients for different energies or different scaling factors for different energies.

[0052] Any type of collimator-detector response function model can be used. In one embodiment, the point response function is measured for a specific collimator and detector, or for a class (i.e., the type of collimator-detector pair). Monte Carlo or other simulations can be used. The point response function varies as a function of energy levels. Other collimator-detector response functions can be used.

[0053] Image formation models for different energies are used for individual reconstructions at different energies. The resulting image object is then combined (332). Instead of a complete separation between reconstruction and post-reconstruction combination at each energy, combination for multiple peaks can be performed within or as part of the reconstruction. Reconstruction is performed iteratively, thus combining multiple peaks within the iterative loop of reconstruction, such as combining feedback (350) of the back projections of different peaks for updating the volume (360) based on comparison (340). Updates are obtained from the comparison (340) of the weighted summation (332) result with SPECT data. By combining feedback (350) from different peaks within the reconstruction, an image object is reconstructed for quantitative SPECT. Using photon counts from multiple peaks in a combined manner during reconstruction increases the signal-to-noise ratio and improves image quality and quantitative accuracy in SPECT imaging.

[0054] For each peak, an image volume is projected 320 and back-projected 350 using a peak-specific system matrix or projection operator (e.g., a projection operator that models attenuation correction, scattering correction, point response function, and / or sensitivity). The residuals, negative gradients (for the conjugate gradient method), or simulations generated by the back-projection 350 of multiple peaks are combined. The image volume is updated by adding the conjugate gradient generated by the combined negative gradient. For update 360, the conjugate gradient is multiplied by the optimal step size based on the combined negative gradient. This reconstruction scheme combines multiple peaks into a single image volume for quantitative SPECT. Alternatively, this combination is a post-reconstruction from image objects of different energies.

[0055] Reconstruction includes a forward projection 320 for region 330. For each region 330, the forward projection 320 is performed for different energies using a different image forming model. A portion of the image forming model includes scattering correction in action 116. Any scattering correction can be used, such as energy window-based scattering correction. In one embodiment, model-based scattering correction is used. Scattering correction is performed as part of the forward projection 320 from image or object space to data space. The scattering correction model is used to apply the image forming process to the activity distribution. The resulting projected data model has reduced scattering.

[0056] For model-based scattering correction, the scattering response function (SRF) is combined with the patient's activity distribution to form a model-based scattering source. The SRF is represented by a scattering kernel. A scattering kernel specific to a given SPECT system is used. The interaction of scattering from different sources with the detector and collimator is simulated. Monte Carlo or other stochastic simulations can be used. Simulations are performed for all systems of a given type, such as all SPECT systems using the same combination of collimator and detector. The simulation is specific to that combination, such as being at least partially based on the size, shape, and / or material properties of the collimator and detector. The simulation is not performed by the SPECT system, but by a computer, workstation, or server. Alternatively, the SPECT system performs the simulation. The result of the simulation is a scattering kernel for the collimator and detector combination. The scattering kernel models the general physical properties of the scattering process in image formation.

[0057] The simulation is for a given radioactive tracer. The simulation provides one or more sources to emit at energy levels of selected photon energies. The simulation provides scattering nuclei at different energy levels for a multi-energy image formation model.

[0058] The scattering nucleus can be fitted to a specific SPECT system using the measured sensitivity. Normalization across nuclei is performed using the measured sensitivity, fitting the scattering nucleus and the resulting SRF to a specific collimator-detector combination. Sensitivity is measured at the energy level of the primary photon. Different sensitivities are provided for different energy levels or energy windows. Normalization is performed by considering the sensitivity to primary photons emitted from radioactive tracers at different energies or to the non-scattering energy.

[0059] A model-based scattering source is created by convolving a scattering kernel with an active distribution. This model-based scattering source is used to model the scattering probes passing through the SPECT system. The model-based scattering source generated from the scattering kernel is forward-projected. This probe is modeled as forward radiative transfer to create a scattering model in the data space.

[0060] The forward projection of the activity distribution from multi-energy image formation is combined with a scattering model to reduce scattering in the resulting projected data model for each region. The image object or activity distribution for a given region (i.e., the region object) is convolved with a scattering kernel, and a forward projection is also performed using a multi-energy formation model.

[0061] The image object used for forward projection is at a given resolution. This resolution can be the resolution of SPECT data for one energy, increased by adding the resolution of SPECT data for another energy. In alternative embodiments, other resolutions are used. Different resolutions are used to model the scattering. The image object in this region is resampled, such as downsampled or upsampled. The scattering kernel is based on the energy resolution of the SPECT system and is adapted to the physical properties of different energy windows. Resampling matches the image object resolution to the energy and spatial resolution of the SPECT system, as represented by the scattering kernel. In multi-energy modeling, different resampling of the reconstructed image is provided for different models.

[0062] Resampling matches the resolution of the image object to the resolution of the SPECT system, rather than to the resolution of the forward projection of the image object used for region division. The region division object is used for forward projection at one resolution and for modeling scattering at one or more other resolutions. Resampling is provided for each of the energies used in multi-energy reconstruction. During image formation modeling, to best incorporate spatial information, the voxel size of the reconstructed image is set to be the same as the voxel size of the input region map. In the model of primary photon image formation, the voxel size of the reconstructed image remains unchanged. However, in the model of scattering photons, to facilitate the necessary convolution with the scattering kernel, the reconstructed image (e.g., the region division image object) is resampled such that the voxel size of the resampled image is the same as the voxel size of the scattering kernel.

[0063] Figure 4An example implementation of multi-energy region reconstruction is shown. This example may or may not include scattering correction of action 116. A loop of three operations 400, 410, and 420 for three energy windows is provided. The loop may include two, three, four, or more operations. The loop represents iterative reconstruction, in which photon emission at different (e.g., three) energies E1, E2, and E3 is reconstructed sequentially within each iteration. In iterative reconstruction, each iteration includes successive projection (forward and backward) operations 400, 410, and 420 from low to high energy. An image object from an operation at one energy (e.g., 400) can be used to form regions based on region-based forward projections in other operations (e.g., 410 and 420) at other energies in the loop. Each iteration of the loop has an objective function that includes photon emission at different energies. For example, the objective function is given by: , thus Indicates the use of image formation models The Mighell-corrected chi-square metric (for analytical simplification and assuming Poisson noise, an L2 norm might be used; however, any L1 or L2 norm appropriate for the corresponding problem could be used), incorporates a suitable image forming model as the system matrix. . Minimization yields a tomographic reconstruction for solving the corresponding inverse problem (i.e., the inverse of the generalized decaying Ladon transform). It has its system matrix. Specific Image Formation (IF) Data Model Allowing weights The joint estimation allows for customization based on the information content obtained. This can be done by designers or artificial intelligence (AI) methods. For example, at high energies, spatial resolution may be very poor, but statistics are high, while at low energies with better spatial resolution, statistics are poor. In this case, one would use the lower energy to define the region and the high energy to estimate the ingestion within the boundary. Cross terms can be further introduced to customize the minimization of efficiency without biasing the results.

[0064] For each iteration of the loop, the same operations 400, 410, and 420 are repeated. In other methods, the operations can be changed for different iterations. For example, after a given number of iterations or other change criteria are met, the region is not updated. The same region continues to be used in later iterations.

[0065] The iteration continues until one or more stopping criteria are met. Once complete, the image object from the last iteration is used for imaging.

[0066] exist Figure 1In action 120, the image processor generates an image from the image object. The reconstructed output is used for imaging. The reconstructed output represents the image object or volume from the patient in the last iteration. This final image object is used to generate the image.

[0067] An image object is a three-dimensional representation of the patient's detected emission. The image object is rendered or otherwise used to generate an image. For example, a multi-planar reconstruction or single-slice image of a plane is generated. The intersection of one or more planes with the image object is visualized. As another example, surface or projection rendering is performed for three-dimensional imaging. Other imaging methods can be used.

[0068] An image was generated. Alternatively, a sequence of images was generated. For example, image objects from different time periods were used to generate an image sequence representing the patient over time.

[0069] Images derived from regional reconstruction of functional information are displayed separately. Alternatively, anatomical images are displayed alongside functional images. For example, functional images are overlaid on CT images. This overlay can be colored for display on grayscale CT images. Other combinations may be used.

[0070] For quantitative SPECT, images can be alphanumeric text representing specific uptake values ​​for a location. Graphical, graphical, or other representations of uptake at multiple locations can be output. Spatial images showing the distribution of uptake can use color or brightness modulation to indicate the level of uptake at a location.

[0071] Image objects for imaging are reconstructed from emissions at one or more energies. For example, image objects are reconstructed from emissions at one energy window, region by region, based on regions determined by reconstructions from emissions at different energy windows. As another example, multiple image objects are reconstructed from different energy windows, region by region. The image objects used for imaging are combinations (e.g., averages or weighted averages) of image objects from different energies (i.e., post-reconstruction combination). In yet another example, image objects are reconstructed from emissions detected at different energy windows, region by region. Combination is within the reconstruction. Emissions at different energies are combined in the reconstruction, such as using a shared objective function.

[0072] Figure 5 An embodiment of a medical imaging system 500 for intramodal regional reconstruction of SPECT is shown. System 500 can achieve multi-energy regional reconstruction, where the regions are intramodal (i.e., from SPECT emission). Figure 1 , Figure 3 and / or Figure 4 One method or another.

[0073] System 500 is a SPECT imaging system or scanner and includes a detector arrangement 510, a reconstruction processor 520, a memory 530, and a display 540. Additional, different, or fewer components may be provided. For example, a PET or Compton imaging system may be provided instead of a SPECT imaging system. In one embodiment, the reconstruction processor 520, memory 530, and / or display 540 are part of the SPECT imaging system. In an alternative embodiment, the reconstruction processor 520, memory 530, and / or display 540 are provided as a workstation, server, or computer separate from the detector arrangement 510. The memory 530 is part of a computer or workstation having the reconstruction processor 520, or a remote database such as a Picture Archiving and Communication System (PACS).

[0074] Detector arrangement 510 includes one or more detectors for detecting radiation emitted from within the patient's body. For SPECT, a gamma camera is used for detection. Detectors detect photon emission. Photons are emitted from tracers or radiopharmaceuticals. Detectors detect photons. A given detector can detect a series of events from the same or different locations within the patient.

[0075] Tracers consist of radionuclides with complex energy spectra. They provide multiple energy peaks or substantially continuous energy regions. Combinations of radionuclides can be provided to generate emissions at different energies. Radionuclides emit energy at or near different energy peaks or within continuous energy regions.

[0076] Detector arrangement 510 includes one or more detectors for detection in two or more different energy windows. It can be used... Figure 2 The detector arrangement may be either physical or other. For example, one detector may detect emissions at different energy windows, such as using a CZT detector. As another example, a combination of detectors for physical collimation and Compton scattering is provided. In yet another example, a multi-camera system has one detector (e.g., an LE detector) that is physically collimated for a relatively low energy range, and another detector (e.g., an ME detector) that is electrically or physically collimated for detection at a relatively high energy range.

[0077] The reconstruction processor 520 is a general-purpose processor, central processing unit, control processor, graphics processor, digital signal processor, application-specific integrated circuit, field-programmable gate array, artificial intelligence processor, digital circuit, analog circuit, timing circuit, combination thereof, or other device now known or hereafter developed for reconstructing patient volume from detected emissions. The reconstruction processor 520 is a single device or multiple devices operating serially, in parallel, or individually. The reconstruction processor 520 is specifically designed or provided for reconstruction, but may be the main processor or general-purpose processor of a computer (such as a laptop or desktop computer), or may be a processor used to handle tasks in a larger system. The reconstruction processor 520 can perform functions other than regional reconstruction.

[0078] The reconstruction processor 520 is configurable. The reconstruction processor 520 is configured by software, firmware, and / or hardware. Different software, firmware, and / or instructions are loaded or stored in memory 530 for configuring the reconstruction processor 520.

[0079] The reconstruction processor 520 is configured to reconstruct an object representing a patient from detected emissions. Reconstruction can be performed against at least two energy windows of the distribution energy of the radionuclide. The reconstruction also includes sub-regional reconstruction, where the regions are intramodal (i.e., from the detected emissions). Detected emissions at one or more energies are reconstructed or projected onto the image space for spatial segmentation. For example, emissions at lower energy windows are used to form an image object or patient representation for segmentation of regions. These regions are then used for sub-regional reconstruction. Regions from lower energy emissions are used to perform sub-regional reconstruction of emissions from one or more higher energy windows. In another embodiment, the reconstruction is multispectral or multi-energy, thus sub-regionally reconstructing the image object from emissions from two or more energy windows (including or excluding the energy window used to identify regions).

[0080] In one embodiment using a scattering model, the reconstruction processor 520 is configured to forward project a region of the image using a multi-energy projector at a first resolution, and to model the scattering of the region of the image at a second resolution different from the first resolution. For the primary multi-energy projection and scattering model, the region of the image is resampled (i.e., sampled differently). For example, the resolution of the region of the image is the resolution of the reconstructed image object from the lower-energy emission, and the resolution used to model the scattering is the system or scattering kernel resolution.

[0081] The reconstruction processor 520 can be configured to modify the reconstruction process. Reconstruction is iterative. Different reconstruction procedures can be used for later iterations compared to earlier iterations.

[0082] Memory 530 is a random access memory, graphics processing memory, video random access memory, system memory, cache memory, hard disk drive, optical media, magnetic media, flash drive, buffer, database, combination thereof, or other memory device now known or hereafter developed for storing data. Memory 530 stores detected emissions (e.g., event data detected by PET, Compton, or SPECT), region information, segmentation information, energy information, and / or reconstruction information. Memory 530 stores data that has been processed, such as updated image objects, segmented image objects, renormalization coefficients, scattering kernels, projection operators or system matrices, segmented data models, combined data models, region functions, resampled image objects, and / or other information.

[0083] Memory 530 or other memory is a non-transitory computer-readable storage medium storing data representing instructions executable by a programmable reconstruction processor 520 for SPECT reconstruction. Instructions for implementing the processes, methods, and / or techniques discussed herein are provided on a computer-readable storage medium or memory, such as a cache, buffer, RAM, removable media, hard disk drive, or other computer-readable storage medium. Computer-readable storage media include various types of volatile and non-volatile storage media. Functions, actions, or tasks illustrated in the figures or described herein are performed in response to one or more sets of instructions stored in or on a computer-readable storage medium. These functions, actions, or tasks are independent of a particular type of instruction set, storage medium, processor, or processing strategy, and can be performed independently or in combination by software, hardware, integrated circuits, firmware, microcode, and the like. Similarly, processing strategies can include multiprocessing, multitasking, parallel processing, and the like.

[0084] In one embodiment, the instructions are stored on a removable media device for reading by a local or remote system. In other embodiments, the instructions are stored at a remote location for transmission over a computer network or telephone line. In still other embodiments, the instructions are stored within a given computer, CPU, GPU, or system.

[0085] Display 540 is a monitor, LCD, plasma display, touchscreen, printer, or other device used to display images for user viewing. Display 540 shows one or more images representing function, such as uptake or activity concentration. The image is a quantitative or qualitative SPECT image of the reconstructed object. The image can be a volumetric rendering, multiplanar reconstruction, cross-section, and / or another image from the final image of the object. The image represents the distribution of radionuclides within the patient's body based on emissions detected from the SPECT system 500.

[0086] While the invention has been described above with reference to various embodiments, it should be understood that many changes and modifications can be made without departing from the scope of the invention. Therefore, it is intended that the foregoing detailed description be considered illustrative rather than restrictive, and that the following claims, including all equivalents, are intended to define the spirit and scope of the invention.

Claims

1. A method for reconstruction using single-photon emission computed tomography (SPECT), the method comprising: Detect SPECT emission from the patient, the SPECT emission being at a first energy and a second energy; The image object is reconstructed by segmenting it from the detected SPECT emission, and the segmented reconstruction area for the second energy is determined from the SPECT emission at the first energy. as well as Generate an image from an image object.

2. The method of claim 1, wherein detection comprises detecting SPECT emission from a radionuclide, wherein the radionuclide comprises a progeny of Lu-177, I-131, Tl-201, Tb161, or Ac225.

3. The method of claim 1, wherein the detection includes detection using a multi-camera SPECT system, wherein a first camera of the multi-camera SPECT system is configured to detect at a lower energy than a second camera of the multi-camera SPECT system.

4. The method of claim 1, wherein the detection comprises detecting SPECT emission at a first energy using physical collimation and detecting SPECT emission at a second energy using Compton scattering.

5. The method of claim 1, wherein the regional reconstruction comprises generating a region map comprising a region from a first reconstruction of the SPECT emission at a first energy, forward projecting the SPECT emission at a second energy for each region, performing a weighted summation on the forward projection, obtaining an update from a comparison of the weighted summation result with the SPECT emission, backward projection updating, and updating the estimate from the update of the backward projection.

6. The method of claim 1, wherein regional reconstruction comprises generating a region map of the region from SPECT emission at a first energy, the first energy being lower than a second energy.

7. The method of claim 6, wherein the SPECT emission at the first energy spatially stabilizes the regional reconstruction from the SPECT emission at the second energy, the second energy being an energy higher than the first energy.

8. The method of claim 1, wherein the region reconstruction includes modeling the scattering of a second energy, which is higher than the first energy, wherein the SPECT emission at the second energy is segmented according to a region determined from the first energy, the region being determined with a higher spatial resolution than the image object of the SPECT emission at the second energy.

9. The method of claim 1, further comprising an initial reconstruction of the SPECT emission at a first energy, and then segmenting the region from the initial reconstruction for a sub-regional reconstruction of the SPECT emission at a second energy.

10. The method of claim 1, wherein the regional reconstruction includes iterative reconstruction, wherein in each iteration, the SPECT emission at a first energy and the SPECT emission at a second energy are reconstructed sequentially.

11. The method of claim 10, wherein iterative reconstruction includes a cyclic sorting from a first energy to a second energy in each iteration, the first energy being lower than the second energy, each iteration having an objective function, the objective function including both SPECT emission at the first energy and SPECT emission at the second energy.

12. The method of claim 1, wherein generating the image comprises generating the image from an image object emitted from a SPECT at a second energy.

13. The method of claim 1, wherein generating an image comprises combining an image object reconstructed by region based on SPECT emission at a detected second energy with another image object reconstructed from SPECT emission at a detected first energy to generate an image.

14. The method of claim 1, wherein generating an image comprises generating an image object reconstructed regionally from SPECT emission at a second energy in combination with reconstruction from SPECT emission at a first energy.

15. A method for reconstruction using single-photon emission computed tomography (SPECT), the method comprising: Detecting SPECT emissions from the patient, the SPECT emissions originating from radioactive nuclides with multiple energies; Image objects are reconstructed using multispectral region reconstruction, wherein the multispectral region reconstruction includes the multiple energies and a region from a first energy among the multiple energies; as well as Generate an image from an image object.

16. The method of claim 15, wherein reconstructing the image object comprises iterative reconstruction, wherein each iteration comprises a successive projection operation from low energy to high energy.

17. The method of claim 15, wherein the reconstruction includes using a region from a first energy, the first energy being an energy lower than a second energy, wherein the multispectral sub-regional reconstruction uses the region from the first energy to reconstruct the second energy.

18. A medical imaging system for intramodal region reconstruction in single-photon emission computed tomography (SPECT), the medical imaging system comprising: The detector array is configured to detect emissions from the patient and emissions of different energies from the radioactive tracer; A processor configured to reconstruct an object representing a patient from emissions detected from at least two energy windows of a distributed energy source, wherein the reconstruction of the lower energy window in the at least two energy windows is segmented, and wherein the reconstruction of the higher energy window in the at least two energy windows is regioned based on the segmentation from the reconstruction of the lower energy window in the at least two energy windows; as well as A display, configured to show an image of the reconstructed object.

19. The medical imaging system of claim 18, wherein the detector arrangement includes detectors configured to detect using the at least two energy windows.

20. The medical imaging system of claim 18, wherein the detector arrangement comprises a physical collimation detector and a Compton scattering detector.

Images