Nucleus imaging-based voxel-level radionuclide dose estimation method and computer device

Abstract

The application provides a voxel-level radionuclide dose estimation method based on radionuclide imaging and a computer device. The method comprises the following steps: acquiring X-ray structural images and gamma-ray functional images at a single time point or multiple time points; segmenting a region of interest to obtain multiple sub-regions of interest; performing regional partial volume correction on the multiple sub-regions of interest of the gamma-ray functional images to obtain corrected gamma-ray functional images; performing registration on the corrected gamma-ray functional images; performing activity-dose rate conversion on the registration results to obtain a dose rate, and performing voxel-level curve fitting and integration on the dose rate to obtain an absorbed dose distribution result, and determining the absorbed dose of each sub-region of interest based on the absorbed dose result. The application can reduce the influence of the partial volume effect on the resolution of radionuclide imaging, improve the accuracy of radionuclide image registration, and further improve the accuracy of radionuclide absorbed dose estimation.

Description

Technical Field

[0001] This application relates to the field of radionuclide dose estimation, and more specifically, to a voxel-level radionuclide dose estimation method and computer equipment based on radionuclide imaging. Background Technology

[0002] Targeted radionuclide therapy (TRT) is a treatment for various advanced metastatic lesions of cancer. TRT selectively delivers a dose of radionuclide to tumor cells, allowing the radionuclide to bind to tumor-targeting molecules and achieve tumor treatment. However, to ensure the safety and effectiveness of tumor treatment, medical staff need to accurately grasp the distribution of radionuclide in various organs of the patient and the absorbed dose of radionuclide at the tumor lesion site.

[0003] In related technologies, it is common to acquire single-photon emission computed tomography (SPECT) images and computed tomography (CT) images at multiple time points, register the SPECT and CT images at the single time point, perform activity-dose rate conversion based on the registration results, and obtain the radioactive dose distribution map by voxel-level curve fitting and integration.

[0004] However, when estimating radionuclide doses using related technologies, some volume effects cause blurring of SPECT image boundaries, leading to registration errors between SPECT and CT images. Therefore, these technologies suffer from poor image registration accuracy and inaccurate radionuclide absorption dose estimation results. Summary of the Invention

[0005] The purpose of this application is to provide a voxel-level nuclide dose estimation method and computer equipment based on nuclide imaging, which can reduce the impact of some volume effects on the resolution of nuclide imaging, thereby improving the accuracy of nuclide image registration and thus improving the accuracy of radionuclide absorbed dose estimation.

[0006] The embodiments of this application are implemented as follows:

[0007] A first aspect of this application provides a voxel-level radionuclide dose estimation method based on radionuclide imaging, the method comprising:

[0008] Acquire X-ray structural images and gamma-ray functional images at a single or multiple time points;

[0009] The region of interest is segmented to obtain multiple sub-regions of interest;

[0010] Partial volume correction is performed on the gamma-ray functional images of each sub-region of interest to obtain the corrected gamma-ray functional images;

[0011] Register corrected gamma-ray functional images and X-ray structural images at a single or multiple time points;

[0012] The registration results are converted from activity to dose rate to obtain the dose rate corresponding to a single time point or multiple time points;

[0013] Curve fitting and integration are performed on the dose rates corresponding to a single time point or multiple time points to obtain the dose distribution results of the radionuclides;

[0014] Based on the dose distribution results, the dose of radionuclide absorbed by each sub-region of interest is determined.

[0015] As one possible implementation, partial volume correction is performed on the gamma-ray functional images of each sub-region of interest to obtain corrected gamma-ray functional images, including:

[0016] Based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest, the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest is determined, and the gamma-ray functional image is iteratively partially volume corrected based on the point spread function corresponding to each voxel to obtain the iterative gamma-ray functional image.

[0017] An overall primality consistency constraint is applied to the iterated gamma-ray functional image to obtain a constrained gamma-ray functional image, so that the overall primality value of the constrained gamma-ray functional image is consistent with the overall primality value of the gamma-ray functional image before iteration, and the constrained gamma-ray functional image is used as the corrected gamma-ray functional image.

[0018] As one possible implementation, based on the three-dimensional coordinates of each voxel in the gamma-ray functional image of each sub-region of interest, the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest is determined. Then, iterative partial volume correction is performed on the gamma-ray functional image based on the point spread function corresponding to each voxel to obtain the iterative gamma-ray functional image, including:

[0019] The average value of the iterative mask for each sub-region of interest is determined based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest and the number of voxel points contained in the gamma-ray functional image of each sub-region of interest.

[0020] Based on the average of the iterative masks of each sub-region of interest, determine the three-dimensional coordinates of the iteratively synthesized reference image corresponding to each sub-region of interest;

[0021] Based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the iteratively synthesized reference image, iterative partial volume correction is performed on the gamma-ray functional image of each sub-region of interest to obtain the iterative gamma-ray functional image.

[0022] As one possible implementation, the gamma-ray functional images of each sub-region of interest are iteratively partially volume-corrected based on the point spread function corresponding to each voxel contained in the gamma-ray functional image of each sub-region of interest and the iteratively synthesized reference image, to obtain the iteratively synthesized gamma-ray functional image, including:

[0023] Based on formula Calculate the gamma-ray functional image after iteration;

[0024] Among them, I k+1 (x,y,z) represents the gamma-ray functional image obtained after the (k+1)th iteration, I k (x,y,z) represents the gamma-ray functional image obtained after the k-th iteration, s k (x,y,z) represents the iteratively synthesized reference image obtained in the k-th iteration, and PSF(x,y,z) represents the point spread function corresponding to voxel (x,y,z) in the gamma-ray functional image of the sub-region of interest.

[0025] As one possible implementation, a global element consistency constraint is applied to the iterative gamma-ray functional image to obtain a constrained gamma-ray functional image, including:

[0026] Based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the initial mask mean of each sub-region of interest, determine the grayscale adjustment area corresponding to the gamma-ray functional image of each sub-region of interest.

[0027] The overall morphological consistency constraint is applied to the gamma-ray functional image after iteration based on the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest, so as to obtain the constrained gamma-ray functional image.

[0028] As one possible implementation, based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the initial mask mean of each sub-region of interest, the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest is determined, including:

[0029] Based on formula Calculate the grayscale adjustment area;

[0030] Where m(x,y,z) represents the grayscale adjustment region, PSF(x,y,z) represents the point spread function corresponding to voxel (x,y,z) in the gamma-ray functional image of the sub-region of interest, and s 0 (x,y,z) represents the initial mask mean of the sub-region of interest.

[0031] As one possible implementation, the overall morphological consistency constraint is applied to the iterated gamma-ray functional image based on the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest, to obtain the constrained gamma-ray functional image, including:

[0032] Based on formula The constrained gamma-ray functional image was calculated.

[0033] Among them, I_CP k+1 (x,y,z) represents the constrained gamma-ray functional image after the (k+1)th iteration, I k+1 (x,y,z) represents the gamma-ray functional image obtained after the (k+1)th iteration, I k (x,y,z) represents the gamma-ray functional image obtained after the k-th iteration, and m(x,y,z) represents the grayscale adjustment area.

[0034] As one possible approach, curve fitting and integration are performed on the dose rates at multiple time points to obtain the dose distribution results of the radionuclide, including:

[0035] Based on the voxel values ​​of each voxel in the gamma-ray functional image after multi-time point registration, the voxel mean value of each voxel is determined. The voxel mean value refers to the voxel mean value between each voxel and its adjacent voxels.

[0036] Based on the average voxel value corresponding to each voxel, iteratively update the voxel value corresponding to each voxel.

[0037] A new voxel matrix is ​​generated based on the updated voxel values ​​corresponding to each voxel.

[0038] The new voxel matrix was fitted and integrated using a volume-by-volume curve fitting method to obtain the dose distribution results of radionuclides at multiple time points.

[0039] In a second aspect of this application, a computer device is provided, comprising: a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of the voxel-level radionuclide dose estimation method based on radionuclide imaging described in the first aspect.

[0040] A third aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the voxel-level nuclide dose estimation method based on nuclide imaging described in the first aspect.

[0041] The beneficial effects of the embodiments of this application include:

[0042] This application provides a voxel-level radionuclide dose estimation method based on radionuclide imaging. By acquiring X-ray structural images and gamma-ray functional images at single or multiple time points after tumor treatment, a user-selected region of interest (ROI) is segmented to obtain multiple sub-ROIs. Partial volume correction is applied to the gamma-ray functional images of each sub-ROI to obtain corrected gamma-ray functional images. A model is used to register the corrected gamma-ray functional images and X-ray structural images at single or multiple time points to obtain registration results. The registration results are then converted from activity to dose rate to obtain a dose distribution map of the radionuclide in the patient's body. Based on the dose distribution map, the absorbed radionuclide dose in each sub-ROI is determined. Partial volume effects can cause blurred boundaries in the gamma-ray functional images, leading to errors in activity measurement within each sub-ROI. This application performs partial volume correction on the gamma-ray functional images of each sub-ROI to ensure clear boundaries in the corrected gamma-ray functional images, thereby improving the registration accuracy between the gamma-ray functional images and the X-ray structural images. In this way, the impact of some volume effects on the resolution of radionuclide imaging can be reduced, thereby improving the accuracy of radionuclide image registration and thus improving the accuracy of radionuclide absorbed dose estimation. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating the first voxel-level radionuclide dose estimation method based on radionuclide imaging provided in this application embodiment;

[0045] Figure 2 A flowchart of a partial volume calibration method based on a variable point diffusion function is provided for embodiments of this application;

[0046] Figure 3 A schematic diagram illustrating the reconstruction result of a radionuclide point source in the coronal plane, provided as an embodiment of this application;

[0047] Figure 4 A schematic diagram of the three-dimensional full width at half maximum (FWHM) distribution of a series of radioactive nuclide point sources provided in an embodiment of this application;

[0048] Figure 5 A schematic diagram of the reconstruction result of a radionuclide point source in a cross section provided for an embodiment of this application;

[0049] Figure 6 A schematic diagram of curve fitting results between the axial half-height full width and the distance to the point source collimator provided in an embodiment of this application;

[0050] Figure 7 A schematic diagram of the curve fitting result between the radial half-height full width and the distance to the point source collimator provided for an embodiment of this application;

[0051] Figure 8 A schematic diagram of curve fitting results between the full width at half maximum (FWHM) of the circumference and the distance to the point source collimator, provided for an embodiment of this application;

[0052] Figure 9 A flowchart illustrating a second method for estimating voxel-level radionuclide dose based on radionuclide imaging, provided in an embodiment of this application;

[0053] Figure 10 A flowchart illustrating the third voxel-level radionuclide dose estimation method based on radionuclide imaging provided in this application embodiment;

[0054] Figure 11 The functional image of gamma rays obtained from the simulation experiment provided in the embodiments of this application;

[0055] Figure 12 A schematic diagram of partial volume correction results of gamma-ray functional images obtained from simulation experiments provided in this application embodiment;

[0056] Figure 13 A histogram showing the average correction error of different volume correction methods provided in this application embodiment;

[0057] Figure 14 A flowchart illustrating the fourth voxel-level radionuclide dose estimation method based on radionuclide imaging provided in this application embodiment;

[0058] Figure 15 This is a schematic diagram of a volume-by-volume curve fitting method provided in an embodiment of this application;

[0059] Figure 16 This is a schematic diagram of the fitting results for an existing voxel block.

[0060] Figure 17 This is a schematic diagram of a volume-by-volume curve fitting method provided in an embodiment of this application;

[0061] Figure 18 Flowchart of the registration method provided in the embodiments of this application;

[0062] Figure 19 A schematic diagram of nuclide linearity before registration at a single time point, provided as an embodiment of this application;

[0063] Figure 20 A schematic diagram of a single time point registration result provided in an embodiment of this application;

[0064] Figure 21 A schematic diagram of nuclide linearity before multi-time-point registration is provided for an embodiment of this application;

[0065] Figure 22 A schematic diagram of multi-time-point registration results provided in an embodiment of this application;

[0066] Figure 23 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0068] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0069] Targeted radionuclide therapy, as a treatment for various advanced metastatic lesions of cancer, involves delivering radionuclides directly to tumor cells, allowing the radionuclides to bind to tumor-targeting molecules and achieve therapeutic goals. Among these, the radionuclides... 177 Lu can produce β due to decay. - With its moderate particle size and penetrating power, radioactive nuclides are highly favored by physicians in the field of oncology treatment. Meanwhile, radioactive nuclides... 177 The decay of Lu produces gamma rays (γ rays) of appropriate energy, which can be used for single-photon computed tomography (SPECT) imaging to achieve tumor treatment.

[0070] Currently, the common practice is to acquire gamma-ray functional images and X-ray structural images at single or multiple time points after tumor treatment, register these images, perform activity-dose rate conversion on the registration results, and then perform voxel-level curve fitting and integration on the conversion results to obtain a dose distribution map. Based on this dose distribution map, the dose of the absorbed radionuclide in the region of interest is determined. However, this approach neglects the influence of partial volume effects on gamma-ray functional images. Partial volume effects can cause blurring of the boundaries in gamma-ray functional images, leading to registration errors between the gamma-ray functional images and X-ray structural images, and consequently, inaccurate estimations of the absorbed dose of radionuclides.

[0071] To address this, this application provides a voxel-level nuclide dose estimation method based on radionuclide imaging. This method involves acquiring X-ray structural images and gamma-ray functional images at single or multiple time points after tumor treatment; segmenting the region of interest (ROI) to obtain multiple sub-ROIs; performing partial volume correction on the gamma-ray functional images of each sub-ROI to obtain corrected gamma-ray functional images corresponding to each sub-ROI; registering the corrected gamma-ray functional images and X-ray structural images at single or multiple time points, and performing an activity-dose rate conversion on the registration results; performing curve fitting and integration on the activity-dose rate conversion results to obtain dose distribution results; and determining the absorbed radionuclide dose in each sub-ROI based on the dose distribution results. This approach reduces the impact of partial volume effects on the resolution of radionuclide imaging, thereby improving the accuracy of nuclide image registration and ultimately enhancing the accuracy of radionuclide absorbed dose estimation.

[0072] The following description, in conjunction with the accompanying drawings, provides a detailed explanation of the voxel-level radionuclide dose estimation method based on radionuclide imaging provided in the embodiments of this application.

[0073] Figure 1 A flowchart illustrating a voxel-level radionuclide dose estimation method based on radionuclide imaging, provided in this application, is shown. This method can be applied to computer equipment. See also... Figure 1 This application provides a method for estimating voxel-level radionuclide dose based on radionuclide imaging, including:

[0074] S101. Acquire X-ray structural images and gamma-ray functional images at a single or multiple time points.

[0075] Optionally, a single time point refers to a static time point, while multiple time points refer to a dynamic, continuous time series. Gamma-ray functional imaging and X-ray structural imaging both belong to radionuclide imaging after tumor treatment. Gamma-ray functional imaging is used to display functional and metabolic information of various parts of the patient's body, while X-ray structural imaging is used to display anatomical information of various parts of the patient's body. The radiation source for gamma-ray functional imaging comes from radiopharmaceuticals within the patient's body, while the radiation source for X-ray structural imaging comes from an X-ray generator outside the patient's body. However, gamma-ray functional imaging has lower resolution and higher noise levels. Affected by collimator color marks and scattering effects, gamma-ray functional imaging is relatively blurry and cannot clearly display the structural morphology of various organs and skeletal structures within the patient's body. X-ray structural imaging, on the other hand, has higher resolution and can clearly display the morphological details of organs, structures, and blood vessels within the patient's body.

[0076] Optionally, X-ray structural imaging is a method of scanning a patient's body from multiple angles using X-rays and then reconstructing cross-sectional anatomical images using a computer. X-ray structural imaging is equivalent to CT images. By detecting differences in tissue absorption of X-rays, X-ray structural imaging reflects the morphological details of organs, structures, and blood vessels. Gamma-ray functional imaging is a single-photon emission computed tomography (SPECT) image that generates three-dimensional images by detecting gamma rays released by radioactive tracers in the patient's body. Gamma-ray functional imaging is equivalent to SPECT images. Gamma-ray functional imaging focuses on the physiological changes in various parts of the patient's body and is used to assess the function and metabolic status of various organs or tissues in the patient's body.

[0077] Optionally, single-time-point CT and SPECT images are instantaneous radionuclide images captured at a specific moment. While single-time-point gamma-ray functional images and X-ray structural images require short scan times, they cannot capture the dynamic physiological processes of various organs and tissues within the patient's body. Multi-time-point CT and SPECT images, on the other hand, are time-series radionuclide images obtained through multiple consecutive imaging sessions at different time intervals. Based on multi-time-point CT and SPECT images, dynamic monitoring of the dynamic physiological processes of various organs and tissues within the patient's body is possible; however, multi-time-point CT and SPECT image scanning requires more time. Single-time-point CT and SPECT images are suitable for scenarios such as fractures, cerebral hemorrhage, and static thyroid imaging, while multi-time-point CT and SPECT images are suitable for tracking tumor metabolic changes, verifying drug targeting, and assessing venous blood flow. Physicians can choose the radionuclide imaging acquisition method based on the patient's actual condition.

[0078] S102. Divide the region of interest into multiple sub-regions of interest.

[0079] Optionally, the region of interest refers to a specific area that the user wants to learn about. The region of interest can be an organ or tissue such as the spleen, stomach, kidneys, heart, or veins. The region of interest can be divided into multiple sub-regions of interest.

[0080] Optionally, image processing tools are used to perform mask calibration on each sub-region of interest to obtain multiple mask regions. A mask region refers to the set of voxels in the image defined by the mask that are selected or excluded. Mask regions can precisely limit the processing, analysis, and display range of gamma-ray functional images to shield background gamma-ray functional images unrelated to organs and tissues. Furthermore, one mask region corresponds to only one gamma-ray functional image at any given time. It is worth noting that mask regions can be automatically calibrated based on the voxel intensity range of the X-ray structural image, or manually calibrated by the user based on the structural morphology displayed in the X-ray structural image; this application does not impose specific limitations on either approach.

[0081] S103. Perform partial volume correction on the gamma-ray functional images of each sub-region of interest to obtain the corrected gamma-ray functional images.

[0082] Optionally, partial volume effect refers to the phenomenon where, due to the limited spatial resolution of the imaging system or the large size of the three-dimensional voxels, signals from different tissues or structures are mixed within a single voxel, causing the measured values ​​of the gamma-ray functional image to deviate from the true values ​​of the gamma-ray functional image. For example, if the size of the lesion in the patient's body is smaller than the full width at half maximum (FWHM) of the point spread function of the imaging system, the gamma-ray functional image will be diluted by the surrounding tissue, resulting in blurred boundaries of the gamma-ray functional image.

[0083] Optionally, partial volume correction can be performed on the gamma-ray functional images of each sub-region of interest to make the edges of the gamma-ray functional images corresponding to each sub-region of interest clear, thereby improving the efficiency and accuracy of SPECT image registration with CT images.

[0084] Optionally, the three-dimensional spatial parameters of the corrected gamma-ray functional image are close to those of the real gamma-ray functional image. The corrected gamma-ray functional image is obtained by partially correcting the volume of the original gamma-ray functional image. The voxels of the corrected gamma-ray functional image are consistent with those of the original gamma-ray functional image to ensure that the resolution of the gamma-ray functional image remains unchanged. That is, the dynamic physiological characteristics of the mask region represented by the gamma-ray functional images before and after correction are consistent. In particular, the corrected gamma-ray functional image can effectively avoid misdiagnosis of lesions and can display lesion areas more clearly.

[0085] It is worth noting that the partial volume effect can cause high-activity areas to diffuse into surrounding low-activity areas, leading to spillover and infiltration phenomena. This results in the underestimation of the radioactivity in high-activity areas and the overestimation of the radioactivity in low-activity areas.

[0086] S104. Register the corrected gamma-ray functional image and X-ray structural image at a single or multiple time points.

[0087] Optionally, the registration result refers to the spatial and temporal alignment of the corrected gamma-ray functional image and the X-ray structural image. The registration result can accurately locate the structural position and dynamic physiological characteristics of various organs and tissues in the patient's body.

[0088] Optionally, for the registration of corrected gamma-ray functional images and X-ray structural images at a single time point, normalized mutual information is often used as an optimization function to maximize the normalized mutual information between the rigid transformation corrected gamma-ray functional images and X-ray structural images. Employing a multi-resolution registration strategy can enhance the registration accuracy, speed, and robustness of the corrected gamma-ray functional images and X-ray structural images.

[0089] Optionally, for the registration of corrected gamma-ray functional images and X-ray structural images at multiple time points, the X-ray structural images are usually registered first, and the registered X-ray structural images are applied to the gamma-ray functional images at the corresponding time points to obtain registered gamma-ray functional images at multiple time points.

[0090] Optionally, see Figure 18 This application employs unsupervised learning to train a deep Laplacian pyramid network for differential homeomorphic registration of X-ray structural images. This network utilizes three identical convolutional networks to simulate registration across multiple resolution modes. The parameters of the deep learning model are optimized using a stable velocity field within a log-Euclidean framework, and the model is further optimized within the differential homeomorphic mapping. Specifically, the registration result is input into the unsupervised learning training of the deep Laplacian pyramid network to obtain the corresponding velocity field. The velocity field is then integrated using the SVF algorithm to obtain the displacement field. Finally, a spatial transformation layer transforms the corrected gamma-ray functional image. A weighted sum of locally normalized cross-correlation and the diffusion regularization constraint of the velocity field is used as the loss function. Stochastic gradient descent is employed to optimize the network parameters on the training dataset by minimizing the expected loss function.

[0091] S105. Perform activity-dose rate conversion on the registration results to obtain the dose rate corresponding to a single time point or multiple time points.

[0092] Optionally, based on the registration results, a three-dimensional activity distribution map of the registered gamma-ray functional image is determined, and the cumulative activity of the source organ to which the gamma-ray functional image belongs is extracted from the three-dimensional activity distribution map; the absorbed dose S of a unit cumulative activity in the source organ to the target organ is obtained by looking up a table. Here, the target organ refers to the organ whose effects of radiopharmaceutical radiation need to be assessed.

[0093] Optionally, the activity-dose rate conversion of the registration results can be achieved through models such as MIRD, dose point kernel, and Monte Carlo to obtain the dose rate corresponding to a single time point and the dose rate corresponding to multiple time points.

[0094] S106. Perform curve fitting and integration on the dose rate corresponding to a single time point or multiple time points to obtain the dose distribution results of the radionuclide.

[0095] Optionally, for the dose rate corresponding to a single time point, a single exponential function is used to perform curve fitting and integration on a voxel-by-voxel basis of the corrected gamma-ray functional image to obtain the dose distribution results of the radionuclide; for the dose rate corresponding to multiple time points, a double exponential function is used to perform curve fitting and integration on the corrected gamma-ray functional image to obtain the dose distribution results of the radionuclide. The dose distribution results are used to characterize the dose absorption distribution of the radionuclide in various organs and tissues within the patient's body.

[0096] S107. Based on the dose distribution results, determine the dose of radionuclide absorbed by each sub-region of interest.

[0097] Optionally, based on the dose distribution results, the amount of radionuclide absorbed by each sub-region of interest can be determined, and thus the dose of radionuclide absorbed by each sub-region of interest can be determined.

[0098] In this embodiment, by acquiring X-ray structural images and gamma-ray functional images at a single or multiple time points after tumor treatment, the user-selected region of interest (ROI) is segmented to obtain multiple sub-ROIs. Partial volume correction is applied to the gamma-ray functional images of each sub-ROI to obtain corrected gamma-ray functional images corresponding to each sub-ROI. A model is used to register the corrected gamma-ray functional images and X-ray structural images at a single or multiple time points to obtain registration results. Activity-dose rate conversion is performed on the registration results to obtain a dose distribution map of the radionuclide in the patient's body. The dose of the radionuclide absorbed by each sub-ROI is determined based on the dose distribution map. Partial volume effects can cause blurred boundaries in the gamma-ray functional images, leading to errors in activity measurement within each sub-ROI. This application performs partial volume correction on the gamma-ray functional images of each sub-ROI to ensure clear boundaries in the corrected gamma-ray functional images, thereby improving the registration accuracy between the gamma-ray functional images and the X-ray structural images. In this way, the impact of some volume effects on the resolution of radionuclide imaging can be reduced, thereby improving the accuracy of radionuclide image registration and thus improving the accuracy of radionuclide absorbed dose estimation.

[0099] In one alternative implementation, see [link to implementation details]. Figure 2 The specific operation of step S103 above can be as follows:

[0100] S201. Based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest, determine the point spread function corresponding to each voxel contained in the gamma-ray functional image of each sub-region of interest, and perform iterative partial volume correction on the gamma-ray functional image according to the point spread function corresponding to each voxel to obtain the iterative gamma-ray functional image.

[0101] Optionally, a three-dimensional spatial coordinate system is pre-constructed for the imaging system. When the imaging system begins to perform gamma-ray functional imaging and X-ray structural imaging, the voxel information of the gamma-ray functional image and the X-ray structural image is characterized by the three-dimensional spatial coordinate system.

[0102] Optionally, the radionuclide point source is a standard point source that provides radionuclide radiation. The radionuclide point source is set on a pre-constructed three-dimensional coordinate grid, and each grid point of the radionuclide point source has its corresponding point spread function. The point spread function is used to characterize the diffusion characteristics of the radionuclide corresponding to each radionuclide point source.

[0103] Optionally, based on the three-dimensional coordinates of each voxel in the gamma-ray functional image of each sub-region of interest, the radionuclide point source corresponding to each voxel in the gamma-ray functional image is determined, and the point spread function of the corresponding radionuclide point source is used as the point spread function of that voxel. Furthermore, the gamma-ray functional image of each sub-region of interest is iteratively corrected according to the point spread function of each voxel contained in the gamma-ray functional image, so that the gamma-ray functional image of each sub-region of interest is infinitely close to the true gamma-ray functional image, compensating for errors caused by the dilution or mixing of radionuclides.

[0104] Optionally, iterative partial volume correction is performed on the gamma-ray functional image simultaneously based on the point spread function corresponding to all voxels contained in the gamma-ray functional image to obtain the corrected gamma-ray functional image of the region of interest. It is worth noting that partial volume correction can be performed on the gamma-ray functional image of each sub-region of interest based on the point spread function of each voxel contained in the gamma-ray functional image of that sub-region of interest.

[0105] In one alternative implementation, see [link to implementation details]. Figure 3 and Figure 5 A 7×7×7 radionuclide point source was deployed at different locations within the field of view of a gamma camera. A uniform background activity was then added to the background region outside the sub-region of interest, with the activity of the point sources set higher than the background activity. Due to the nonlinear characteristic of the gamma imaging system's expectation-maximization reconstruction of ordered subsets, if the radionuclide point sources lack background activity perturbation, the reconstructed data will not accurately reflect the point diffusion characteristics of their respective grid points. See also... Figure 4 The full width at half maximum (FWHM) distribution of the same column of radionuclide point sources in the axial, radial, and circumferential dimensions can determine the system resolution of the radionuclide point sources in the three directions of three-dimensional space.

[0106] In one alternative implementation, see [link to implementation details]. Figure 6 , Figure 7 and Figure 8 The radial, circumferential, and axial curves of the point spread function of each radionuclide point source are fitted using a one-dimensional Gaussian function to obtain the system resolution of each radionuclide point source along the three directions.

[0107] It is worth noting that when the distance between the radionuclide point source and the collimator is the same, the full width at half maximum (FWHM) of the radionuclide point source remains consistent in the radial, axial, and circumferential directions.

[0108] S202. Apply a global primality consistency constraint to the iterated gamma-ray functional image to obtain a constrained gamma-ray functional image, so that the global primality value of the constrained gamma-ray functional image is consistent with the global primality value of the gamma-ray functional image before iteration, and use the constrained gamma-ray functional image as the corrected gamma-ray functional image.

[0109] Optionally, during the iterative partial volume correction process of the gamma-ray functional image, the voxel values ​​of the gamma-ray functional image after iteration may be inaccurate compared with the original voxel values ​​of the gamma-ray functional image before iteration. The voxel values ​​of the gamma-ray functional image before and after iteration remain unchanged through the overall voxel consistency constraint, that is, the resolution of the gamma-ray functional image before and after iteration does not change.

[0110] Optionally, the constrained gamma-ray functional image refers to the gamma-ray functional image obtained after iterative partial volume correction and then subject to global element consistency constraints.

[0111] Optionally, the constrained gamma-ray functional image is verified to determine whether the voxel information contained in the constrained gamma-ray functional image obtained after iterative partial volume correction and overall voxel consistency constraint is consistent with the voxel information of the real gamma-ray functional image. If they are the same, no further correction is required; if they are different, partial volume correction is performed.

[0112] In one alternative implementation, see [link to implementation details]. Figure 9 The specific operation of step S201 above can be as follows:

[0113] S901. Determine the average value of the iterative mask for each sub-region of interest based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest and the number of voxel points contained in the gamma-ray functional image of each sub-region of interest.

[0114] Optionally, the average value of the iterative mask for each sub-region of interest can be calculated using the following formula (1):

[0115]

[0116] Among them, T k,i p represents the average value of the iterative mask for the region of interest in the sub-subject during the k-th iteration. i (x,y,z) represents the mask of the i-th sub-region of interest, V i I represents the number of voxel points contained in the gamma-ray functional image of the sub-region of interest. k (x,y,z) represents the gamma-ray functional image obtained in the k-th iteration.

[0117] S902. Based on the average value of the iterative mask of each sub-region of interest, determine the three-dimensional coordinates of the iteratively synthesized reference image corresponding to each sub-region of interest.

[0118] Optionally, the three-dimensional coordinates of the iteratively synthesized reference image corresponding to each sub-region of interest can be calculated using the following formula (2):

[0119]

[0120] Among them, s k (x,y,z) is used to represent the three-dimensional coordinates of the reference image for iterative synthesis, T k,i p represents the average value of the iterative mask for the region of interest in the sub-subject during the k-th iteration. i (x,y,z) represents the mask for the i-th region of interest.

[0121] Optionally, when the imaging system starts generating a gamma-ray functional image, the three-dimensional coordinates contained in the gamma-ray functional image are automatically acquired; the iteratively synthesized reference image is an idealized gamma-ray functional image generated by multiple iterations during the partial volume correction process. The iteratively synthesized reference image is obtained by iterative partial volume correction based on the original gamma-ray functional image, and the three-dimensional coordinates of the iteratively synthesized reference image are obtained by iteratively updating the three-dimensional coordinates of the original gamma-ray functional image.

[0122] S903. Based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the iteratively synthesized reference image, perform iterative partial volume correction on the gamma-ray functional image of each sub-region of interest to obtain the iterative gamma-ray functional image.

[0123] In an optional implementation, step S903 above calculates the iterated gamma-ray functional image using the following formula (3):

[0124]

[0125] Among them, I k+1 (x,y,z) represents the gamma-ray functional image obtained after the (k+1)th iteration, I k (x,y,z) represents the gamma-ray functional image obtained after the k-th iteration, s k (x,y,z) represents the iteratively synthesized reference image obtained in the k-th iteration, and PSF(x,y,z) represents the point spread function corresponding to voxel (x,y,z) in the gamma-ray functional image of the sub-region of interest.

[0126] In one alternative implementation, see [link to implementation details]. Figure 10 The specific operation of step S202 above can be as follows:

[0127] S1001. Based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the initial mask mean of each sub-region of interest, determine the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest.

[0128] Optionally, the initial mask mean refers to the mask mean of the mask region that is marked and not iterated, and the grayscale adjustment region is used to measure the voxel error before and after the gamma-ray functional image iteration.

[0129] In one optional implementation, the grayscale adjustment region of the gamma-ray functional image in the sub-region of interest is calculated using the following formula (4):

[0130]

[0131] Where m(x,y,z) represents the grayscale adjustment region, PSF(x,y,z) represents the point spread function of voxels (x,y,z) in the gamma-ray functional image of the sub-region of interest, and s 0 (x,y,z) represents the initial mask mean of the sub-region of interest.

[0132] S1002. Based on the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest, apply an overall morphological consistency constraint to the iterated gamma-ray functional image to obtain the constrained gamma-ray functional image.

[0133] In one optional implementation, the constrained gamma-ray functional image is calculated using the following formula (5):

[0134]

[0135] Among them, I_CP k+1 (x,y,z) represents the intermediate gamma-ray functional image of the (k+1)th iteration, I k+1 (x,y,z) represents the gamma-ray functional image obtained after the (k+1)th iteration, I k (x,y,z) represents the gamma-ray functional image obtained after the k-th iteration, and m represents the grayscale adjustment area.

[0136] In one alternative implementation, see [link to implementation details]. Figure 14 The operation in step S106 above, "to perform curve fitting and integration on the dose rates corresponding to multiple time points to obtain the dose distribution results of the radionuclide," can be specifically described as follows:

[0137] S1401. Based on the voxel values ​​of each voxel in the gamma-ray functional image after multi-time point registration, determine the voxel mean value of each voxel. The voxel mean value refers to the voxel mean value between each voxel and its adjacent voxels.

[0138] Optionally, based on the three-dimensional coordinates of each voxel in the gamma-ray functional image after multi-time point registration, at least one voxel point adjacent to each voxel can be determined, and the average voxel value between each voxel and its adjacent voxels can be calculated to obtain the average voxel value corresponding to each voxel.

[0139] S1402. Iteratively update the voxel values ​​corresponding to each voxel based on the voxel mean value.

[0140] Optionally, the original voxel values ​​of each voxel can be replaced with the voxel mean value corresponding to each voxel to reduce the voxel error of the gamma-ray functional image.

[0141] S1403. Generate a new voxel matrix based on the updated voxel values ​​corresponding to each voxel.

[0142] Optionally, a voxel matrix corresponding to the gamma-ray functional image is generated based on the voxel values ​​of each voxel point contained in the gamma-ray functional image, and a new voxel matrix is ​​generated based on the voxel values ​​after each voxel point is replaced.

[0143] S1404. The new voxel matrix is ​​fitted and integrated using a volume-by-volume curve fitting method to obtain the dose distribution results of radionuclides at multiple time points.

[0144] Optionally, see Figure 15 For dose rates corresponding to multiple time points, a volume-by-volume curve fitting method is used to calculate the average value of each voxel and its neighboring voxels. This calculated average value is then filled into the current position of the voxel, and the value at the current position of each voxel is iteratively updated to obtain a new matrix. A voxel-by-volume fitting and integration is then performed on this new matrix to obtain the dose distribution results of the radionuclide at multiple time points. However, see [link to relevant documentation]. Figure 16 and Figure 17 Existing curve fitting methods fit each voxel point in the image, which can easily lead to abnormal truncation and discontinuity of voxel values. The volume-by-volume curve fitting method used in this application for dose rates at multiple time points can avoid this phenomenon.

[0145] In one alternative implementation, see [link to implementation details]. Figure 11 , Figure 12 as well as Figure 13 The organ contours of the partial volume-corrected results in this application are clearer than those in the prior art, and the average error of the activity measurement results obtained after partial volume correction in this application is the smallest.

[0146] In one alternative implementation, see [link to implementation details]. Figure 19 and Figure 21 For unregistered CT and SPECT images acquired after tumor treatment, see [link to relevant documentation]. Figure 20 and Figure 22 The registration results are more accurate and refined than those obtained by calibrating the images mentioned in this application.

[0147] The following describes the computer equipment and computer-readable storage medium used to implement the voxel-level radionuclide dose estimation method based on radionuclide imaging provided in this application. The specific implementation process and technical effects are described above and will not be repeated below.

[0148] Figure 23 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. See also... Figure 23 The computer device includes: a memory 2301 and a processor 2302. The memory 2301 stores a computer program that can run on the processor 2302. When the processor 2302 executes the computer program, it implements the steps in any of the above method embodiments.

[0149] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps in the various method embodiments described above.

[0150] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs an embodiment of any of the above-described voxel-level radionuclide dose estimation methods based on radionuclide imaging.

[0151] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0152] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute partial steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0153] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0154] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for estimating voxel-level radionuclide dose based on radionuclide imaging, characterized in that, The method includes: Acquire X-ray structural images and gamma-ray functional images at a single or multiple time points; The region of interest is segmented to obtain multiple sub-regions of interest; Partial volume correction is performed on the gamma-ray functional images of each sub-region of interest to obtain the corrected gamma-ray functional images; Register corrected gamma-ray functional images and X-ray structural images at a single or multiple time points; The registration results are converted from activity to dose rate to obtain the dose rate corresponding to a single time point or multiple time points; Curve fitting and integration are performed on the dose rates corresponding to a single time point or multiple time points to obtain the dose distribution results of the radionuclides; Based on the dose distribution results, the dose of radionuclide absorbed by each sub-region of interest is determined; The partial volume correction of the gamma-ray functional images of each sub-region of interest to obtain the corrected gamma-ray functional images includes: Based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest, the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest is determined, and the gamma-ray functional image is iteratively partially volume corrected based on the point spread function corresponding to each voxel to obtain the iterative gamma-ray functional image. The overall primality consistency constraint is applied to the iterated gamma-ray functional image to obtain a constrained gamma-ray functional image, so that the overall primality value of the constrained gamma-ray functional image is consistent with the overall primality value of the gamma-ray functional image before iteration, and the constrained gamma-ray functional image is used as the corrected gamma-ray functional image. The registration of corrected gamma-ray functional images and X-ray structural images at a single or multiple time points includes: Normalized mutual information is used as the optimization function to maximize the normalized mutual information of the corrected gamma-ray functional image and X-ray structural image at a single time point of rigid transformation, so as to achieve the registration of the corrected gamma-ray functional image and X-ray structural image at a single time point. Alternatively, X-ray structural images at multiple time points can be registered, and the registered X-ray structural images at multiple time points can be applied to the corrected gamma-ray functional images at the corresponding time points to achieve registration between the corrected gamma-ray functional images and the X-ray structural images at multiple time points.

2. The method for estimating voxel-level radionuclide doses according to claim 1, characterized in that, The step of determining the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest based on the three-dimensional coordinates of each voxel in the gamma-ray functional image of each sub-region of interest, and performing iterative partial volume correction on the gamma-ray functional image based on the point spread function corresponding to each voxel to obtain the iterative gamma-ray functional image includes: The average value of the iterative mask for each sub-region of interest is determined based on the three-dimensional coordinates of each voxel contained in the gamma-ray functional image of each sub-region of interest and the number of voxel points contained in the gamma-ray functional image of each sub-region of interest. Based on the average of the iterative masks of each sub-region of interest, determine the three-dimensional coordinates of the iteratively synthesized reference image corresponding to each sub-region of interest; Based on the point spread function corresponding to each voxel contained in the gamma-ray functional image of each sub-region of interest and the iterative synthesis reference image, the gamma-ray functional image of each sub-region of interest is iteratively partially volume corrected to obtain the iterative gamma-ray functional image.

3. The method for estimating voxel-level radionuclide doses according to claim 2, characterized in that, The step of performing iterative partial volume correction on the gamma-ray functional images of each sub-region of interest based on the point spread function corresponding to each voxel contained in the gamma-ray functional image of each sub-region of interest and the iteratively synthesized reference image to obtain the iterative gamma-ray functional image includes: Based on formula Calculate the gamma-ray functional image after iteration; in, This represents the gamma-ray functional image obtained after the (k+1)th iteration. This represents the gamma-ray functional image obtained after the k-th iteration. This represents the iteratively synthesized reference image obtained in the k-th iteration. Voxels in a gamma-ray functional image representing a sub-region of interest The corresponding point spread function.

4. The method for estimating voxel-level radionuclide doses according to claim 1, characterized in that, The step of applying a global element consistency constraint to the iterated gamma-ray functional image to obtain a constrained gamma-ray functional image includes: Based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the initial mask mean of each sub-region of interest, determine the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest. The overall morphological consistency constraint is applied to the iterated gamma-ray functional image based on the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest, so as to obtain the constrained gamma-ray functional image.

5. The method for estimating voxel-level radionuclide doses according to claim 4, characterized in that, The step of determining the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest based on the point spread function corresponding to each voxel in the gamma-ray functional image of each sub-region of interest and the initial mask mean of each sub-region of interest includes: Based on formula Calculate the grayscale adjustment area; in, This indicates the grayscale adjustment area. Voxels in a gamma-ray functional image representing a sub-region of interest The corresponding point spread function, This represents the initial mask mean of the sub-region of interest.

6. The method for estimating voxel-level radionuclide doses according to claim 4, characterized in that, The step of applying an overall morphological consistency constraint to the iterated gamma-ray functional image based on the grayscale adjustment region corresponding to the gamma-ray functional image of each sub-region of interest, to obtain the constrained gamma-ray functional image, includes: Based on formula The constrained gamma-ray functional image was calculated. in, This represents the constrained gamma-ray functional image after the (k+1)th iteration. This represents the gamma-ray functional image obtained after the (k+1)th iteration. This represents the gamma-ray functional image obtained after the k-th iteration, where m represents the grayscale adjustment area.

7. The method for estimating voxel-level radionuclide doses according to claim 1, characterized in that, The process of curve fitting and integration of dose rates at multiple time points to obtain the dose distribution results of radionuclides includes: Based on the voxel values ​​of each voxel contained in the gamma-ray functional image after multi-time point registration, the voxel mean value corresponding to each voxel is determined. The voxel mean value refers to the voxel mean value between each voxel and its adjacent voxels. Based on the average voxel value corresponding to each voxel, iteratively update the voxel value corresponding to each voxel. A new voxel matrix is ​​generated based on the updated voxel values ​​corresponding to each voxel. The new voxel matrix was fitted and integrated using a volume-by-volume curve fitting method to obtain the dose distribution results of radionuclides at multiple time points.

8. A computer device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program that can run on the processor, and when the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1 to 7.

Images