Dead time correction method and device, computer equipment, storage medium and program product
By training the correction model and applying it to the SPECT imaging system, the problem of inaccurate death time correction in traditional techniques is solved, and the imaging quality is improved, especially under high radioactivity conditions.
Patent Information
- Application Number
- CN202411917569.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-23
- Publication Date
- 2025-04-11
AI Technical Summary
The dead time correction model used in SPECT imaging systems in traditional techniques is inaccurate, resulting in a degradation of imaging quality.
By obtaining the projection map of the imaging object and training the correction model with training samples (including the lost projection map sample and the un lost projection map sample), the projection map after dead time correction is determined. This method considers the impact of the dead time effect between pixels and improves the accuracy of correction.
Improves the accuracy of dead time correction, thereby improving imaging quality and ensuring the imaging performance of SPECT system under high radioactivity conditions.
Smart Images
Figure CN119810237A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of imaging technology, and in particular to a dead time correction method, device, computer equipment, storage medium and program product. Background Art
[0002] Single-Photon Emission Computed Tomography (SPECT) is one of the most commonly used diagnostic imaging methods, and it is also one of the most important tools in the emerging integrated diagnosis and treatment technology. During SPECT imaging, if the activity of the injected radionuclide is high, dead time effect (loss of data) is likely to occur, which will affect the imaging quality.
[0003] In conventional technology, a correction model is used to perform dead time correction for the entire detector or each pixel of the detector in SPECT, and the correction models for each pixel of the detector are independent of each other.
[0004] However, the correction model in traditional technology is not accurate for the dead time correction of SPECT. Summary of the invention
[0005] Based on this, it is necessary to provide a dead time correction method, device, computer equipment, storage medium and program product that can improve the accuracy of dead time correction in order to solve the above technical problems.
[0006] In a first aspect, the present application provides a dead time correction method, which is applied to a single photon emission tomography system, and the method comprises:
[0007] obtaining a projection image of an imaging object;
[0008] According to the projection map and the correction model, the projection map after dead time correction is determined; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0009] In one embodiment, the method for obtaining training samples includes:
[0010] Obtain a sample of projection images without missing numbers;
[0011] The Monte Carlo method is used to determine the lost number projection map samples based on the non-lost number projection map samples and the detector response model; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
[0012] In one embodiment, obtaining a non-lost number projection map sample includes:
[0013] Obtain the radiation distribution map and different radiation activities of the preset radiation source;
[0014] According to the radiation distribution diagram and different radioactivity, the samples of projection diagram without missing numbers are determined.
[0015] In one embodiment, determining the dead time corrected projection map according to the projection map and the correction model includes:
[0016] The projection image is input into the correction model to determine the loss ratio of each pixel in the projection image;
[0017] According to the loss ratio of each pixel and the projection image, a projection image after dead time correction is determined.
[0018] In one embodiment, the method of training a calibration model includes:
[0019] Determine the missing number ratio sample according to the missing number projection map sample and the non-missing number projection map sample;
[0020] The initial network model is trained according to the loss ratio samples and the loss projection map samples to obtain a correction model.
[0021] In one embodiment, determining the dead time corrected projection map according to the projection map and the correction model includes:
[0022] The projection map is input into the calibration model to determine the dead time corrected projection map.
[0023] In a second aspect, an embodiment of the present application provides a dead time correction device, which is applied to a single photon emission tomography system, and the device includes:
[0024] An acquisition module, used for acquiring a projection image of an imaging object;
[0025] The input module is used to determine the projection map after dead time correction according to the projection map and the correction model; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0026] In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method provided in the first aspect when executing the computer program.
[0027] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method provided in the first aspect above are performed.
[0028] In a fifth aspect, an embodiment of the present application further provides a computer program product, including a computer program, which implements the steps of the method provided in the first aspect when the computer program is executed by a processor.
[0029] The above-mentioned dead time correction method, device, computer equipment, storage medium and program product, the method is applied to a single photon emission tomography system. By acquiring a projection image of an imaging object, the projection image after dead time correction is determined according to the projection image and a correction model; wherein the correction model is obtained by training an initial network model according to training samples, and the training samples include lost number projection image samples and non-lost number projection image samples. In this embodiment, the training samples used in the training process of the correction model are lost number projection image samples and non-lost number projection image samples, taking into account the influence of the dead time effect between pixels, and using the correction model trained with the training samples to perform dead time correction on the projection image can improve the accuracy of dead time correction on the projection image, thereby improving the quality of the projection image after dead time correction. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 is a schematic diagram of the distribution of photomultiplier tubes in one embodiment;
[0031] Figure 2 is a schematic diagram of count distribution of a detector in one embodiment;
[0032] Figure 3 A schematic diagram of the distribution of photomultiplier tubes in a detector corresponding to different triggering modes in one embodiment;
[0033] Figure 4 A schematic diagram of an application environment of a dead time correction method in an embodiment;
[0034] Figure 5 is a schematic diagram of the structure of a computer device in one embodiment;
[0035] Figure 6 A schematic diagram of a dead time correction method according to an embodiment of the present invention;
[0036] Figure 7 A schematic flow chart of the steps of a dead time correction method in another embodiment;
[0037] Figure 8 A schematic flow chart of the steps of a dead time correction method in another embodiment;
[0038] Fig. 9 A schematic flow chart of the steps of a dead time correction method in another embodiment;
[0039] Fig.10 A schematic flow chart of the steps of a dead time correction method in another embodiment;
[0040] Fig.11 A schematic flow chart of the steps of a dead time correction method in another embodiment;
[0041] Fig.12 FIG. 4 is a schematic diagram of the structure of a dead time correction device in an embodiment. DETAILED DESCRIPTION
[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0043] First of all, before specifically introducing the technical solutions of the disclosed embodiments of the present application, the background technology or technical evolution context based on the embodiments of the present application is introduced. In the medical field, single-photon emission computed tomography (SPECT) is one of the most commonly used diagnostic imaging methods, and in today's emerging integrated diagnosis and treatment technology, SPECT is also one of the most important tools. SPECT imaging technology is to introduce radioactive nuclide-labeled drugs into the phantom. After the radioactive nuclide-labeled drugs are redistributed in the phantom, the detector can detect the gamma rays generated by the radioactive nuclide, and obtain a SPECT image that can characterize the distribution and dynamic changes of the radioactive nuclide-labeled drugs in the phantom through image reconstruction. In the usual clinical diagnosis scenario, the activity of the injected radionuclides in SPECT imaging is different, and the detector count rate will not have a dead time effect (i.e., loss of counts). However, in the integrated diagnosis and treatment scenario, the requirements for quantitative imaging of therapeutic nuclides are higher than those of the diagnosis scenario. The activity of the injected radionuclides is often higher, which can be more than 10 times higher than that of the clinical diagnosis scenario. Dead time effects are prone to occur, which will affect the imaging quality.
[0044] In traditional technology, the dead time correction method is mainly for the entire detector or each pixel of the detector in SPECT, and the correction model is used to perform dead time correction. The correction models of each pixel of the detector are independent of each other and have no correlation. However, the pixels of the detector are not relatively independent. The dead time effects of two pixels at different distances are not the same. The two adjacent pixels will affect each other. The closer the distance, the greater the mutual influence. Assume that the photomultiplier tube (PMT) is used to receive gamma rays in the SPECT detector. The distribution of PMT is as follows: Figure 1 As shown, when the imaging phantom is a uniformly distributed and high-activity radiation source, the detector count distribution (the projection diagram after imaging by the signal received by the detector) is as follows Figure 2 As shown, from Figure 1 and Figure 2 It can be seen that the count distribution corresponding to the area where the PMT is located is more, and the count distribution corresponding to the area between adjacent PMTs is less, and the number of counts in the area where the PMT is located is less than that in the gap between adjacent PMTs. For imaging phantoms with unevenly distributed radiation sources, the number of counts in the projection image after imaging with the signal received by the detector is different from that in imaging phantoms with evenly distributed radiation sources.
[0045] Due to different detector triggering methods and different radiation source distributions, the loss ratio of each pixel will be different. Figure 2 The detector is triggered by a visible gamma photon detection event. Figure 3 As shown, Figure 3 In the image on the left, after PMT22 in the detector is triggered by a gamma photon detection event, the six PMTs adjacent to PMT22, that is, the other PMTs in the circle except PMT22, are used to calculate photon energy and no longer receive triggers. Figure 3 In the middle image, after PMT22 and PMT46 in the detector are simultaneously triggered by two gamma photon detection events, the six adjacent PMTs are used to calculate the photon energy and no longer receive triggers. PMT22 and PMT46 are far apart, and the two gamma photon detection events do not interfere with each other. Figure 3 In the right image, after PMT22 and PMT36 in the detector are triggered by two gamma photon detection events at the same time, the six adjacent PMTs are used to calculate the photon energy and no longer receive triggers. PMT22 and PMT36 are close to each other, and the two gamma photon detection events will interfere with each other. These two events need to be discarded, resulting in loss of data. Therefore, when there is a Figure 3 The large number of gamma photon detection events at different positions shown, whether the detection event is lost depends not only on the arrival time interval between the detection events, but also on the relative distance between the detection events, that is, the relative distance between the two triggered PMTs. However, in the conventional technology, the dead time of a single pixel is corrected by a correction model, which has the problem of inaccurate correction. In this regard, the present application provides a dead time correction method.
[0046] The dead time correction method provided in this application can be applied to SPECT systems, such as Figure 4As shown, the imaging system includes an imaging device 11 and a computer device 12, and the computer device 12 is connected to the imaging device 11 through a network. The imaging device 11 scans the imaging object to obtain scanning data, and transmits the scanning data to the computer device 12. The computer device 12 reconstructs the image according to the received scanning data to obtain a reconstructed image. The imaging device can be SPECT. The computer device can be, but is not limited to, an industrial computer, a laptop computer, and a tablet computer. The internal structure of the computer device can be as follows Figure 5 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a dead time correction method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.
[0047] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0048] In one embodiment, Figure 6 As shown, a dead time correction method is provided. This embodiment uses the method applied to a computer device in a SPECT system as an example. In this embodiment, the method includes the following steps:
[0049] Step 600: Obtain a projection image of the imaging object;
[0050] The projection image may be an image obtained by scanning the imaging object using an imaging device, and the imaging device used may be SPECT. The type of the projection image corresponds to the type of the imaging device, that is, the projection image may be a SPECT image. This embodiment does not limit the type of the projection image of the imaging object obtained, as long as its function can be achieved.
[0051] The computer device may obtain the projection image from the image device post-processing workstation, or from a server such as a Picture Archiving and Communication Systems (PACS). Alternatively, the projection image may be pre-stored in the memory of the computer device, and the computer device may directly obtain it from the memory when needed. This embodiment does not limit the specific method of obtaining the projection image, as long as its function can be achieved.
[0052] Step 610: Determine the projection map after dead time correction according to the projection map and the correction model; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the projection map samples with lost numbers and the projection map samples without lost numbers.
[0053] The correction model may be pre-trained and stored in the memory of the computer device. The training process of the correction model includes:
[0054] Obtain training samples; the training samples include lost number projection map samples and non-lost number projection map samples.
[0055] The lost number projection image sample refers to the image obtained by imaging the imaging phantom when the imaging device has a dead time effect during the imaging process. The imaging phantom can be a radiation source preset by the staff, or an animal body, etc. The non-lost number projection image sample refers to the image obtained by imaging the imaging phantom when the imaging device has no dead time effect during the imaging process. The types of the lost number projection image sample and the non-lost number projection image sample are the same as the types of the projection image.
[0056] The computer device may obtain the training samples from the image device post-processing workstation, or from a server such as PACS, etc. Alternatively, the training samples may be pre-stored in the memory of the computer device, and the computer device may directly obtain them from the memory when needed.
[0057] The initial network model is trained using the training samples to obtain a correction model.
[0058] After acquiring the projection image of the imaging object, the computer device determines the projection image after dead time correction according to the projection image and the pre-trained correction model. This embodiment does not limit the specific method of determining the projection image after dead time correction according to the projection image and the correction model, as long as its function can be achieved.
[0059] The dead time correction method provided in the embodiment of the present application is applied to the SPECT system, by acquiring the projection image of the imaging object, and determining the projection image after dead time correction according to the projection image and the correction model; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection image samples and the non-lost number projection image samples. In this embodiment, the training samples used in the training process of the correction model are the lost number projection image samples and the non-lost number projection image samples, and the influence of the dead time effect between each pixel is taken into account. The correction model trained with the training samples is used to perform dead time correction on the projection image, which can improve the accuracy of the dead time correction of the projection image, thereby improving the quality of the projection image after the dead time correction.
[0060] In one embodiment, Figure 7 As shown, a method for obtaining training samples is involved, and the steps of the method include:
[0061] Step 700: Obtain a sample of the projection image without missing numbers.
[0062] The non-lost number projection map sample can be pre-stored in the memory of the computer device, and the computer device directly obtains it from the memory when needed. The description of the non-lost number projection map sample can refer to the specific description of the above embodiment, which will not be repeated here.
[0063] Step 710: using the Monte Carlo method, determine the lost number projection map samples according to the non-lost number projection map samples and the detector response model; the detector response model is used to characterize the detector space count distribution corresponding to different rays.
[0064] The Monte Carlo method is also called the statistical simulation method. Its working process is divided into three main steps: constructing or describing the probability process; implementing sampling from a known probability distribution; and establishing various estimators. The detector response model is a physical model pre-set by the staff, which can characterize the spatial count distribution of the detector under the incidence of photons with different spatial and temporal probability distributions.
[0065] After obtaining the non-lost number projection map sample, the computer device determines the lost number projection map sample according to the non-lost number projection map sample and the detector response model. Specifically, a probability process is constructed according to the non-lost number projection map sample, that is, the average interval time of the emission time of adjacent photons on the pixel is determined according to the radioactivity of each pixel in the non-lost number projection map sample, and the exponentially distributed photon arrival time is constructed with the average interval time; a random event is generated as the non-lost number projection map sample according to the non-lost number projection map sample and the exponentially distributed photon arrival time, and the event distribution of each pixel in the non-lost number projection map sample is the random photon arrival time with the average interval time exponential distribution; the random event is input into the detector response model, and the detector spatial count distribution under the input exponentially distributed random photon incidence is obtained through the detector response model, so that the lost number projection map sample can be obtained.
[0066] In this embodiment, the Monte Carlo method is used to determine the lost number projection map samples according to the obtained non-lost number projection map samples and the detector response model. This method of using the Monte Carlo method to determine the lost number projection map samples is fast and has strong randomness, which can improve the efficiency and quantity of the obtained training samples, thereby improving the accuracy of the correction model trained using the training samples.
[0067] In one embodiment, Figure 8 As shown, an implementation method for obtaining a sample of a projection graph without missing numbers includes:
[0068] Step 800: Obtain a radiation distribution diagram and different radiation activities of a preset radiation source.
[0069] The preset radioactive source refers to a phantom that can generate radiation. Each pixel value in the radioactive distribution map of the preset radioactive source represents the proportion of the activity of the preset radioactive source on the pixel. The radioactive distribution map and different radioactive activities can be pre-stored in the memory of the computer device, and the computer directly obtains them from the memory. Among them, the radioactive distribution map can be a random normalized radioactive distribution ratio map.
[0070] Step 810: Determine the projection map samples without missing numbers according to the radiation distribution map and different radiation activities.
[0071] After obtaining the radioactivity distribution map and different radioactivity activities, the computer device can determine the projection map sample without missing numbers according to the radioactivity distribution map and the different radioactivity activities. In other words, by calculating the product between the radioactivity distribution map and the different radioactivity activities, the radioactivity distribution map with the same radioactivity distribution ratio and different total activities, i.e., the projection map sample without missing numbers, can be obtained.
[0072] In this embodiment, the non-lost projection map samples are determined by obtaining the radiation distribution diagram of the preset radiation source and different radiation activities. This method of determining the non-lost projection map samples is quick and easy to implement, and can improve the efficiency of obtaining the non-lost projection map samples.
[0073] In one embodiment, Fig. 9 As shown, another implementation method of the training correction model is involved, and the steps of the implementation method include:
[0074] Step 900: Determine the number loss ratio samples according to the number loss projection graph samples and the number non-loss projection graph samples.
[0075] After obtaining the training samples, the computer device determines the missing number ratio samples according to the missing number projection map samples and the non-missing number projection map samples in the training samples. Specifically, the ratio between the missing number projection map samples and the non-missing number projection map samples is determined, that is, the ratio of each pixel value in the missing number projection map samples to the pixel value at the corresponding position in the non-missing number projection map samples is calculated, and the difference between 1 and the ratio is calculated to obtain the missing number ratio samples.
[0076] Step 910: Train the initial network model according to the loss ratio samples and the loss projection graph samples to obtain a correction model.
[0077] After obtaining the loss ratio sample, the computer device inputs the loss projection map sample and the loss ratio sample into the initial network model, processes the loss projection map sample through the initial network model, and obtains the initial loss ratio; determines the loss function according to the initial loss ratio and the loss ratio sample, and adjusts the parameters of the initial network model according to the loss function to realize the training of the initial network model. When the training end condition is met, a correction model can be obtained. The training end condition can be that the number of iterations reaches a preset number threshold, or it can be that the loss function converges, which is not limited in this embodiment.
[0078] In this embodiment, the loss ratio samples are determined according to the loss ratio samples and the non-loss projection map samples; the initial network model is trained according to the loss ratio samples and the loss projection map samples to obtain the correction model. In this embodiment, another method for training the correction model is provided, and the user can choose according to the actual application scenario, which can improve the practicality of the dead time correction method.
[0079] In one embodiment, using Fig. 9 In the case where the correction model is obtained by training the method of the embodiment shown in the figure, Fig.10 As shown, it is related to an implementation method of determining a projection map after dead time correction according to the projection map and the correction model, and the steps of the implementation method include:
[0080] Step 101: Input the projection image into the calibration model to determine the loss ratio of each pixel in the projection image.
[0081] After acquiring the projection image of the imaging object, the computer device inputs the projection image into the correction model, analyzes the projection image through the correction model, and determines the loss ratio of each pixel in the projection image.
[0082] Step 102: Determine a projection image after dead time correction according to the loss ratio of each pixel and the projection image.
[0083] After obtaining the number loss ratio of each pixel, the computer device determines the projection image after the dead time correction according to the number loss ratio of each pixel and the projection image. That is, by using the pixel value of each pixel in the projection image and the number loss ratio of the corresponding pixel, a non-number loss projection image, that is, a dead time corrected projection image, can be obtained.
[0084] In this embodiment, the projection image is input into the correction model to determine the loss ratio of each pixel in the projection image; based on the loss ratio of each pixel and the projection image, the projection image after dead time correction is determined. In this embodiment, another method for correcting the projection image through the correction model is provided, and the user can choose according to the actual application requirements. In this embodiment, not only the projection image after dead time correction can be obtained, but also the loss ratio of each pixel in the projection image can be obtained, which is convenient for the subsequent improvement of the detector structure in the imaging device according to the loss ratio of each pixel, and can improve the practicality of the dead time correction method.
[0085] In one embodiment, another implementation method of determining a projection map after dead time correction according to a projection map and a correction model includes:
[0086] The projection map is input into the calibration model to determine the dead time corrected projection map.
[0087] After obtaining the projection image of the imaging object, the computer device inputs the projection image into the correction model, and corrects the dead time of the projection image through the correction model, so that the projection image after the dead time correction can be directly obtained.
[0088] In an optional embodiment, the projection image of the imaging object acquired by the computer device includes the number of photons received by the detector in the case of loss of numbers during the imaging process of the single photon emission tomography system, and the duration of the signal acquisition when generating the projection image. The computer device directly inputs the projection image into the correction model to determine the projection image after dead time correction. The projection image after dead time correction includes the number of photons in the case of no loss of numbers. The projection image of the imaging object acquired by the computer device may also include a first projection sub-image and a second projection sub-image, the first projection sub-image includes the number of photons received by the detector in the case of loss of numbers during the imaging process of the single photon emission tomography system, and the second projection sub-image includes the duration of the signal acquisition when the projection image is generated using the single photon emission tomography system. The computer device inputs the first projection sub-image and the second projection sub-image into the correction model, analyzes the first projection sub-image and the second projection sub-image through the correction model, and determines the projection image after dead time correction.
[0089] The training process of the correction model corresponding to the present embodiment may include: after obtaining the training sample, the computer device inputs the training sample into the initial network model, and the initial network model processes the loss projection map sample to output the initial non-loss projection map; and determines the loss function according to the initial non-loss projection map and the non-loss projection map sample, and adjusts the parameters of the initial network model according to the loss function to realize the training of the initial network model. When the training end condition is met, the correction model can be obtained. The training end condition can be that the number of iterations reaches a preset number threshold, or it can be that the loss function converges, which is not limited in the present embodiment.
[0090] In this embodiment, the projection image is directly input into the correction model, and the correction model can output the projection image after the dead time correction. This method of determining the projection image after the dead time correction is quick and easy to implement. In addition, the above embodiment provides two implementation methods for determining the projection image after the dead time correction based on the projection image and the correction model. The user can choose any one or both implementation methods, and compare the initial dead time corrected projection images obtained by the two implementation methods to determine the dead time corrected projection image. Among them, it is possible to select a projection image with the best quality from the two initial dead time corrected projection images as the dead time corrected projection image, or it is possible to determine the dead time corrected projection image by determining the average of the two initial dead time corrected projection images.
[0091] In one embodiment, the projection image includes a plurality of projection sub-images, and the dead-time-corrected projection image includes a plurality of dead-time-corrected projection sub-images.
[0092] The computer device sends control information to the imaging device to control the imaging device to perform imaging at different angles on the imaging object, and multiple projection sub-images corresponding to different angles can be obtained. The computer device inputs the multiple projection sub-images into the correction model, and the correction model can be used to perform dead time correction on each projection sub-image, and obtain multiple corresponding dead time corrected projection sub-images. When the computer device inputs the multiple projection sub-images into the correction model, the multiple projection sub-images can be sorted first, and each projection sub-image can be input into the correction model in sequence according to the sorted order; or the multiple projection sub-images can be sorted first, and each projection sub-image can be input into the correction model in sequence according to the sorted order; or Figure 1 The data are input into the calibration model, and this embodiment does not impose any limitation on this, as long as the function can be achieved.
[0093] In the case where the acquired projection image of the imaging object includes a plurality of projection sub-images, and the dead time corrected projection image includes a plurality of dead time corrected projection sub-images, the method further includes:
[0094] Image reconstruction is performed based on multiple dead-time corrected projection sub-images to obtain a three-dimensional projection image of the imaging object.
[0095] After obtaining multiple dead time corrected projection sub-images through the correction model, the computer device reconstructs the multiple dead time corrected projection sub-images to obtain a three-dimensional projection image of the imaging object. This embodiment does not limit the specific method of image reconstruction, as long as its function can be achieved.
[0096] In an optional embodiment, the computer device may perform feature detection and extraction on each dead-time-corrected projection sub-image, and perform feature matching on multiple dead-time-corrected projection sub-images based on the extracted images to obtain a three-dimensional projection image of the imaging object.
[0097] In another optional embodiment, the computer device may input a plurality of dead-time-corrected projection sub-images into a pre-trained reconstruction model to obtain a three-dimensional projection image.
[0098] In this embodiment, a three-dimensional projection image of the imaging object is obtained by performing image reconstruction on a plurality of projection sub-images after dead time correction, and the three-dimensional projection image obtained in this way is more accurate.
[0099] See also Fig.11 An embodiment of the present application provides a dead time correction method, which is applied to a SPECT system. The method comprises the following steps:
[0100] Step 110, obtaining a radiation distribution diagram and different radiation activities of a preset radiation source;
[0101] Step 111, determining a sample of projection images without missing numbers according to the radiation distribution map and different radiation activities;
[0102] Step 112, using the Monte Carlo method, according to the non-lost number projection map samples and the detector response model, determine the lost number projection map samples; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons;
[0103] Step 113, using the lost number projection map samples and the non-lost number projection map samples to train the initial network model to obtain a correction model;
[0104] Step 114, obtaining a plurality of projection sub-images of the imaging object, and inputting the plurality of projection sub-images into a correction model to obtain a plurality of projection sub-images after dead time correction;
[0105] Step 115 : reconstruct an image based on the multiple projection sub-images after dead time correction to obtain a three-dimensional projection image of the imaging object.
[0106] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0107] Based on the same inventive concept, the embodiment of the present application also provides a dead time correction device for implementing the dead time correction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more dead time correction device embodiments provided below can refer to the limitations of the dead time correction method above, and will not be repeated here.
[0108] In one embodiment, Fig.12 As shown, a dead time correction device 20 is provided, which is applied to a SPECT system, and includes: an acquisition module 21 and an input module 22, wherein:
[0109] An acquisition module 21, used to acquire a projection image of an imaging object;
[0110] The input module 22 is used to determine the projection map after dead time correction according to the projection map and the correction model; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0111] In one embodiment, the dead time correction device 20 further includes a training module. The training module includes an acquisition unit and a first determination unit. The acquisition unit is used to acquire non-lost number projection map samples. The first determination unit is used to determine the lost number projection map samples according to the non-lost number projection map samples and the detector response model using the Monte Carlo method; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
[0112] In one embodiment, the acquisition unit is specifically used to acquire a radiation distribution diagram and different radiation activities of a preset radiation source; and determine a non-missing projection diagram sample according to the radiation distribution diagram and the different radiation activities.
[0113] In one embodiment, the input module 22 includes an input unit and a second determination unit. The input unit is used to input the projection image into the correction model to determine the loss ratio of each pixel in the projection image; the input unit is used to determine the projection image after dead time correction based on the loss ratio of each pixel and the projection image.
[0114] In one embodiment, the training module is further used to determine the loss ratio samples according to the loss ratio samples and the non-loss projection map samples; and train the initial network model according to the loss ratio samples and the loss projection map samples to obtain the correction model.
[0115] In one embodiment, the input module 22 is specifically used to input the projection image into the correction model to determine the projection image after dead time correction.
[0116] In one embodiment, the projection image includes a plurality of scanning sub-images, and the corrected projection image includes a plurality of corrected projection sub-images. The dead time correction device 20 also includes a reconstruction module. The reconstruction module is used to perform image reconstruction based on the plurality of dead time corrected projection sub-images to obtain a three-dimensional projection image of the imaging object.
[0117] Each module in the dead time correction device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.
[0118] In one embodiment, a computer device is provided, whose internal structure diagram can be as follows: Figure 5 shown.
[0119] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:
[0120] obtaining a projection image of an imaging object;
[0121] According to the projection map and the correction model, the projection map after dead time correction is determined; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0122] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining non-lost number projection map samples; using the Monte Carlo method, determining the lost number projection map samples based on the non-lost number projection map samples and the detector response model; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
[0123] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining a radiation distribution map and different radioactivity of a preset radiation source; and determining a non-missing projection map sample according to the radiation distribution map and the different radioactivity.
[0124] In one embodiment, when the processor executes the computer program, the following steps are also implemented: inputting the projection image into the correction model to determine the loss ratio of each pixel in the projection image; and determining the projection image after dead time correction based on the loss ratio of each pixel and the projection image.
[0125] In one embodiment, when the processor executes the computer program, the following steps are also implemented: determining the loss ratio samples according to the loss ratio samples and the non-loss projection map samples; training the initial network model according to the loss ratio samples and the loss projection map samples to obtain the correction model.
[0126] In one embodiment, when the processor executes the computer program, the following steps are further implemented: inputting the projection map into the correction model, and determining the projection map after dead time correction.
[0127] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:
[0128] obtaining a projection image of an imaging object;
[0129] According to the projection map and the correction model, the projection map after dead time correction is determined; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0130] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining non-lost number projection map samples; using the Monte Carlo method, based on the non-lost number projection map samples and the detector response model, determining the lost number projection map samples; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
[0131] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining a radiation distribution map and different radioactivity of a preset radiation source; and determining a non-missing projection map sample according to the radiation distribution map and the different radioactivity.
[0132] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: inputting the projection image into the correction model to determine the loss ratio of each pixel in the projection image; and determining the projection image after dead time correction based on the loss ratio of each pixel and the projection image.
[0133] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: determining the loss ratio samples based on the loss ratio samples and the non-loss projection map samples; training the initial network model based on the loss ratio samples and the loss projection map samples to obtain a correction model.
[0134] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: inputting the projection map into the correction model, and determining the projection map after dead time correction.
[0135] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:
[0136] obtaining a projection image of an imaging object;
[0137] According to the projection map and the correction model, the projection map after dead time correction is determined; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
[0138] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining non-lost number projection map samples; using the Monte Carlo method, based on the non-lost number projection map samples and the detector response model, determining the lost number projection map samples; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
[0139] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining a radiation distribution map and different radioactivity of a preset radiation source; and determining a non-missing projection map sample according to the radiation distribution map and the different radioactivity.
[0140] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: inputting the projection image into the correction model to determine the loss ratio of each pixel in the projection image; and determining the projection image after dead time correction based on the loss ratio of each pixel and the projection image.
[0141] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: determining the loss ratio samples based on the loss ratio samples and the non-loss projection map samples; training the initial network model based on the loss ratio samples and the loss projection map samples to obtain a correction model.
[0142] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: inputting the projection map into the correction model, and determining the projection map after dead time correction.
[0143] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0144] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0145] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A dead time correction method, characterized in that: Applied to a single photon emission tomography system, the method comprises: obtaining a projection image of an imaging object; According to the projection map and the correction model, the projection map after dead time correction is determined; wherein the correction model is obtained by training the initial network model according to the training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
2. The method according to claim 1, characterized in that The method for obtaining the training sample includes: Obtaining the non-lost number projection image sample; The lost number projection image samples are determined according to the non-lost number projection image samples and a detector response model using a Monte Carlo method; the detector response model is used to characterize the detector spatial count distribution corresponding to different photons.
3. The method according to claim 2, characterized in that The obtaining of the non-lost number projection image sample comprises: Obtain the radiation distribution map and different radiation activities of the preset radiation source; The non-missing projection map sample is determined according to the radiation distribution map and the different radioactivity.
4. The method according to claim 1, characterized in that: Determining the projection map after dead time correction according to the projection map and the correction model includes: Inputting the projection image into the calibration model to determine the loss ratio of each pixel in the projection image; The projection image after the dead time correction is determined according to the loss number ratio of each pixel and the projection image.
5. The method according to claim 4, characterized in that The method of training the correction model comprises: Determine a lost number ratio sample according to the lost number projection map sample and the non-lost number projection map sample; The initial network model is trained according to the loss number ratio samples and the loss number projection map samples to obtain the correction model.
6. The method according to claim 1, characterized in that Determining the projection map after dead time correction according to the projection map and the correction model includes: The projection image is input into the correction model to determine the projection image after the dead time correction.
7. A dead time correction device, characterized in that: Applied to a single photon emission tomography system, the device comprises: An acquisition module, used for acquiring a projection image of an imaging object; An input module is used to determine the projection map after dead time correction based on the projection map and the correction model; wherein the correction model is obtained by training the initial network model based on training samples, and the training samples include the lost number projection map samples and the non-lost number projection map samples.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.