Dead time correction method and system, image reconstruction method and system, PET device, electronic device and storage medium

Abstract

The invention discloses a dead time correction method and system, an image reconstruction method and system, PET equipment, electronic equipment and a storage medium. The dead time correction method comprises the following steps: acquiring measurement true event data and corresponding activity which are acquired for multiple times; performing multiple fitting processing on the measured true event data and the corresponding activity, the multiple fitting processing including at least one linear fitting and non-linear dead time model fitting to obtain a dead time constant and predicted true event data, the predicted true event data being obtained through a non-linear dead time model having the dead time constant; and determining a dead time correction factor at least according to the predicted true event data. According to the dead time correction method provided by the invention, the precision and robustness of dead time correction are effectively improved by combining linear and nonlinear multiple fitting processing strategies.

Description

Technical Field

[0001] This application relates to the field of positron emission tomography (PET) technology, specifically to dead time correction methods and systems, image reconstruction methods and systems, PET equipment, electronic equipment, and storage media. Background Technology

[0002] During PET imaging, the detector inherently suffers from a dead-time effect when receiving coincidence events. During this period, the detector cannot record newly arriving signals. The dead-time effect is more pronounced at higher radioactivity levels and count rates, potentially leading to the loss of coincidence events and causing a discrepancy between the measured and true count rates. This, in turn, affects image quality and the accuracy of quantitative analysis.

[0003] Furthermore, PET detection systems can be affected by various factors in practical applications, leading to fluctuations in detection performance. These fluctuations may cause non-ideal variations in the actual measured count rate.

[0004] Therefore, current dead-time correction methods still have room for improvement in terms of correction accuracy and stability.

[0005] Here, we hope to provide a more accurate and robust dead-time correction scheme to further improve the imaging performance of PET systems.

[0006] The background art description is provided solely for the purpose of understanding the relevant technologies in this field and is not intended as an admission of prior art. Summary of the Invention

[0007] In this regard, the embodiments of this application aim to provide a dead time correction method and system, an image reconstruction method and system, a PET device, an electronic device, and a storage medium, which can improve the correction accuracy and stability of the dead time correction method.

[0008] In a first aspect, embodiments of this application provide a dead time correction method, comprising:

[0009] Acquire multiple sets of measured true event data and corresponding activity;

[0010] The measured true event data and the corresponding activity are subjected to multiple fitting processes, including at least one linear fitting and one nonlinear dead time model fitting, to obtain the dead time constant and the predicted true event data, wherein the predicted true event data is obtained by a nonlinear dead time model with the dead time constant.

[0011] The dead time correction factor is determined based at least on the predicted true event data.

[0012] In some embodiments, the step of performing multiple fitting processes on the measured true event data and the corresponding activity includes:

[0013] Perform a first linear fit on the measured true event data and the corresponding activity to obtain first fitted true event data; and

[0014] The nonlinear dead-time model is fitted to the first fitted true event data and the measured true event data to solve for the dead-time constant.

[0015] Based on the first fitted true event data and the dead time constant, the predicted true event data is calculated using the nonlinear dead time model.

[0016] In some embodiments, performing a first linear fit on the measured true event data and the corresponding activity to obtain first fitted true event data includes:

[0017] The first fitting slope is obtained by linearly fitting the multiple collected measurement true event data and the corresponding activity using the linear least squares method.

[0018] Based on the first fitting slope and activity, the first fitted true event data is calculated.

[0019] In some embodiments, performing the nonlinear dead-time model fitting on the first fitted true event data and the measured true event data to solve for the dead-time constant includes:

[0020] The first fitted true event data is used as the true value, and the measured true event data is used as the observed value and substituted into the nonlinear dead time model.

[0021] The nonlinear dead-time model is fitted using the nonlinear least squares method to solve for the dead-time constant.

[0022] In some embodiments, the nonlinear dead-time model is a paralysis model.

[0023] In some embodiments, the paralysis model has the following expression:

[0024]

[0025] Among them, R True Corresponding to the true value, R Meas For the observed value, τ is the dead time constant.

[0026] In some embodiments, the nonlinear dead-time model is a non-paralyzed model or a hybrid model comprising a weighted combination of a paralyzed model and a non-paralyzed model;

[0027] The non-paralyzed model has the following expression:

[0028] R Meas =R True / (1+R True ·τ),

[0029] Among them, R True Corresponding to the true value, R Meas For the observed value, τ is the dead time constant;

[0030] The hybrid model has the following expression:

[0031]

[0032] Among them, R True Corresponding to the true value, R Meas For the observed values, τ is the dead time constant, and λ1 and λ2 are weights.

[0033] In some embodiments, the step of performing multiple fitting processes on the measured true event data and the corresponding activity further includes:

[0034] A second linear fit is performed on the predicted true event data and the corresponding activity to obtain the second fitted true event data;

[0035] The step of determining the dead time correction factor based at least on the predicted true event data includes: determining the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

[0036] In some embodiments, performing a second linear fit on the predicted true event data and the corresponding activity to obtain second fitted true event data includes:

[0037] The second fitting slope is obtained by linearly fitting the predicted true event data collected multiple times and the corresponding activity using the linear least squares method.

[0038] Based on the second fitting slope and activity, the second fitted true event data are calculated.

[0039] In some embodiments, determining the dead time correction factor based at least on the predicted true event data includes:

[0040] The dead time correction factor is determined based on the ratio of the first fitted true event data to the predicted true event data.

[0041] In some embodiments, the true event data is the true event count rate.

[0042] In some embodiments, acquiring multiple sets of measured true event data and corresponding activity includes:

[0043] Obtain the number of timely and delayed matching events for each data collection.

[0044] The number of timely compliant events and the number of delayed compliant events are used to determine the measurement true event data collected each time;

[0045] Obtain the activity level corresponding to each data collection.

[0046] In a second aspect, embodiments of this application provide an image reconstruction method, comprising:

[0047] During image reconstruction, the dead time correction factor determined according to the dead time correction method described in the first aspect is used to correct for dead time effects on coincident events.

[0048] In a third aspect, embodiments of this application provide a dead time correction system, which includes:

[0049] The data acquisition module is configured to acquire multiple sets of measurement true event data and corresponding activity.

[0050] The fitting module is configured to perform multiple fitting processes on the measured true event data and the corresponding activity. The multiple fitting processes include at least one linear fitting and one nonlinear dead time model fitting to obtain the dead time constant and the predicted true event data. The predicted true event data is obtained by the nonlinear dead time model with the dead time constant.

[0051] The correction factor determination module is configured to determine the dead time correction factor based at least on the predicted true event data.

[0052] In some embodiments, the fitting module includes:

[0053] A first linear fitting unit is configured to perform a first linear fitting on the measured true event data and the corresponding activity to obtain first fitted true event data; and

[0054] The nonlinear fitting unit is configured to perform the nonlinear dead-time model fitting on the first fitted true event data and the measured true event data, and solve for the dead-time constant.

[0055] The prediction unit is configured to calculate the predicted true event data based on the first fitted true event data and the dead time constant using the nonlinear dead time model.

[0056] In some embodiments, the fitting module further includes:

[0057] The second linear fitting unit is configured to perform a second linear fitting on the predicted true event data and the corresponding activity to obtain the second fitted true event data.

[0058] The correction factor determination module is further configured to determine the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

[0059] In some embodiments, the correction factor determination module is further configured to determine the dead time correction factor based on the ratio of the first fitted true event data to the predicted true event data.

[0060] In some embodiments, the nonlinear dead-time model is a paralyzed model, a non-paralyzed model, or a hybrid model comprising a weighted combination of a paralyzed model and a non-paralyzed model.

[0061] In some embodiments, the true event data is the true event count rate.

[0062] In some embodiments, the data acquisition module includes:

[0063] The event acquisition unit is configured to acquire the number of timely matching events and the number of delayed matching events for each acquisition.

[0064] The preprocessing unit is configured to determine the measured true event data for each acquisition based on the number of timely matching events and the number of delayed matching events;

[0065] The activity acquisition unit is configured to acquire the activity corresponding to each data collection.

[0066] In a fourth aspect, embodiments of this application provide an image reconstruction system, comprising:

[0067] According to the dead time correction system described in the third aspect; and

[0068] The image reconstruction module is configured to perform dead-time effect correction on coincident events during the image reconstruction process using the dead-time correction factor determined by the dead-time correction system.

[0069] In a fifth aspect, embodiments of this application provide a PET device, which includes:

[0070] According to the dead time correction system described in the third aspect; or

[0071] The image reconstruction system according to the fourth aspect.

[0072] In a sixth aspect, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores computer program instructions, and the processor, when executing the computer program instructions, implements the method as described in the first or second aspect.

[0073] In a seventh aspect, embodiments of this application provide a computer storage medium storing computer program instructions, wherein when the computer program instructions are executed by a processor, the method described in the first or second aspect is implemented.

[0074] The dead-time correction method provided in this application determines the dead-time correction factor by performing multiple fitting processes on measured true event data and corresponding activities. Specifically, it uses at least one linear fitting and one nonlinear dead-time model fitting to fit the measured true event data and corresponding activities, obtaining a dead-time constant for the nonlinear dead-time model. Predicted true event data is then obtained using the nonlinear dead-time model with the determined dead-time constant, and the dead-time correction factor is determined based on the obtained predicted true event data. This scheme solves the technical problem of insufficient accuracy in traditional single-fitting methods when dealing with data fluctuations and detector instabilities. By employing a multiple-fitting strategy, this scheme can gradually suppress random fluctuations in the measured data, making the solved dead-time constant and correction factor more accurate and stable, thereby significantly improving the accuracy and robustness of dead-time correction.

[0075] In a further embodiment of this application, the multiple fitting process may include: performing a first linear fit on the measured true event data and the corresponding activity to obtain first fitted true event data; performing a nonlinear dead-time model fit on the first fitted true event data and the measured true event data to solve for the dead-time constant; and calculating the predicted true event data based on the first fitted true event data and the dead-time constant using the nonlinear dead-time model. This further solution solves the technical problem that directly performing nonlinear fitting on raw measurement data containing fluctuations and noise can easily lead to error amplification and model distortion. By performing a two-stage process of smoothing the data through linear fitting before performing nonlinear fitting to extract physical parameters, this further solution can better separate noise suppression from accurate modeling of dead-time effects, avoiding noise interference in the nonlinear fitting process.

[0076] In a further embodiment of this application, the multiple fitting process may further include: performing a second linear fit on the predicted true event data and the corresponding activity to obtain second fitted true event data; and determining the dead-time correction factor based on the ratio of the second fitted true event data to the predicted true event data. This further approach further avoids the problem of insufficient consistency of the correction factor across the entire activity range due to possible local biases in the activity space. By adding a subsequent linear fit to re-optimize the predicted true event data, this further approach eliminates fitting residuals and nonlinear biases, making the correction factor smoother and more consistent across the entire activity range. This results in a higher accuracy and stronger robustness of the obtained dead-time correction factor, significantly improving the quantitative accuracy of images under high activity conditions.

[0077] Optional features and other effects of the embodiments of this application are described in part below, and in part will be apparent from reading this document. Attached Figure Description

[0078] The embodiments of this application will be described in detail with reference to the accompanying drawings. The elements shown are not limited to the scale shown in the drawings, and the same or similar reference numerals in the drawings denote the same or similar elements, wherein:

[0079] Figure 1 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0080] Figure 2 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0081] Figure 3 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0082] Figure 4 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0083] Figure 5 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0084] Figure 6 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0085] Figure 7 A flowchart of a dead time correction method according to an embodiment of this application is shown;

[0086] Figure 8 An exemplary block diagram of a dead time correction system according to an embodiment of this application is shown;

[0087] Figure 9 An exemplary block diagram of a dead time correction system according to an embodiment of this application is shown;

[0088] Figure 10 An exemplary block diagram of a dead time correction system according to an embodiment of this application is shown;

[0089] Figure 11 An exemplary processing flow is shown for determining the dead time correction factor using a dead time correction system according to embodiments of this application; and

[0090] Figure 12 An exemplary structural diagram of an electronic device that can implement the methods according to embodiments of this application is shown. Detailed Implementation

[0091] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0092] The term "comprising" and its variations as used herein signify open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "at least partially based on". The terms "one example embodiment" and "one embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc., may refer to different or the same objects.

[0093] As mentioned earlier, the detector inherently exhibits a dead-time effect during PET imaging. When the radioactivity is high and the count rate is large, the dead-time effect leads to the loss of a large number of coincidence events, resulting in a measured count rate that is far lower than the true count rate, which seriously affects the accuracy of quantitative image analysis and image quality.

[0094] Furthermore, the performance of PET detection systems in practical applications can be affected by a variety of factors. For example, components such as the photomultiplier tube and front-end electronics of the detector may be affected by factors such as ambient temperature and voltage fluctuations, causing the detection efficiency to drift. This instability results in non-ideal fluctuations in the actual measured count rate, rather than a smooth change with activity.

[0095] Some known dead-time correction schemes typically employ dead-time models for estimation. However, some known methods determine the dead-time constant and correction factor based on a single activity point or a single fit. These methods present the following technical problems in practical applications.

[0096] For example, some known schemes suffer from inaccuracies in their models under extremely high count rates, i.e., model accuracy issues. This could be explained, rather than limited, by neglecting the complex nonlinear characteristics of the dead-time effect, leading to insufficient calibration accuracy in the high-activity region.

[0097] For example, traditional single-fit dead-time correction methods mix detector fluctuation noise and dead-time nonlinear effects in the same fitting process for estimation. This can lead to unsatisfactory correction results and may even introduce new biases. As an explanation, and not a limitation, in traditional single-activity-point or single-fit dead-time correction methods, the model may be affected by two types of error sources simultaneously: on the one hand, due to insufficient consideration of count rate fluctuations caused by detector instability, measurement data containing noise and fluctuations are directly used for fitting; on the other hand, the fitting process simultaneously attempts to characterize the physical effects of dead time. As a result, noise interferes with the nonlinear fitting process, significantly amplifying the error and making the solved dead-time constant and correction factor unstable and unreliable.

[0098] To address this issue, embodiments of this application provide a dead-time correction scheme that effectively alleviates or overcomes the aforementioned problems. As an explanation, and not a limitation, the dead-time correction scheme provided in this application can employ multiple fitting processes to address different types of error sources in stages; more specifically, for example, by combining at least one linear fitting and a nonlinear dead-time model fitting, the processes of "noise suppression" and "physical modeling" are separated and progressively optimized, thereby obtaining a more stable and reliable dead-time correction factor. In some embodiments, this dead-time correction scheme can be a dead-time correction factor determination scheme.

[0099] Therefore, embodiments of this application propose a dead-time correction method and a corresponding dead-time correction system, as well as related image reconstruction methods and systems, electronic devices, storage media, and program products. These can at least partially implement the aforementioned dead-time correction scheme, at least partially alleviate or solve the aforementioned technical problems, and / or at least partially possess the aforementioned technical effects or advantages. In some embodiments, the dead-time correction method of this application can be implemented by the dead-time correction system of this application, but embodiments of this application are not limited thereto.

[0100] The dead time correction method of this application embodiment will be further described below with reference to the accompanying drawings.

[0101] like Figure 1 The diagram shows a flowchart of a dead time correction method according to an embodiment of this application, which may include steps S110 to S130:

[0102] S110: Acquire measured true event data and corresponding activity from multiple acquisitions.

[0103] In this embodiment of the application, in step S110, measurement data generated by the PET detection system during multiple acquisitions can be acquired. In some specific embodiments, multiple data acquisitions can be performed according to a predetermined acquisition plan, recording the coincidence event information received by the detector at each acquisition and obtaining the radioactivity value at that acquisition time. Preferably, the multiple acquisitions are a given number of recent multiple acquisitions.

[0104] In this application embodiment, "true event data" refers to data representing true matching events. In some embodiments, true event data is the true event count rate. However, it is conceivable that in different embodiments, true event data includes, but is not limited to, the true event count rate, the true event count, etc. In the following embodiments, the true event count rate is used as an example for illustration, but the embodiments of this application are not limited to this.

[0105] More specifically, such as Figure 2 As shown, step S110 may include steps S111 to S113:

[0106] Step S111: Obtain the number of timely matching events and the number of delayed matching events for each collection.

[0107] In this embodiment of the application, in step S111, the number of timely matching events and the number of delayed matching events are read from the data file generated each time.

[0108] In this embodiment, a timely coincidence event has the conventional meaning in the art, such as a pair of photon events received by a detector within a set coincidence time window. In some embodiments, a timely coincidence event includes true coincidence events, random coincidence events, and may also include scattering coincidence events. In this embodiment, a delayed coincidence event has the conventional meaning in the art, such as a coincidence event recorded by delaying the signal of one or more detector channels by a time much longer than the coincidence time window. Accordingly, delayed coincidence events are mainly composed of random coincidences and can be used to estimate the random coincidence component in timely coincidence events.

[0109] In some embodiments, timely conformance events and / or delayed conformance events can have any suitable data format, such as list mode or histogram mode.

[0110] Step S112: Determine the measurement true event data collected each time based on the number of timely and delayed matching events.

[0111] In this embodiment of the application, in step S112, the number of timely compliance events and the number of delayed compliance events obtained in step S111 can be used to calculate the measurement true event data collected each time.

[0112] In some embodiments, the difference between the number of timely coincidences and the number of delayed coincidences can be calculated. In some embodiments, this difference represents the net number of coincidences after deducting random coincidences, for example, primarily comprising true coincidences and scattered coincidences. In some embodiments, if scattering correction has been performed, this difference can directly reflect true coincidences.

[0113] In a specific example, when the true event data is the true event count rate, the above difference can be divided by the collection duration to obtain the true event count rate. The collection duration refers to the duration of each collection session. Specifically, it is expressed by the following formula:

[0114]

[0115] Among them, R Meas To measure real event data.

[0116] The above calculations yield the true event count rate for each acquisition, reflecting the frequency of true coincidence events actually measured by the detector under the conditions of that acquisition. As an explanation, this measured value is affected by the dead time effect and is lower than the true event count rate under ideal conditions.

[0117] Step S113: Obtain the activity level corresponding to each collection.

[0118] In this embodiment of the application, in step S113, the radioactivity value at each acquisition can be obtained. For explanation, radioactivity represents the radioactive intensity of a radioactive source, typically expressed in becquerels (Bq). In PET imaging, radioactivity decreases exponentially over time as the radionuclide decays. The method for obtaining the activity can be a known one, and will not be elaborated here.

[0119] S120: Perform multiple fitting processes on the measured true event data and corresponding activity, including at least one linear fitting and one nonlinear dead time model fitting, to obtain the dead time constant for the nonlinear dead time model.

[0120] In this embodiment, in step S120, multiple fitting processes can be performed on the measured true event data and corresponding activity obtained in step S110. In this embodiment, the multiple fitting process suppresses noise and fluctuations in the data through a staged fitting strategy, thereby more accurately characterizing the physical properties of the dead time effect.

[0121] In a further embodiment of this application, the multiple fitting process may include at least one linear fitting and a nonlinear dead-time model fitting. This combined fitting approach effectively separates the noise suppression and physical modeling processes.

[0122] In the embodiments of this application, the dead time constant for the nonlinear dead time model can be obtained through this multiple fitting process. By way of explanation and not limitation, the dead time constant is generally a physical parameter characterizing the intensity of the detector's dead time effect.

[0123] In some embodiments, the fitted true event data can also be obtained through this multiple fitting process. In these embodiments, the fitted true event data can also be referred to as theoretical true event data, which can be theoretical true event data obtained after at least one linear fitting process to reduce or eliminate random fluctuations.

[0124] In a further embodiment of this application, at least one linear fit may include a preliminary linear fit.

[0125] like Figure 3 In the illustrated embodiment, step S120 may include steps S121 to S123:

[0126] S121: Perform a first linear fit on the measured true event data and the corresponding activity to obtain the first fitted true event data.

[0127] In some embodiments, such as in step S121, a first linear fit can be performed on the measured true event data and corresponding activity obtained in step S110 to obtain first fitted true event data. In these embodiments, the first linear fit can also be referred to as the initial linear fit, and the first fitted true event data can also be referred to as the initial fitted true event data.

[0128] This first linear fit can serve as the first stage of multiple fitting processes. For illustrative purposes, and not as limiting, this first linear fit is used to smooth the measurement data, suppressing short-term random fluctuations and noise caused by detector instability. Thus, the first fitted true event data obtained through the first linear fit provides a more stable and reliable input for subsequent nonlinear fitting, avoiding interference from noise in the nonlinear fitting process.

[0129] Preferably, the first linear fit is a linear least squares fit. Specifically, as shown below... Figure 4 As shown, step S121 may include the following specific steps S1211 and S1212:

[0130] S1211: Linear fitting is performed on the measured true event data and corresponding activity from multiple acquisitions using the linear least squares method to obtain the first fitting slope.

[0131] In some embodiments, such as in step S1211, linear least squares can be used to linearly fit the multiple collected measurement true event data and corresponding activities using linear regression. By way of explanation and not limitation, linear least squares is a classic numerical fitting method whose basic principle is to find a straight line that minimizes the sum of squared distances between that line and all data points. In some embodiments, the measurement true event data R collected most recently can be selected. Meas The first linear fit is performed on the data and its corresponding activity A. In one example, the first-fit true event data model based on the first linear fit can be expressed as follows:

[0132] R True (A)=a·A

[0133] Among them, R True The output corresponds to the first fitted true event data, where A is the activity and a is the first fitted slope.

[0134] S1212: Calculate the first fitted true event data based on the first fitted slope and activity.

[0135] In some embodiments, such as in step S1212 above, the first fitting slope obtained in step S1211 can be used in combination with the activity to calculate the first fitted true event data collected each time.

[0136] In some embodiments, the activity used to calculate the fitted true event data is preferably the activity corresponding to the acquisition for which the dead time factor needs to be corrected, such as the activity corresponding to each acquisition (e.g., the most recent one). In other embodiments of this application, the activity in step S1212 can also be the activity of any one or more acquisitions (e.g., the activity of any one of the aforementioned multiple acquisitions or the activity of subsequent acquisitions). In other embodiments of this application, the activity in step S1212 can also be a statistically or otherwise processed activity. For example, in some application scenarios, statistical values ​​(e.g., mean, median, etc.) of the activities of some or all acquisitions involved in the fitting, or activity values ​​after other mathematical processing, can be used.

[0137] In some embodiments, a first fitted true event data is obtained through a first linear fitting process, which can provide a stable input with noise-free interference for subsequent nonlinear dead-time model fitting.

[0138] S122: Perform nonlinear dead-time model fitting on the first fitted true event data and the measured true event data, and solve for the dead-time constant.

[0139] In some embodiments, such as in step S122 above, the first fitted true event data obtained in step S121 and the measured true event data obtained in step S110 can be used to perform nonlinear dead-time model fitting to solve for the system's dead-time constant. In some embodiments, nonlinear fitting is the second stage of multiple fitting processes. For explanation and not limitation, unlike the conventional method described above which directly performs a single fitting on the noisy original measurement data, the input data used in this step is the first fitted true event data smoothed by the first linear fitting, which has removed most of the random fluctuations. Therefore, the nonlinear fitting process can focus on accurately characterizing the essence of the dead-time effect, thereby solving for a more robust and reliable dead-time constant.

[0140] Specifically, such as Figure 5 As shown, step S122 may include the following specific steps S1221 and S1222:

[0141] S1221: Substitute the first fitted true event data as the true value and the measured true event data as the observed value into the nonlinear dead-time model.

[0142] In some embodiments, such as in step S1221 above, the first fitted true event data can be used as the true value, and the measured true event data can be used as the observed value, and substituted into the nonlinear dead-time model for fitting preparation. As an explanation and not a limitation, in the physical model of the dead-time effect, there is a nonlinear relationship between the true count rate (the ideal count rate unaffected by dead time) and the actual measured count rate (the observed count rate affected by dead time).

[0143] S1222: The nonlinear dead-time model is fitted using the nonlinear least squares method to solve for the dead-time constant.

[0144] In some embodiments, such as in step S1222, the system may use the nonlinear least squares method to fit the nonlinear dead-time model and solve for the dead-time constant τ.

[0145] By way of explanation and not limitation, nonlinear least squares is a numerical optimization method for solving parameters of nonlinear equations. Its basic principle is to iteratively adjust parameter values ​​to minimize the sum of squared residuals between model predictions and actual observations. In this embodiment, the dead time constant τ is the parameter to be solved. The optimal value of τ is obtained by minimizing the error between the predictions and observations of the nonlinear dead time model.

[0146] In some embodiments, the above-described nonlinear dead-time model can be a paralysis model.

[0147] This application recognizes that the paralyzable dead time model is currently the most accurate dead time model for the physical characteristics of PET detectors. For illustrative purposes, and not restrictive, the paralyzable dead time model is more physically realistic: in a PET system, if the detector receives a new event during the dead time, the event causes the dead time to "restart," meaning the system will be "paralyzed" for a longer period; the paralyzable model provides a more accurate mathematical representation of this phenomenon, consistent with the detector's operating principles. Furthermore, the paralyzable model is more accurate: compared to the non-paralyzable model, it more accurately reflects actual saturation behavior at high count rates, thus avoiding underestimation of losses in high activity regions. In addition, the paralyzable model exhibits better parameter stability: its form can be solved using nonlinear least squares methods, resulting in stronger convergence and interpretability.

[0148] In a further embodiment of this application, the paralysis model may have the following expression:

[0149]

[0150] Among them, R True Corresponding to the true values ​​(in this case, the first fitted true event data), R Meas For the corresponding observations (in this case, measured true event data), τ is the dead time constant.

[0151] In another alternative embodiment, the aforementioned nonlinear dead-time model can be a non-paralyzing model. By way of explanation and not limitation, the non-paralyzing model is based on different physical assumptions: when the detector is in a dead-time state, newly arriving events are lost but the dead time is not prolonged; this model is suitable for certain types of detector systems or simplified dead-time correction schemes.

[0152] In these embodiments, the non-paralyzed model can have the following expression:

[0153] R Meas =R True / (1+R True ·τ),

[0154] Among them, R True Corresponding to the true values ​​(in this case, the first fitted true event data), R Meas For the corresponding observations (in this case, measured true event data), τ is the dead time constant.

[0155] In another alternative embodiment, the aforementioned nonlinear dead-time model can be a hybrid model comprising a weighted combination of a paralyzed model and a non-paralyzed model. Here, the hybrid model is a weighted combination of the paralyzed model and the non-paralyzed model to accommodate situations where some detector systems may simultaneously possess both dead-time characteristics.

[0156] In these embodiments, the hybrid model may have the following expression:

[0157]

[0158] Among them, R True Corresponding to the true value, R Meas For the observed values, τ is the dead time constant, λ1 is the first weight corresponding to the paralyzed model, and λ2 is the second weight corresponding to the non-paralyzed model.

[0159] In some embodiments, the weighting parameters λ1 and λ2 can be preset, for example, based on empirical values ​​or set as needed. In other embodiments of this application, the weighting parameters λ1 and λ2 can be used as parameters to be solved, thereby simultaneously optimizing the dead time constant τ and the weighting parameters λ1 and λ2 during nonlinear fitting, minimizing the sum of squared residuals between the predicted and observed values ​​of the mixture model. All of the above fall within the scope of this application.

[0160] The method in this embodiment of the application can solve for the dead time constant τ used in the dead time model through the above-described nonlinear fitting process. This allows the construction of a nonlinear dead time model for predicting real event data, and subsequently for calculating the dead time correction factor.

[0161] Accordingly, in some embodiments, predicted true event data can be obtained at least based on the determined dead time constant τ described above through a nonlinear dead time model. Furthermore, predicted true event data can be obtained based on fitted true event data through a nonlinear dead time model with the same dead time constant τ, i.e., the fitted true event data is used as input to the nonlinear dead time model.

[0162] Continue to refer to Figure 3 In some embodiments, the first fitted true event data can be used as input to a nonlinear dead-time model with a dead-time constant τ. That is, step S120 may further include the following step S123:

[0163] Step S123: Based on the first fitted true event data and the dead time constant, calculate the predicted true event data using a nonlinear dead time model.

[0164] In some embodiments, in step S123, the fitted true event data obtained in step S121 can be used to calculate the predicted true event data by substituting it into a nonlinear dead-time model, wherein the nonlinear dead-time model may have a dead-time constant obtained in step S122.

[0165] In one specific embodiment, the first fitted true event data, which is determined using the predicted true event data corresponding to each acquisition (as in step S1212), can be substituted into the predicted true event data corresponding to each acquisition calculated by a nonlinear dead-time model, which has a dead-time constant determined according to the aforementioned step S122 and its sub-steps.

[0166] In a specific example, taking the paralysis model as an example, the formula for predicting true event data is as follows:

[0167]

[0168] Among them, R True For the first fitted true event data (i.e., the input values ​​of the model), R Fit For predicting true event data (i.e., the predicted values ​​of the model), τ is the dead time constant obtained in step S122.

[0169] As mentioned earlier, nonlinear models can also be non-paralyzing models or hybrid models, which will not be elaborated here.

[0170] In some embodiments, a mapping relationship between fitted (theoretical) true event data and predicted measurements can be established in this step for subsequent calculation of the dead time correction factor.

[0171] In a further embodiment of this application, at least one linear fit in the multiple fitting process may further include a subsequent linear fit, i.e., a second linear fit after the nonlinear fit. Accordingly, referring to the reference Figure 3 and Figure 6 Step S120, which involves multiple fitting processes, may also include S124. However, those skilled in the art will understand that the subsequent second linear fitting (such as step S124) may be optional, as further described below, and the embodiments of this application also explicitly cover schemes without the subsequent second linear fitting (such as step S124).

[0172] In some embodiments, reference Figure 6 Step S120 may also include step S124:

[0173] S124: Perform a second linear fit on the predicted true event data and the corresponding activity to obtain the second fitted true event data.

[0174] In some embodiments, such as in step S124, a second linear fit can be performed on the predicted true event data obtained in step S123 and the activity obtained in step S110 to obtain second fitted true event data. In these embodiments, the second linear fit can also be referred to as optimized linear fit, and the second fitted true event data can also be referred to as optimized fitted true event data.

[0175] In some embodiments, the second linear fit can serve as a third stage of multiple fitting processes. By way of explanation and not limitation, in some cases, although predictions based on the dead-time model have been obtained through the fitting in the first two stages, these predictions may still contain local biases or residuals; by performing a second linear fit on the predicted true event data, these biases or residuals can be further eliminated, making the final theoretical model smoother, more consistent, and more robust across the full activity range.

[0176] In some embodiments, similar to step S121, the second linear fit is a linear least squares fit. Specifically, as... Figure 7 As shown, step S124 may include the following specific steps S1241 and S1242:

[0177] S1241: The second fitting slope is obtained by linearly fitting the predicted true event data and corresponding activity of multiple collections using the linear least squares method.

[0178] In some embodiments, such as in step S1241, linear least squares method can be used to linearly fit the predicted true event data and corresponding activity collected multiple times.

[0179] In some embodiments, similar to step S1211, the predicted true event data can be used as the dependent variable and the activity as the independent variable to establish a linear relationship model, and the second fitting slope can be solved by the linear least squares method.

[0180] In some embodiments, the second fitted true event data model based on the second linear fit can be as follows:

[0181] R True_Fit (A)=a fit ·A,

[0182] Among them, R True_Fit The output corresponds to the second fitted true event data, where A is the activity and a is the value of the activity. fit The second fitted slope.

[0183] S1242: Calculate the second fitted true event data based on the second fitted slope and activity.

[0184] In some embodiments, such as in step S1242, the system may use the second fitting slope obtained in step S1241 and combine it with the activity to calculate the second fitted true event data collected each time.

[0185] In some embodiments, similar to step S1212, the activity used to calculate the fitted true event data is preferably the activity corresponding to the acquisition for which the dead time factor needs to be corrected, such as the activity corresponding to each acquisition (e.g., the most recent one). In other embodiments, the activity mentioned in step S1242 can also be the activity of any one or more acquisitions (e.g., the activity of any one of the aforementioned multiple acquisitions or the activity of subsequent acquisitions). In other embodiments, the activity mentioned in step S1242 can also be a statistically or otherwise processed activity. For example, in some application scenarios, statistical values ​​(e.g., mean, median, etc.) of the activities of some or all acquisitions involved in the fitting, or activity values ​​after other mathematical processing, can be used.

[0186] S130: Determine the dead time correction factor based at least on the predicted true event data.

[0187] In this embodiment of the application, in step S130, the dead time correction factor can be determined at least based on the predicted true event data determined in step S120, and more preferably based on the fitted true event data and the predicted true event data.

[0188] In a further embodiment, the dead time correction factor can be determined based on the second fitted true event data determined in step S124 and the aforementioned predicted true event data. In these embodiments, the fitted true event data used to determine the dead time correction factor is the second fitted true event, but other data, such as other fitted data, such as the first fitted true event data, are also conceivable.

[0189] In some embodiments, step S130 may include the following specific step S130' (not shown):

[0190] S130': Determine the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

[0191] In some embodiments, such as in step S130', the dead time correction factor can be determined based on the ratio of the second fitted true event data obtained in step S124 to the predicted true event data obtained in step S123.

[0192] In a specific example, the formula for calculating the dead time correction factor CF_DT can be:

[0193]

[0194] Among them, R True_Fit For the second set of true event data (i.e., optimizing the theoretical values), R Fit To predict real event data (i.e., predictions from models such as paralysis models), CF DT This is the dead time correction factor.

[0195] As an explanation and not a limitation, the correction factor determined by the above formula can represent the ratio between the true count rate (as the true value under ideal conditions) and the predicted measurement value after considering dead time (as the theoretical measurement value after considering dead time), thus quantifying the proportion of count loss caused by the dead time effect. Therefore, in the subsequent image reconstruction process, applying this correction factor to the actual acquired coincidence event data can recover the counts lost due to the dead time effect, thereby obtaining a quantitatively accurate PET image.

[0196] In this embodiment, since the second fitted true event data is the final theoretical value obtained through three-stage fitting optimization, the correction factor calculated using this data has high accuracy and robustness. This correction factor remains smooth and consistent across the entire activity range, effectively correcting for dead-time effects under different activity conditions. Therefore, by applying this correction factor, the counting information lost due to dead-time effects can be accurately recovered, making the reconstructed PET images more accurate in quantitative analysis and significantly improving image quality.

[0197] In the above embodiments, the system can obtain the second fitted true event data through a three-stage fitting process (first linear fitting, nonlinear fitting, and second linear fitting), and calculate the dead time correction factor based on the data.

[0198] As mentioned above, the aforementioned embodiment including step S124 and using the second fitted true event data in step S130' is an optional embodiment. This optional embodiment, through three-stage fitting processing, can further eliminate fitting residuals and small deviations, making the final theoretical model smoother and more consistent across the full activity range. The resulting dead time correction factor has high accuracy and strong robustness, and is particularly suitable for application scenarios with high activity conditions or large fluctuations in detector performance, providing excellent image quantitative accuracy.

[0199] In another embodiment, step S124 is optional. That is, the method provided in the relevant alternative embodiments may not perform a second linear fitting, but instead directly use the results of the first linear fitting and nonlinear fitting to calculate the dead-time correction factor. This embodiment is suitable for application scenarios with high requirements for computational efficiency or where the data quality is already good, and can still obtain a dead-time correction factor with satisfactory accuracy and robustness. Several embodiments of this application explicitly cover this embodiment.

[0200] In these alternative embodiments, the features and execution process of steps S110 and its sub-steps, S121, S122, S123 and their sub-steps can be referred to the foregoing embodiments, and will not be repeated here. The difference is that in these alternative embodiments, step S124 is not executed.

[0201] Accordingly, in a further embodiment, step S130 may include step S130 (not shown):

[0202] Step S130”: Determine the dead time correction factor based on the ratio of the first fitted true event data to the predicted true event data.

[0203] In step S130", the dead time correction factor can be determined based on the ratio of the first fitted true event data obtained in step S121 to the predicted true event data obtained in step S123.

[0204] In a specific example, for acquisition, the formula for calculating the dead time correction factor CF_DT can be:

[0205]

[0206] Among them, R True For the first fitted true event data (i.e., the initial theoretical value), R Fit To predict real event data (i.e., the predicted values ​​of the aforementioned models such as the paralysis model), CF DT This is the dead time correction factor.

[0207] For other features, execution and details involved in step S130", please refer to the relevant records of steps S130 and S130' above, which will not be repeated here.

[0208] Therefore, the dead time correction method of this application embodiment can have at least some of the following technical effects: In the dead time correction method of this application embodiment, in the first stage, the measurement data can be smoothed by linear fitting to suppress short-term random fluctuations and noise, extract the overall trend of the data, and obtain preliminary fitting true event data, thereby providing a more stable input data basis with suppressed noise for subsequent nonlinear model (modeling) fitting. In the second stage, based on the smoothed noise, the nonlinear dead time model fitting focuses on characterizing the true physical characteristics of dead time to obtain a high-precision dead time constant; at this time, since the input data has been smoothed, the nonlinear fitting process can more accurately reflect the essence of the dead time effect and avoid fitting distortion caused by noise interference, such as, but not limited to, avoiding the nonlinear model (modeling) fitting from getting trapped in local optima or divergence.

[0209] Furthermore, in some embodiments, the dead-time correction method can also have further technical effects: in an optional third stage, the results of the nonlinear fitting can be optimized by linear fitting again, which further eliminates fitting residuals and small biases, making the final theoretical model smoother, more consistent and reliable across the full activity range.

[0210] In summary, by employing a multi-stage fitting strategy of "stabilization first, then precision," and more preferably "stabilization first, then precision, then smoothing optimization," this application constructs a dead-time correction process of "smoothing-modeling," and more preferably "smoothing-modeling-smoothing again." Through an iterative optimization process of "linear fitting → nonlinear fitting," and preferably "linear fitting → nonlinear fitting → linear fitting," the random fluctuations in the count rate introduced by detector instability and other reasons are smoothed and suppressed layer by layer. This makes the finally solved dead-time constant and correction factor more accurate, reliable, and stable, avoiding the error amplification problem caused by directly fitting fluctuating data in traditional methods. Furthermore, the method provided by this application is insensitive to data fluctuations and noise, and the calculation results are more stable and reliable, significantly improving the robustness of the correction algorithm in real unstable environments. In addition, the method provided by this application has a clear process flow and can be easily integrated into existing PET image reconstruction related processing flows without modifying the hardware, thereby significantly improving the accuracy and robustness of dead-time correction, effectively addressing count rate fluctuations caused by detector instability, and especially improving the image quantitative accuracy and imaging performance of the PET system under high activity conditions.

[0211] In this application embodiment, an image reconstruction method is also provided, which applies the dead time correction factor determined according to the dead time correction method of this application embodiment to the PET image reconstruction process to achieve accurate correction of the dead time effect. For specific implementation methods of the dead time correction method, please refer to the detailed description of the foregoing embodiments, which will not be repeated here.

[0212] In some embodiments, the image reconstruction method may include the following steps (not identified): during the image reconstruction process, performing dead-time effect correction on coincident events using a dead-time correction factor determined according to the dead-time correction method described in the foregoing embodiments.

[0213] In some embodiments, during the PET image reconstruction process, the dead time correction factor calculated in the foregoing embodiments is applied to correct the acquired coincidence event data.

[0214] In some examples, for each acquisition or each time period of coincident event data, the PET device or its image reconstruction system can read the corresponding dead time correction factor (CF). DT The dead time correction factor CF DT It can be determined according to the multiple fitting process described in the foregoing embodiments.

[0215] As an explanation and not a limitation, the dead time correction factor CF DTThe proportion of count loss caused by the dead-time effect under the current acquisition conditions is quantified. In some examples, the PET device or its image reconstruction system can multiply the actual acquired coincidence event data by the dead-time correction factor to obtain the corrected coincidence event data. Through the dead-time correction process, the coincidence event counts lost due to the dead-time effect are corrected.

[0216] In different embodiments of this application, the dead time correction factor CF is used. DT Implementing dead-time correction during reconstruction can take various forms, including but not limited to any suitable existing or new schemes, and can be implemented independently or in combination with the reconstruction algorithm and / or independently or in combination with other corrections. In some examples, in image reconstruction, a dead-time correction factor can be applied to the raw list pattern data or sine curve data during the data preprocessing stage, i.e., before executing the image reconstruction algorithm, to generate corrected data, which is then input into the "pure" image reconstruction algorithm to obtain the reconstructed image. In other examples, this dead-time correction factor CF... DT It can be integrated into image reconstruction algorithms, so that during the iterative process of the image reconstruction algorithm, when updating the image in each iteration, the dead time correction factor CF... DT As part of the correction factor matrix, it is applied to image reconstruction along with other correction factors (such as attenuation correction, scattering correction, normalization correction, etc.). Furthermore, it utilizes the dead-time correction factor CF... DT Dead time correction during reconstruction can be performed on the PET device or its detector, either as a whole or in parts. All of the above variations fall within the scope of the invention.

[0217] In the embodiments of this application, any suitable image reconstruction algorithm can be used, such as, but not limited to, one or more of the following: Filtered Back Projection (FBP) algorithm, Iterative Reconstruction Algorithm (MLEM) algorithm, Ordered Subset Expectation Maximization (OSEM) algorithm, etc.

[0218] By applying the high-precision, robust dead-time correction factor determined by the dead-time correction method in the aforementioned embodiments to the image reconstruction process, the quantitative accuracy of PET images can be significantly improved. Especially under high activity conditions, traditional methods, due to insufficient dead-time correction accuracy, may lead to a systematically low activity value in the image, affecting lesion detection and quantitative analysis. Since the dead-time correction factor in the method of this application embodiment is calculated through a multiple fitting processing strategy, the influence of data fluctuations can be effectively suppressed, thus significantly improving the quality of the reconstructed image after correction.

[0219] In the embodiments of this application, reference is made to Figure 8Furthermore, a dead-time correction system 800 is provided, which may include a data acquisition module 810, a fitting module 820, and a correction factor determination module 830.

[0220] The data acquisition module 810 is configured to acquire multiple collections of measurement true event data and corresponding activity;

[0221] The fitting module 820 is configured to perform multiple fitting processes on the measured true event data and the corresponding activity. The multiple fitting processes include at least one linear fitting and one nonlinear dead time model fitting to obtain the dead time constant and the predicted true event data. The predicted true event data is obtained by the nonlinear dead time model with the dead time constant.

[0222] The correction factor determination module 830 is configured to determine a dead time correction factor based at least on the predicted true event data.

[0223] In the embodiments of this application, reference is made to Figure 9 The data acquisition module 810 may include a coincidence event acquisition unit 811, a preprocessing unit 812, and an activity acquisition unit 813.

[0224] The matching event acquisition unit 811 is configured to acquire the number of timely matching events and the number of delayed matching events for each acquisition;

[0225] The preprocessing unit 812 is configured to determine the measurement true event data collected each time based on the number of timely matching events and the number of delayed matching events;

[0226] The activity acquisition unit 813 is configured to acquire the activity corresponding to each acquisition.

[0227] In some embodiments, reference Figure 10 The fitting module 820 may include a first linear fitting unit 821, a nonlinear fitting unit 822, and a prediction unit 823.

[0228] The first linear fitting unit 821 is configured to perform a first linear fitting on the measured true event data and the corresponding activity to obtain first fitted true event data; and

[0229] The nonlinear fitting unit 822 is configured to perform the nonlinear dead-time model fitting on the first fitted true event data and the measured true event data to solve for the dead-time constant.

[0230] The prediction unit 823 is configured to calculate the predicted true event data based on the first fitted true event data and the dead time constant using the nonlinear dead time model.

[0231] In some embodiments, continue to refer to Figure 10 The fitting module 820 may further include a second linear fitting unit 824. The second linear fitting unit 824 is configured to perform a second linear fitting on the predicted true event data and the corresponding activity to obtain second fitted true event data; wherein the correction factor determination module 830 is further configured to determine the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

[0232] In some embodiments, the correction factor determination module 830 is further configured to determine the dead time correction factor based on the ratio of the first fitted true event data to the predicted true event data.

[0233] In some embodiments, the nonlinear dead-time model is a paralyzed model, a non-paralyzed model, or a hybrid model comprising a weighted combination of a paralyzed model and a non-paralyzed model.

[0234] In some embodiments, the true event data is the true event count rate.

[0235] In this application embodiment, an image reconstruction system is also provided, which may include the dead time correction system 800 according to the embodiment of this application and an image reconstruction module, wherein the image reconstruction module is configured to perform dead time effect correction on coincident events using the dead time correction factor determined by the dead time correction system during the image reconstruction process.

[0236] In some embodiments, reference Figure 11 An exemplary processing flow is shown for determining a dead time correction factor using a dead time correction system according to embodiments of this application. In this embodiment, the dead time correction system can be used to implement the steps and sub-steps of the dead time correction method described in any of the foregoing embodiments, and the determined dead time correction factor can be used in an image reconstruction system according to embodiments of this application to perform dead time effect correction on coincident events.

[0237] In this application embodiment, a PET device is also provided, which may include the dead time correction system according to the embodiment of this application; or the image reconstruction system according to the embodiment of this application.

[0238] In this application embodiment, an electronic device is also provided, which may include a processor and a memory, wherein the memory stores computer program instructions, and the processor, when executing the computer program instructions, implements the method described in the present application embodiment.

[0239] Figure 12A schematic diagram of an exemplary electronic device 1200 that can implement the methods of embodiments of this application is shown. In some embodiments, it may include more or fewer electronic devices than shown. In some embodiments, it may be implemented using a single or multiple electronic devices. In some embodiments, it may be implemented using cloud-based or distributed electronic devices.

[0240] like Figure 12 As shown, the electronic device 1200 includes a processor 1201, which can perform various appropriate operations and processes based on programs and / or data stored in read-only memory (ROM) 1202 or programs and / or data loaded from storage portion 1208 into random access memory (RAM) 1203. The processor 1201 can be a single-core or multi-core processor, or may include multiple processors. In some embodiments, the processor 1201 may include a general-purpose main processor (such as a CPU) and one or more special coprocessors, such as a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), or other general-purpose or application-specific integrated circuits. Various programs and data required for the operation of the electronic device 1200 are also stored in RAM 1203. The processor 1201, ROM 1202, and RAM 1203 are interconnected via a bus 1204. An input / output (I / O) interface 1205 is also connected to the bus 1204.

[0241] The processor and memory described above are used together to execute a program stored in the memory. When the program is executed by a computer, it can implement the steps or functions of the methods described in the above embodiments.

[0242] The following components are connected to I / O interface 1205: an input section 1206 including a keyboard, mouse, etc.; an output section 1207 including a display and speakers, etc.; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN card and a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to I / O interface 1205 as needed. Removable media 1211, such as a hard disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1210 as needed so that computer programs read from them can be installed into storage section 1208 as needed.

[0243] Figure 12 The electronic device shown is merely illustrative, but the electronic device according to embodiments of this application may include more than [other components]. Figure 12 The electronic device shown has more or fewer components or has more or fewer components than the one shown. Figure 12 The embodiments shown have the same, partially the same, or different architectures.

[0244] In some embodiments, the electronic device may be combined with various components to obtain methods, apparatus and systems that have the advantages of the present invention.

[0245] Although not shown, some embodiments also provide a computer-readable storage medium storing a computer program configured to be executed to perform the methods of any of the embodiments of this application. The computer program includes various program modules / units constituting the apparatus according to the embodiments of this application, and when executed, the computer program comprised of the various program modules / units can perform the functions corresponding to the various steps in the methods described in the above embodiments. The computer program can also run on electronic devices as described in the embodiments of this application.

[0246] Although not shown, some embodiments also provide a program product comprising a computer program configured to be run to perform the methods of any of the embodiments of this application.

[0247] The storage medium in embodiments of this application includes non-volatile and / or volatile articles that can store information by any method or technology. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0248] Those skilled in the art will understand that the embodiments of this specification can be implemented in various forms, such as methods, systems, or computer program products. Therefore, those skilled in the art will realize that the functional modules / units or controllers and related method steps described in the above embodiments can be implemented in software, hardware, or a combination of software and hardware.

[0249] Unless explicitly stated otherwise, the actions or steps of the methods and procedures described in the embodiments of this application do not necessarily have to be performed in a specific order and can still achieve the desired results. In some implementations, multitasking and parallel processing are also possible or may be advantageous.

[0250] This document describes several embodiments, but for the sake of brevity, the descriptions of the embodiments are not exhaustive, and identical or similar features or parts between the embodiments may be omitted. In this document, "one embodiment," "some embodiments," "example," "specific example," or "some examples" refers to at least one embodiment or example applicable to this application, but not all embodiments. The above terms do not necessarily mean referring to the same embodiment or example. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of the different embodiments or examples.

[0251] The exemplary systems and methods of this application have been specifically shown and described with reference to the above embodiments, which are merely examples of the best mode for implementing the systems and methods. Those skilled in the art will understand that various changes can be made to the embodiments of the systems and methods described herein without departing from the spirit and scope of the invention as defined in the appended claims when implementing the systems and / or methods.

Claims

1. A dead time correction method, characterized in that, include: Acquire multiple sets of measured true event data and corresponding activity; The measured true event data and the corresponding activity are subjected to multiple fitting processes, including at least one linear fitting and one nonlinear dead time model fitting, to obtain the dead time constant and the predicted true event data, wherein the predicted true event data is obtained by a nonlinear dead time model with the dead time constant. The dead time correction factor is determined based at least on the predicted true event data.

2. The method according to claim 1, characterized in that, The process of performing multiple fitting operations on the measured true event data and the corresponding activity includes: Perform a first linear fit on the measured true event data and the corresponding activity to obtain first fitted true event data; and The nonlinear dead-time model is fitted to the first fitted true event data and the measured true event data to solve for the dead-time constant. Based on the first fitted true event data and the dead time constant, the predicted true event data is calculated using the nonlinear dead time model.

3. The method according to claim 2, characterized in that, The step of performing a first linear fit on the measured true event data and the corresponding activity to obtain the first fitted true event data includes: The first fitting slope is obtained by linearly fitting the multiple collected measurement true event data and the corresponding activity using the linear least squares method. Based on the first fitting slope and activity, the first fitted true event data is calculated.

4. The method according to claim 2, characterized in that, The step of performing the nonlinear dead-time model fitting on the first fitted true event data and the measured true event data to solve for the dead-time constant includes: The first fitted true event data is used as the true value, and the measured true event data is used as the observed value and substituted into the nonlinear dead time model. The nonlinear dead-time model is fitted using the nonlinear least squares method to solve for the dead-time constant.

5. The method according to claim 4, characterized in that, The nonlinear dead-time model is a paralysis model.

6. The method according to claim 5, characterized in that, The paralysis model has the following expression: Among them, R True Corresponding to the true value, R Meas For the observed value, τ is the dead time constant.

7. The method according to claim 4, characterized in that, The nonlinear dead-time model is a non-paralyzed model or a hybrid model that includes a weighted combination of paralyzed and non-paralyzed models. The non-paralyzed model has the following expression: R Meas =R True / (1+R True ·t), Among them, R True Corresponding to the true value, R Meas For the observed value, τ is the dead time constant; The hybrid model has the following expression: Among them, R True Corresponding to the true value, R Meas For the observed values, τ is the dead time constant, and λ1 and λ2 are weights.

8. The method according to claim 2, characterized in that, The process of performing multiple fitting operations on the measured true event data and the corresponding activity also includes: A second linear fit is performed on the predicted true event data and the corresponding activity to obtain the second fitted true event data; The step of determining the dead time correction factor based at least on the predicted true event data includes: determining the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

9. The method according to claim 8, characterized in that, The step of performing a second linear fit on the predicted true event data and the corresponding activity to obtain the second fitted true event data includes: The second fitting slope is obtained by linearly fitting the predicted true event data collected multiple times and the corresponding activity using the linear least squares method. Based on the second fitting slope and activity, the second fitted true event data are calculated.

10. The method according to claim 2, characterized in that, The determination of the dead time correction factor, based at least on the predicted true event data, includes: The dead time correction factor is determined based on the ratio of the first fitted true event data to the predicted true event data.

11. The method according to any one of claims 1 to 10, characterized in that, The true event data is the true event count rate.

12. The method according to any one of claims 1 to 10, characterized in that, The acquisition of multiple collected measurement true event data and corresponding activity includes: Obtain the number of timely and delayed matching events for each data collection. The number of timely compliant events and the number of delayed compliant events are used to determine the measurement true event data collected each time; Obtain the activity level corresponding to each data collection.

13. An image reconstruction method, characterized in that, include: During image reconstruction, the dead time correction factor determined by the dead time correction method according to any one of claims 1 to 12 is used to correct the dead time effect of the matching events.

14. A dead-time correction system, characterized in that, include: The data acquisition module is configured to acquire multiple sets of measurement true event data and corresponding activity. The fitting module is configured to perform multiple fitting processes on the measured true event data and the corresponding activity. The multiple fitting processes include at least one linear fitting and one nonlinear dead time model fitting to obtain the dead time constant and the predicted true event data. The predicted true event data is obtained by the nonlinear dead time model with the dead time constant. The correction factor determination module is configured to determine the dead time correction factor based at least on the predicted true event data.

15. The system according to claim 14, characterized in that, The fitting module includes: A first linear fitting unit is configured to perform a first linear fitting on the measured true event data and the corresponding activity to obtain first fitted true event data; and The nonlinear fitting unit is configured to perform the nonlinear dead-time model fitting on the first fitted true event data and the measured true event data, and solve for the dead-time constant. The prediction unit is configured to calculate the predicted true event data based on the first fitted true event data and the dead time constant using the nonlinear dead time model.

16. The system according to claim 15, characterized in that, The fitting module further includes: The second linear fitting unit is configured to perform a second linear fitting on the predicted true event data and the corresponding activity to obtain the second fitted true event data. The correction factor determination module is further configured to determine the dead time correction factor based on the ratio of the second fitted true event data to the predicted true event data.

17. The system according to claim 15, characterized in that, The correction factor determination module is further configured to determine the dead time correction factor based on the ratio of the first fitted true event data to the predicted true event data.

18. The system according to any one of claims 14 to 17, characterized in that, The nonlinear dead-time model is a paralyzed model, a non-paralyzed model, or a hybrid model that includes a weighted combination of paralyzed and non-paralyzed models.

19. The system according to any one of claims 14 to 17, characterized in that, The true event data is the true event count rate.

20. The system according to any one of claims 14 to 17, characterized in that, The data acquisition module includes: The event acquisition unit is configured to acquire the number of timely matching events and the number of delayed matching events for each acquisition. The preprocessing unit is configured to determine the measured true event data for each acquisition based on the number of timely matching events and the number of delayed matching events; The activity acquisition unit is configured to acquire the activity corresponding to each data collection.

21. An image reconstruction system, characterized in that, include: The dead time correction system according to any one of claims 14 to 20; and The image reconstruction module is configured to perform dead-time effect correction on coincident events during the image reconstruction process using the dead-time correction factor determined by the dead-time correction system.

22. A PET device, characterized in that, include: The dead time correction system according to any one of claims 14 to 20; or The image reconstruction system according to claim 21.

23. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores computer program instructions, and the processor, when executing the computer program instructions, implements the method as described in any one of claims 1 to 13.

24. A computer storage medium, characterized in that, The system stores computer program instructions, wherein when the computer program instructions are executed by a processor, the method as described in any one of claims 1 to 13 is implemented.

Images