A dynamic gain compensation method for a silicon photomultiplier

Abstract

The application relates to a dynamic gain compensation method of a silicon photomultiplier, which acquires gain compensation information in a first state of all unit detectors of a PET system and corrects online in a single event of a second state of all unit detectors, and the method comprises the following steps: an FPGA judges whether the unit detector is in the first state, if yes, gain compensation parameters of gain compensation information are acquired according to a gain acquisition mode, or gain compensation parameters in known gain compensation information are updated in real time according to the known gain compensation information; when the unit detector is in the second state, the FPGA corrects data in the single event of the second state according to the latest gain compensation parameters, so that detection data used for reconstructing a PET image is acquired based on the corrected data. The above method solves the problem that real-time temperature compensation occupies system resources, reduces system dead time and maintains the stability of the gain of a silicon photomultiplier of a PET system detector.

Description

Technical Field

[0001] This invention relates to the field of nuclear radiation detector technology, and in particular to a dynamic gain compensation method for silicon photomultiplier tubes. Background Technology

[0002] Silicon photomultiplier tubes (SPVs) are being widely studied for use in positron emission tomography (PET). Compared to PET systems using traditional photomultiplier tubes (PMTs), SPVs offer lower operating voltages and are less sensitive to magnetic fields, allowing them to be used in conjunction with MRI for better human imaging. However, SPV performance is temperature-sensitive; its gain, breakdown voltage, and dark count are all temperature-dependent, limiting its applications. Under the same temperature fluctuations, the gain variation range of SPVs is greater than that of traditional PMTs. PET systems use a fixed energy window to filter all acquired events, eliminating valid events and blocking invalid ones. Gain fluctuations in SPVs effectively cause energy window fluctuations, resulting in some valid events being blocked and some invalid events being included, thus reducing the sensitivity and stability of the PET system.

[0003] Typical temperature compensation for silicon photomultiplier tubes (SPVs) involves using a temperature sensor to detect changes in the ambient temperature. The signal is then processed through an analog-to-digital converter (ADC), microcontroller, digital-to-analog converter (DAC), and switching boost circuit to adjust the SPV's bias voltage in real-time, thereby stabilizing its gain. While this method is sufficient for general applications, in the PET (Polymer Electron Device) field, where high-activity data acquisition and high-speed transmission require real-time temperature monitoring and adjustment, the process of adjusting the voltage to change the gain consumes significant data transmission and processing time. This results in a dead time for the PET system, reducing its sensitivity. Therefore, a dynamic gain compensation method for SPVs is urgently needed to improve the sensitivity of PET systems. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a dynamic gain compensation method for silicon photomultiplier tubes.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0008] In a first aspect, embodiments of the present invention provide a dynamic gain compensation method for silicon photomultiplier tubes. The dynamic gain compensation method acquires gain compensation information in the first state of all unit detectors in a PET system and corrects it online during a single-stage event in the second state of all unit detectors. The method includes:

[0009] The FPGA determines whether the unit detector is in the first state. If so, it obtains the gain compensation parameters of the gain compensation information according to the gain acquisition mode, or updates the gain compensation parameters in the known gain compensation information in real time according to the known gain compensation information.

[0010] When the unit detector is in the second state, the FPGA corrects the data in the single-lift event of the second state according to the latest gain compensation parameters, so as to obtain detection data for reconstructing PET images based on the corrected data;

[0011] The first state is the non-acquisition state of the unit detector, the second state is the acquisition state of the unit detector, and the FPGA is the FPGA to which all unit detectors in the PET system belong.

[0012] Optionally, the FPGA determines whether the unit detector is in the first state by:

[0013] The FPGA determines whether it has received a data acquisition command. If it has, it determines that the unit detector is in the second state; otherwise, it is in the first state.

[0014] Optionally, the gain compensation parameters for obtaining gain compensation information according to the gain acquisition mode include:

[0015] A01. Obtain the first curve of the operating temperature and gain of each unit detector by calibration measurement;

[0016] A02. Obtain the second curve showing the change of the operating temperature of each unit detector over time;

[0017] A03. Perform an approximate linear fit on the second curve of all unit detectors to obtain intermediate parameters for calculating the gain compensation parameters.

[0018] A04. Based on the first curve and the intermediate parameters, obtain the gain compensation parameters that show the change of gain over time.

[0019] Optionally, A01, the first curve of the operating temperature versus gain of each detector unit is obtained by calibration measurement, including:

[0020] A011. The operating temperature of each unit detector is read using the temperature sensor of each unit detector;

[0021] A012. With the Ge barrel placed in the middle of all unit detectors, record the energy peak at 511 keV in the single crystal energy spectrum within one cycle to obtain the operating temperature and peak position variation point of each single crystal strip.

[0022] A013. The average value of all single crystal peaks of the unit detector is used to characterize the peak position of the unit detector, and the first curve of the change of operating temperature and gain of the unit detector is obtained.

[0023] Using the first function G 增益 (Temp) = a 增益 *Temp+b 增益 ;Formula (m1).

[0024] Optionally, the linear relationship of the second curve in A02 can be expressed as:

[0025] Temp 降 =a 降 t+b 降 Formula (m2)

[0026] Temp 升 =a 升 t+b 升 ;

[0027] The second curve in A02 was obtained using an online real-time calibration method. During the online calibration process, the initial period rise time T needs to be input. 0上升 and descent time T 0下降 Maximum initial temperature (Temp) 0max and minimum initial temperature Temp 0min ,

[0028] FPGA based on the first cycle range (T) 0上升 +T 0下降 Get the maximum temperature (Temp) within the first cycle. 1max and the time t when the temperature reaches its maximum 1max And the minimum temperature of the first cycle, Temp. 1min and t 1min ; ΔT=t 1min -t 1max If the time difference is greater than 0, then ΔT is used as the new descending cycle. 1下降 This means that if the difference ΔT is less than 0, then ΔT is used as the new downward cycle. 1上升 express;

[0029] t 1max The time is taken as the starting point of the new cycle, and a new minimum value Temp is determined within the extended time of the selected new cycle. 2min and time t 2min and the maximum value Temp2max and time t 2max T 2下降

[0030] =t 2min -t 1max T 2上升 =t 2max -t 2min , t 2max Using the time point as the starting point of a new cycle, five sets of data for the second curve are obtained by iterating through the data in sequence.

[0031] The data for all five sets of the second curves include: the period start time t. star descent time T 下降 and rise time T 上升 Maximum temperature within the cycle: Temp max and minimum value Temp min .

[0032] Optionally, A03, an approximate linear fit is performed on the second curves of all unit detectors to obtain intermediate parameters for calculating the gain compensation parameters, including:

[0033] t star =t max(上组) The starting time is the maximum temperature (Temp) of the last complete cycle at the corresponding scale time. max The corresponding time t max ;

[0034] T 下降 =t min -t star ;

[0035] T 上升 =t max -t min ;

[0036] The remaining 4 sets of values ​​T 下降 T 上升 Temp max and Temp min Each time a new value is obtained, it is updated by adding 1 / 16 of the original value to 15 / 16. All four sets of values ​​are averaged using this method.

[0037] That is: Current value = New value * 1 / 16 + Old value * 15 / 16;

[0038] Get intermediate parameter a 降 b 降 a 升 b 升 ;

[0039] a 降 =(Tempmin -Temp max ) / T 下降 ;

[0040] b 降 =Temp max ;

[0041] a 升 =(Temp max -Temp min ) / T 上升 ;

[0042] b 升 =Temp min .

[0043] Optionally, in A04, based on the first curve and the intermediate parameters, a gain compensation parameter that shows the gain changing over time is obtained, including:

[0044] According to formula (m1), formula (m2), t = t 当前 -t star and intermediate parameter a 降 b 降 a 升 b 升 ; and the empirical formula for gain (m3)

[0045] G 标准 =G 实测 R 补偿 ;R 补偿 =G 标准 / G 增益 ;Formula (m3)

[0046] G 标准 It is the gain value of the unit detector at the expected average temperature, which is a known value. 标准 This is a definite value, representing the expected average operating temperature of the unit detector. Based on formula (m1), the following formula (m4) is derived:

[0047] G 标准 =a 增益 Temp 标准 +b 增益 (m4)

[0048] R 补偿 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 Temp+b 增益 );

[0049] R 补偿The variable is temperature (Temp), and Temp is a piecewise function divided into Temp... 降 and Temp 升 Combining this with formula (m2), we obtain R. 补偿降 R 补偿升 ;

[0050] R 补偿降 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 (a 降 t+b 降 )+b 增益 );

[0051] R 补偿升 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 (a 升 t+b 升 )+b 增益 );

[0052] The above R 补偿降 R 补偿升 Except for t, all coefficients are known values. Based on the known value of a... 降 b 降 a 升 b 升 , t = t 当前 -t star ;

[0053] The formula (m5) for obtaining the gain compensation parameter as part of the gain compensation information can be obtained:

[0054] The formula (m5) is:

[0055]

[0056]

[0057] t 当前 To obtain the data in real time;

[0058] R 补偿降 and R 补偿升 Judgment conditions:

[0059] Remainder = (t) 当前 -t star ) / (T 下降 +T 上升 The remainder of ).

[0060] Remainder ≤ T 下降For the descent period, use R in formula (m5). 补偿降 ;

[0061] Remainder > T 下降 For the rising period, use R in formula (m5). 补偿升 .

[0062] Optionally, the FPGA corrects the data in the single-launch event of the second state according to the latest gain compensation parameters, including:

[0063] Formula (m5) is written into the front-end FPGA as a correction formula to obtain the latest gain compensation parameters. The FPGA corrects the data in the single-instance event of the second state based on the latest gain compensation parameters to obtain the original data as probe data.

[0064] Alternatively, formula (m5) can be written into the front-end FPGA as a correction formula to obtain the latest gain compensation parameters. The FPGA automatically corrects the gain according to the latest gain compensation parameters. The single-instance data after gain correction is then filtered by the energy window and time coincidence to obtain the raw data as the probe data.

[0065] In a second aspect, embodiments of the present invention provide a detection device, comprising: a unit detector based on a silicon photomultiplier tube and an FPGA, wherein all unit detectors are electrically connected to the FPGA, and the FPGA executes a dynamic gain compensation method for a silicon photomultiplier tube as described in any of the first aspects above.

[0066] Thirdly, embodiments of the present invention provide a PET system, which includes the detection device described in the second aspect above.

[0067] (III) Beneficial Effects

[0068] The method of the present invention solves the technical defects of the prior art in which real-time temperature acquisition compensation occupies system resources and reduces system dead time, while maintaining the stability of the gain of silicon photomultiplier tubes in PET system detectors.

[0069] In other words, the method of the present invention minimizes the system's reading, transmission, and judgment time when acquiring data in real time in the PET system. It can synchronously compensate the gain of the silicon photomultiplier tube according to temperature changes without affecting data acquisition. By continuously acquiring and correcting the temperature change cycle during idle acquisition time, it ensures the accuracy of the prediction model during acquisition. It also ensures that the gain of the PET detector remains stable, thereby ensuring the sensitivity and stability of the PET system. Attached Figure Description

[0070] Figure 1 This is a flowchart of the real-time temperature compensation process.

[0071] Figure 2This is a schematic diagram showing the temperature change of a detector at a certain location over time.

[0072] Figure 3 This is a schematic diagram showing how the gain of a detector at a certain location changes with temperature.

[0073] Figure 4 This is a schematic diagram showing the change in single-lift over-energy window count rate of a detector at a certain location before and after correction, as a function of temperature.

[0074] Figure 5 This is a flowchart illustrating a dynamic gain compensation method for a silicon photomultiplier tube according to an embodiment of the present invention. Detailed Implementation

[0075] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0076] The method of this invention can be applied to radiation detectors in nuclear medicine imagers that employ radiation transmission or radiopharmaceuticals, such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT) imagers.

[0077] Typically, PET systems use silicon photomultiplier tubes (SMTs) as detectors for photoelectric conversion. The method described in this invention can also be applied to other radiation imaging modalities, and to systems or detection methods employing radiation detectors, such as those used in astronomy and airport baggage screening.

[0078] The method of this invention can be understood as a method for dynamically compensating the gain of silicon photomultiplier tubes under non-real-time temperature acquisition conditions. The method of this embodiment can be illustrated using a PET system as an example. A PET system consists of dozens to hundreds of unit detectors. Each unit detector has an independent temperature sensor and readout circuit to test the operating temperature of the unit detector. The photosensitive element of a unit detector is generally composed of an array of dozens to hundreds of silicon photomultiplier tubes. Each silicon photomultiplier tube may correspond to one or more scintillator crystal strips.

[0079] Example 1

[0080] like Figure 1 and Figure 5 As shown, this embodiment of the invention provides a dynamic gain compensation method for silicon photomultiplier tubes. The dynamic gain compensation method acquires gain compensation information in the first state of all unit detectors in the PET system and corrects it online in the single-launch event of the second state of all unit detectors.

[0081] Specifically, the dynamic gain compensation method for silicon photomultiplier tubes includes the following steps:

[0082] S01. The FPGA determines whether the unit detector is in the first state. If so, it obtains the gain compensation parameters of the gain compensation information according to the gain acquisition mode, or updates the gain compensation parameters in the known gain compensation information in real time according to the known gain compensation information.

[0083] For example, the FPGA determines whether it has received a data acquisition command. If it has received and is executing the acquisition command, the FPGA knows that the unit detector is in the second state; otherwise, it is in the first state. If the FPGA has completed data acquisition or has not received a data acquisition command, it confirms that it is in the first state. The FPGA here is the FPGA electrically connected to all unit detectors.

[0084] It should be noted that when using the method of this embodiment for the first time, the gain compensation parameters of the gain compensation information are obtained according to the gain acquisition mode. After the gain compensation parameters are available, the gain compensation parameters in the known gain compensation information can be updated in real time.

[0085] In addition, the known gain compensation information in this embodiment can be gain compensation information obtained in the manner of this application, or gain compensation information obtained in other ways, as long as it can be updated in accordance with the update method of this application.

[0086] S02. When the unit detector is in the second state, the FPGA corrects the data in the single-lift event of the second state according to the latest gain compensation parameters, so as to obtain the detection data for reconstructing the PET image based on the corrected data.

[0087] The first state is the non-acquisition state of the unit detector, and the second state is the acquisition state of the unit detector.

[0088] The method in this embodiment solves the technical defects of existing technologies, such as real-time temperature acquisition compensation occupying system resources and reducing system dead time, while maintaining the stability of the gain of the silicon photomultiplier tube of the PET system detector.

[0089] Typically, changing the external cooling system of a detector will alter its actual operating temperature and gain. Since the actual operating temperature varies with the cycle and is not a fixed value, to better implement the method of this invention, in practical applications, gain should be acquired at multiples of the temperature change over an entire cycle, and the operating temperature can be replaced by a one-cycle temperature averaging method.

[0090] To better understand the process of obtaining the gain compensation parameters based on the gain acquisition mode in step S01 above, the following will provide a detailed explanation in conjunction with sub-steps A01 to A04.

[0091] Sub-step A01: Obtain the first curve of the operating temperature and gain of each unit detector by means of a calibration measurement method.

[0092] In this embodiment, the relationship between the gain and operating temperature of the unit detector is achieved through radiation source calibration. Specifically, as the temperature increases, the measured channel value of the energy peak decreases. A unit detector typically has about 200 single crystal strips, each with several energy spectra. Usually, a unit detector will have several temperature-gain curves for each strip, while only one temperature probe (temperature sensor) is used. In this embodiment, to simplify the temperature-gain curves, the average channel value of all single crystal strips is used to represent the channel value of the detector at that temperature.

[0093] Furthermore, in practice, energy spectrum acquisition cannot be done in a short period of time. Depending on the source intensity, it takes at least several minutes to more than ten minutes to collect and analyze the data to ensure the statistical results. During these few minutes to more than ten minutes, the actual temperature also changes periodically. Therefore, in order to accurately characterize the temperature, the data collection time for statistical energy spectrum must be the entire cycle of temperature change. Only in this way can the average channel value of the energy spectrum correspond to the average temperature of one cycle.

[0094] That is, the first curve is the curve of gain changing with operating temperature. This curve is determined by the physical properties of the unit detector and is stable in the long term. It is generally measured once every few months or once a year, such as once every 6 months or 12 months.

[0095] Sub-step A02: Obtain the second curve of the operating temperature of each unit detector changing over time.

[0096] In this sub-step, the second curve is the operating temperature changing over time. This curve stabilizes within a short period, typically within several hours, and requires frequent calibration. The parameters of the temperature-time function are calculated from the calibrated values ​​to predict the temperature change curve during patient data acquisition. Typically, each user is in the second state for approximately half an hour, meaning data acquisition occupies no more than half an hour of the system. Therefore, in the first state (outside of the second state), when the system's FPGA is idle, calibration can be performed to acquire the second curve.

[0097] The acquisition process of the second curve and the acquisition process of the first curve are independent of each other and do not conflict. They do not have to be carried out in the same time period. The first curve is updated and the second curve is updated based on actual needs. The update frequency of the first curve is different from that of the second curve. The update frequency of the second curve is much less than that of the first curve.

[0098] Understandably, PET systems all have cooling structures, such as air cooling, water cooling, or a combination of both, to ensure that the PET system detector and its photosensitive elements, such as silicon photomultiplier tubes, do not continuously overheat. In other words, the detector temperature of a PET system actually varies periodically around a mean. Under conditions where the external environment does not change drastically, the temperature variation of the PET system detector is stable and regular.

[0099] The temperature change period at the detector end is consistent with that of the cooling system, but there is a phase difference; it occurs later than the cooling system. The maximum value is less than or equal to the upper threshold of the cooling system, and the minimum value is greater than or equal to the lower threshold of the cooling system. Therefore, the temperature change of the unit detector over time is predictable.

[0100] Sub-step A03: Perform an approximate linear fit on the second curves of all unit detectors to obtain intermediate parameters for calculating the gain compensation parameters.

[0101] That is, before the PET system collects data, it continuously collects the temperature changes of each unit detector in real time. This allows it to fit a predictive function of the detector temperature changing over time through filtering and averaging. Based on the relationship between temperature and gain curves, a predictive function of the detector gain changing over time can be derived. This predictive function is then used to calculate the temperature change over time. 上升 (t) and function T 下降 (t) describes the temperature rise and fall processes over time, respectively. Where T 上升 and T 下降 The temperature represents the rise and fall, and t represents time. Intermediate parameters are obtained by fitting a series of measurements.

[0102] Sub-step A04: Based on the first curve and the intermediate parameters, obtain the gain compensation parameters that change with time, i.e., the third curve.

[0103] The third curve, obtained by substituting the temperature parameter, becomes a gain versus time curve. In this embodiment, to better achieve gain correction, the temperature value is not required during correction. Utilizing substitution relationships and based on the first and second curves, the temperature change prediction can be transformed into a gain versus time prediction. Without calculating the temperature, the measured gain at the current moment can be calculated using the current time. The ratio of the fixed-temperature gain to the predicted measured gain is the prediction ratio. Multiplying the measured gain by the prediction ratio equals the fixed-temperature gain. Therefore, the gain can be maintained at a fixed temperature, and its minor fluctuations are within an acceptable range.

[0104] The above method minimizes the system's reading, transmission, and judgment time when acquiring data in real time in the PET system. It can synchronously compensate the gain of the silicon photomultiplier tube according to temperature changes without affecting data acquisition. By continuously acquiring and correcting the temperature change cycle during idle periods, it ensures the accuracy of the prediction model during acquisition. It also ensures that the gain of the PET detector remains stable, thereby guaranteeing the sensitivity and stability of the PET system.

[0105] Example 2

[0106] To better understand the method of Embodiment 1 above, the following will be combined with... Figures 2 to 5 The dynamic gain compensation method for silicon photomultiplier tubes is explained in detail.

[0107] Step 1: Measure the operating temperature and gain variation curve of the unit detector. .

[0108] This step pertains to the fixed variations of the unit detector. This embodiment primarily describes the process of acquiring parameters. For example, it may include the following steps:

[0109] A011. The operating temperature of each unit detector is read using the temperature sensor of each unit detector.

[0110] In practice, the PET system employs a water-cooled and air-cooled system. A water-cooled plate, cooled by circulating cooling water, is placed close to the silicon photomultiplier tube on the detector for cooling. Air cooling is used to cool the circuit board and FPGA that process the signal. Assuming each unit detector has only one temperature sensor, if multiple temperature sensors are used, the average value of all sensors can be used to characterize the operating temperature of the unit detector. Alternatively, different operating temperatures can be used to characterize the temperature of a unit detector at different locations based on the sensor settings.

[0111] A012. With the Ge barrel placed in the middle of all unit detectors, record the energy peak at 511 keV in the single crystal energy spectrum within one cycle to obtain the operating temperature and peak position variation point of each single crystal strip.

[0112] A013. The average value of all single crystal peaks of the unit detector is used to characterize the peak position of the unit detector, and the first curve of the change of operating temperature and gain of the unit detector is obtained.

[0113] That is, each detector in a PET system has a temperature sensor, and the photosensitive element of a single detector unit can consist of an array of tens to hundreds of silicon photomultiplier tubes. The average temperature change of the single detector unit and the average gain change of the silicon photomultiplier tubes are used to represent the temperature-gain relationship of each single detector unit in the PET system, which can be expressed using a function G. 增益 (Temp) indicates that G 增益 It represents the gain of the silicon photomultiplier tube, and Temp represents the temperature.

[0114] That is, using the first function G 增益 (Temp) = a 增益 *Temp+b 增益 .

[0115] The water-cooling system exhibits highly stable periodic temperature changes. By altering the outlet water temperature of the water-cooling system, the operating temperature and gain variation curve of the detector can be measured. Through a series of outlet water temperature changes, the operating temperature and peak position variation point of each single crystal strip can be measured, thus obtaining a variation curve.

[0116] Specifically, a uniform Ge source is placed in the center, and the data acquisition time is an integer multiple of the water temperature change cycle. The detector's operating temperature is the average temperature of one cycle. The energy spectrum of the single crystal obtained during this time period corresponds to this average operating temperature. The Ge source emits a pair of photons with an energy of 511 keV through electron-positron annihilation. The single crystal energy spectrum will generate an energy peak at 511 keV, and the change in the position of this energy peak corresponds to a change in gain. Before the initial real-time calibration, the necessary parameters need to be input. The value 'a' is required to generate the calibration parameters. 增益 and b 增益 .

[0117] The parameters of the first curve are relatively stable and remain largely unchanged, so they can be measured once every few months or a year and then input into the system for calculation.

[0118] Understandably, acquiring the energy spectrum cannot be done in a short time. Depending on the source intensity, it takes at least several minutes to more than ten minutes to collect and analyze the data to ensure the statistical results. During these few minutes to more than ten minutes, the actual temperature also changes periodically. Therefore, in order to accurately characterize the temperature, the time length for collecting the statistical energy spectrum data must be the entire cycle of temperature change. Only in this way can the average channel value of the energy spectrum correspond to the average temperature of one cycle.

[0119] Typically, a single-unit detector has only one temperature sensor, and the percentage difference in energy peak changes with temperature among different crystals within the same detector unit is relatively small. To reduce computational load and optimize system response speed, the average value of all single-crystal peaks is used to characterize the peak position of the detector, thus obtaining the detector's operating temperature and gain variation points. Figure 3 As shown, the gain G 增益 The relationship between temperature (Temp) and temperature change is linear, using the function G. 增益 (Temp) = a 增益 Temp+b 增益 The coefficient a is obtained through linear fitting. 增益 and b 增益 value.

[0120] Step 2: Obtain the second curve showing the change of the operating temperature of each detector unit over time.

[0121] It should be noted that the second step here is unrelated to the first step mentioned above; they are executed independently. The operating temperature here is not the same as the operating temperature in the first step, nor is it the operating temperature at a specific time point. The operating temperatures of the unit detectors in both the first and second steps are real-time measured and recorded values.

[0122] like Figure 2 As shown, the first step is to calibrate each detector unit in the PET system before data acquisition. For example, high-speed acquisition of temperature changes in the PET system allows for offline analysis to determine the temperature change period and fluctuation range. Since the system's heat generation and dissipation are essentially fixed, the period and range of each detector are also essentially fixed, resulting in testing errors only within a very small range. The average temperature, however, will vary due to differences in water temperature settings and ambient temperature. By analyzing the offline temperature change data over time, the average period of temperature change and the average maximum and minimum values ​​of temperature fluctuations can be obtained, providing initial calculation values ​​for online real-time calibration.

[0123] Understandably, before the first online calibration, real-time acquisition and recording of detector temperature and time data must be performed offline for analysis to provide initial values. Subsequent online calibrations can continue using offline analysis, but in practice, parameters uploaded to the front-end FPGA are retrieved back to the back-end FPGA to provide initial values, thus performing the rolling calibration. The front-end FPGA and back-end FPGA are two separate parts of the FPGA architecture.

[0124] In addition, complex offline calculations involve sparsely acquiring temperature data over time, fitting the rise and fall of each cycle, and then calculating the maximum and minimum values ​​according to the fitting formula. The rise and fall time differences are then averaged across four sets of values. Simpler calculations involve densely acquiring data to directly read the maximum and minimum values, the rise and fall times, and then averaging them. After real-time calibration, correction function parameters are obtained. Initial parameters are required for real-time calibration. These initial parameters can be calculated offline by acquiring a portion of the temperature data over time, or they can be the last recorded value from the previous real-time calibration. Five sets of values ​​are retained for real-time calibration: 1) the start time of one cycle; 2) the average rise time; 3) the average fall time; 4) the average maximum value of the cycle; and 5) the average minimum value of the cycle.

[0125] In this embodiment, the linear relationship of the second curve is expressed as: Temp 降 =a 降 t+b 降 Temp 升 =a 升 t+b 升 ;

[0126] Online real-time calibration requires input of the initial period rise time T. 0上升 and descent time T0下降 Maximum initial temperature (Temp) 0max and minimum initial temperature Temp 0min The starting value can be the average period and average temperature extreme value of the offline analysis, or it can be the last result of the previous temperature value.

[0127] The FPGA will automatically be in the first cycle range (T) 0上升 +T 0下降 Find the maximum temperature (Temp) within the first cycle. 1max and the time t at that point 1max And the minimum temperature of the first cycle, Temp. 1min and t 1min ΔT=t 1min -t 1max If the time difference is greater than 0, then ΔT is used as the new descending cycle. 1下降 This means that if the difference ΔT is less than 0, then ΔT is used as the new downward cycle. 1上升 express;

[0128] It should be noted that the starting point for temperature measurement is random; it could be during an upward or downward trend. Only a pair of highest and lowest temperatures can be obtained for each temperature cycle, allowing the determination of the cycle's starting position. The cycle can begin with either the highest or lowest temperature. By determining the order of the highest and lowest temperatures, it's possible to identify whether the cycle is complete (upward or downward). If ΔT is greater than 0, t1max is used as the starting time, and this complete downward cycle is used for calibration calculations. If ΔT is less than 0, the data from this complete upward cycle can also be used for calibration calculations.

[0129] t 1max The time is used as the starting point of the new cycle (determined during the calibration process). A new minimum value, Temp, is determined within the extended time of the selected new cycle (i.e., a time interval slightly larger than the interval of the new cycle). 2min and its corresponding time t 2min and the maximum value Temp 2max and the time t at that point 2max T 2下降 =t 2min -t 1max T 2上升 =t 2max -t 2min , t 2max Using the time point as the starting point of a new cycle, five sets of data for the second curve are obtained by iterating through the data in sequence.

[0130] The data for all five sets of the second curves include: the period start time t. star descent time T 下降and rise time T 上升 Maximum temperature within the cycle: Temp max and minimum value Temp min .

[0131] Step 3: Perform an approximate linear fit on the data from the second curve to obtain the parameters used to calculate the gain compensation. Intermediate parameters.

[0132] t star =t max(上组) The starting time is the maximum temperature (Temp) of the last complete cycle at the corresponding scale time. max The corresponding time t max ;

[0133] T 下降 =t min -t star ;

[0134] T 上升 =t max -t min ;

[0135] The remaining 4 sets of values ​​T 下降 T 上升 Temp max and Temp min Each new value is updated by adding 1 / 16 of the original value to 15 / 16 of the new value. All four sets of values ​​are averaged using this method. The real-time rolling average yields five sets of data. To reduce computation, only the maximum and minimum values, and the time difference between rise and fall, are retained. A two-point linear fit can then be used to simplify the calculation. In other words, while polynomial fitting is optimal for the rising and falling curves in offline data analysis, the large number of fitting coefficients in online real-time scaling necessitates the use of an approximate linear fit to reduce the difficulty of implementing online real-time scaling.

[0136] That is: Current value = New value * 1 / 16 + Old value * 15 / 16;

[0137] Get a 降 =(Temp min -Temp max ) / T 下降 ;

[0138] b 降 =Temp max ;

[0139] a 升 =(Temp max -Temp min ) / T 上升 ;

[0140] b 升=Temp min .

[0141] The rolling average can be stacked at 1 / 8 or 1 / 16, and the denominator can vary. It is a weighted average, and its value is related to the temperature change cycle. A shorter cycle results in a larger denominator, and a longer cycle results in a smaller denominator. After each cycle is completed, 5 new values ​​are output. The current time value is directly overwritten by the scale value, and the remaining 4 values ​​are weighted and averaged with the scale value before being saved as the new scale value.

[0142] The fourth step is to calculate the detector gain correction function over time.

[0143] Based on the approximate linear formula (1) for the relationship between temperature and time:

[0144] Temp 降 =a 降 t+b 降 ;

[0145] Temp 升 =a 升 t+b 升 (1)

[0146] t = t 当前 -t star ;

[0147] Based on the 5 sets of values ​​recorded by the system, the coefficients of the approximate linear formula (2) are calculated:

[0148] a 降 =(Temp min -Temp max ) / T 下降 ;

[0149] b 降 =Temp max (2)

[0150] a 升 =(Temp max -Temp min ) / T 上升 ;

[0151] b 升 =Temp min ;

[0152] Simultaneously, based on the gain-temperature variation formula parameter G already stored in the register... 增益 =a 增益 Temp+b 增益 Calculate the parameters of the gain-time formula.

[0153] Formula (3) is an empirical formula, meaning that the expected gain is to stabilize at G.标准 However, the actual gain changes with temperature as G 实测 Therefore G 增益 This is a result of temperature effects; we want to compensate for the temperature effects and stabilize the gain at G. 标准 This requires the current gain value G. 实测 Multiply by the compensation coefficient R 补偿 ;

[0154] G 标准 = G 实测 R 补偿 R 补偿 =G 标准 / G 增益 (3)

[0155] G 标准 It is the detector gain value at the expected average temperature, which is a known value.

[0156] Temp 标准 It is a definite value, representing the detector's expected average operating temperature, based on G. 增益 =a 增益 Temp+b 增益 It can be known that:

[0157] G 标准 =a 增益 Temp 标准 +b 增益 (4)

[0158] R 补偿 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 Temp+b 增益 );

[0159] R 补偿 The variable is temperature (Temp), and Temp is a piecewise function divided into Temp... 降 =a 降 t+b 降 and

[0160] Temp 升 =a 升 t+b 升 ;

[0161] Formula (5) is:

[0162] R 补偿降 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 (a 降 t+b降 )+b 增益 );

[0163] R 补偿升 =(a 增益 Temp 标准 +b 增益 ) / (a 增益 (a 升 t+b 升 )+b 增益 );

[0164] In the above formula (5), all coefficients except t are known values. Substitute them into the 5 sets of data recorded in real time correction:

[0165] Based on the formula (2), we know that: a 降 b 降 a 升 b 升 , t = t 当前 -t star ;

[0166] Then we can calculate formula (6):

[0167] Formula (6) is:

[0168]

[0169]

[0170] t 当前 The time is used to obtain the data in real time.

[0171] R 补偿降 and R 补偿升 Judgment conditions:

[0172] Remainder = (t) 当前 -t star ) / (T 下降 +T 上升 The remainder of )

[0173] Remainder ≤ T 下降 For the descent period, use formula R. 补偿降 ;

[0174] Remainder > T 下降 For the upward cycle, use formula R. 补偿升 ;

[0175] The above formula (6) is written into the front-end FPGA as a correction formula and used when collecting patient data.

[0176] Step 5: Implement real-time gain correction using FPGA.

[0177] When the PET system is idle, the back-end FPGA controls the front-end FPGA to continuously perform real-time online calibration. The back-end FPGA analyzes and calculates the real-time calibration. When the control software interface issues a data acquisition command to the system, the rolling calibration calculation stops. The last set of analyzed and calculated calibration values, totaling 5 sets, are retrieved from the back-end FPGA. star and descent time T 下降 and rise time T 上升 and Temp max and Temp min The data is uploaded as a parameter to the front-end FPGA and then the data acquisition begins.

[0178] The front-end FPGA determines whether the temperature should be rising or falling based on the actual data acquisition time, and calculates R in real time. 补偿 It can perform normal integration of the energy path by triggering a threshold-based single-step event, and multiply the integral value by R. 补偿 Then, the data is filtered through a window, followed by time-matching filtering, and finally stored as raw data on a computer, awaiting reconstruction into PET images.

[0179] For example, a PET system has a front-end processing FPGA (6 detectors share one front-end FPGA, totaling 20-30), a back-end processing FPGA (only one in the entire system), and a computer control panel. Calibration data is acquired by the front-end FPGA and sent to the back-end FPGA for processing. After calibration processing, a series of parameters are obtained and transmitted back to the front-end FPGA, which then completes the calibration process. Figure 1 The correction and energy window match are completed by the back-end FPGA. Figure 1 In the middle, the back-end FPGA transmits data to the computer console, and the computer console saves the data.

[0180] Therefore, whether real-time data acquisition can correctly correct the gain can be compared by observing changes in the count rate of a fixed bucket source. Without real-time correction, it can be observed that the gain changes due to temperature variations in the single-source over-the-window data, and the count rate over the over-the-window exhibits a certain regularity. When real-time correction is used, the difference in count rate decreases significantly. Figure 4 As shown.

[0181] In this embodiment, the functions of steps one through four are written into the FPGA. During data acquisition, the acquired data is predicted and real-time compensated and corrected. The temperature is no longer read in real time during the data acquisition process, thereby reducing the system dead time caused by the system reading temperature. At the same time, the gain is compensated and corrected by predicting the temperature change curve to ensure the gain stability.

[0182] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0183] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, 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 different embodiments or examples.

[0184] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0185] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for dynamic gain compensation of a silicon photomultiplier tube, characterized in that, The dynamic gain compensation method acquires gain compensation information in the first state of all unit detectors in the PET system and corrects it online in the single-launch event of the second state of all unit detectors. The method includes: The FPGA determines whether the unit detector is in the first state. If so, it obtains the gain compensation parameters of the gain compensation information according to the gain acquisition mode, or updates the gain compensation parameters in the known gain compensation information in real time according to the known gain compensation information. When the unit detector is in the second state, the FPGA corrects the data in the single-lift event of the second state according to the latest gain compensation parameters, so as to obtain detection data for reconstructing PET images based on the corrected data; The first state is the non-acquisition state of the unit detector, the second state is the acquisition state of the unit detector, and the FPGA is the FPGA to which all unit detectors in the PET system belong; The gain compensation parameters, which are used to obtain gain compensation information based on the gain acquisition mode, include: A01. Obtain the first curve of the operating temperature and gain of each unit detector by calibration measurement; A02. Obtain the second curve showing the change of the operating temperature of each unit detector over time; A03. Perform an approximate linear fit on the second curve of all unit detectors to obtain intermediate parameters for calculating the gain compensation parameters. A04. Based on the first curve and the intermediate parameters, obtain the gain compensation parameters that show the change of gain over time.

2. The method according to claim 1, characterized in that, The FPGA determines whether the unit detector is in the first state, including: The FPGA determines whether it has received a data acquisition command. If it has, it determines that the unit detector is in the second state; otherwise, it is in the first state.

3. The method according to claim 1, characterized in that, A01. Obtain the first curve of the operating temperature and gain of each detector unit through calibration measurement, including: A011. The operating temperature of each unit detector is read using the temperature sensor of each unit detector; A012. With the Ge barrel placed in the middle of all unit detectors, record the energy peak at 511 keV in the single crystal energy spectrum within one cycle to obtain the operating temperature and peak position variation point of each single crystal strip. A013. The average value of all single crystal peaks of the unit detector is used to characterize the peak position of the unit detector, and the first curve of the change of operating temperature and gain of the unit detector is obtained. Using the first function representation: G 增益 (Temp) = a 增益 *Temp +b 增益 ; Formula (m1) The coefficient 'a' was obtained through linear fitting. 增益 and b 增益 Temp is the operating temperature, G 增益 (Temp) Gain as a function of operating temperature.

4. The method according to claim 3, characterized in that, The linear relationship of the second curve in A02 is expressed as follows: Temp 降 =a 降 t + b 降 ; Temp 升 =a 升 t + b 升 ; Formula (m2) The second curve in A02 was obtained using an online real-time calibration method. The initial period rise time T was determined using the average period and average temperature extreme values ​​from offline analysis, or the records left after the last calibration. 0上升 and descent time T 0下降 Maximum initial temperature (Temp) 0max and minimum initial temperature Temp 0min , FPGA based on the first cycle range (T) 0上升 +T 0下降 Get the maximum temperature (Temp) within the first cycle. 1max and the time t when the temperature reaches its maximum 1max And the minimum temperature of the first cycle, Temp. 1min and t 1min ; ΔT=t 1min -t 1max If the time difference is greater than 0, then ΔT is used as the new descending cycle, represented by T. 1下降 This means that if the difference ΔT is less than 0, then ΔT is used as the new downward cycle. 1上升 express; t 1max The time is taken as the starting point of the new cycle, and a new minimum value Temp is determined within the extended time of the selected new cycle. 2min and time t 2min and the maximum value Temp 2max and time t 2max T 2下降 =t 2min -t 1max T 2上升 =t 2max -t 2min , t 2max Using the time point as the starting point of a new cycle, five sets of data for the second curve are obtained by iterating through the data in sequence. The data for all five sets of the second curves include: the period start time t. star descent time T 下降 and rise time T 上升 Maximum temperature within the cycle: Temp max and minimum value Temp min .

5. The method according to claim 4, characterized in that, A03. Perform an approximate linear fit on the second curves of all unit detectors to obtain intermediate parameters for calculating the gain compensation parameters, including: t star =t max（上组） The starting time is the maximum temperature (Temp) of the last complete cycle at the corresponding scale time. max The corresponding time t max ; T 下降 =t min -t star ; T 上升 =t max -t min ; The remaining 4 sets of values ​​T 下降 T 上升 Temp max and Temp min Each time a new value is obtained, it is updated by adding 1 / 16 of the original value to 15 / 16. All four sets of values ​​are averaged using this method. That is: Current value = New value * 1 / 16 + Old value * 15 / 16; Get intermediate parameter a 降 b 降 a 升 b 升 ; a 降 =(Temp min -Temp max ) / T 下降 ; b 降 =Temp max ; a 升 =(Temp max -Temp min ) / T 上升 ; b 升 =Temp min 。 6. The method according to claim 5, characterized in that, In A04, based on the first curve and the intermediate parameters, gain compensation parameters that vary with time are obtained, including: According to formula (m1), formula (m2), t=t 当前 -t star and intermediate parameter a 降 b 降 a 升 b 升 The formula for the gain compensation parameter, which serves as the gain compensation information, can be obtained from the empirical formula for gain (m3). , : ； ； t 当前 To obtain the data in real time; R 补偿降 and R 补偿升 Judgment condition: Remainder = (t) 当前 -t star ) / ( T 下降 +T 上升 The remainder of ). Remainder ≤ T 下降 For the descent period, use R 补偿降 ; Remainder > T 下降 For the rising cycle, use R 补偿升 ; The formula (m3) is: G 标准 = G 实测 R 补偿 ;R 补偿 =G 标准 / G 增益 ; G 标准 It is the gain value of the unit detector at the expected average temperature, which is a known value. Temp 标准 It is a definite value, representing the expected average operating temperature of the unit detector.

7. The method according to claim 6, characterized in that, The FPGA corrects the data in the single-launch event of the second state according to the latest gain compensation parameters, including: The formula for the gain compensation parameter in the gain compensation information. , The data is written to the front-end FPGA, and the FPGA corrects the data in the single-instance event of the second state based on the latest gain compensation parameters to obtain the original data as probe data. Alternatively, the formula for the gain compensation parameter in the gain compensation information. , The data is written to the front-end FPGA, which automatically corrects the gain according to the latest gain compensation parameters. The gain-corrected single-instance data is then filtered through the energy window and time-matched to obtain the raw data as probe data.

8. A detection device, characterized in that, include: A silicon photomultiplier tube-based unit detector and an FPGA are used, with all unit detectors electrically connected to the FPGA, and the FPGA executes a dynamic gain compensation method for silicon photomultiplier tubes as described in any one of claims 1 to 7.

9. A PET system, characterized in that, include: The detection device as described in claim 8.

Images