Nuclear medicine imaging adaptive control method and system based on modular gamma probe

Abstract

The application provides a nuclear medicine imaging adaptive control method and system based on a modular gamma probe, and relates to the technical field of nuclear medicine imaging. The method comprises the following steps: acquiring multi-scale signal features and identifying the uptake mode of the current tracer distribution; dynamically adjusting the weight of the multi-scale signal features in the positioning decision of the modular gamma probe array according to the characteristics of the identified uptake mode; after the weight adjustment is completed, generating and executing adjustment instructions for the modular gamma probe array, and controlling the modular gamma probe to collect data, obtain image data and reconstruct a diagnostic image. The method aims to solve the problem of image quality degradation caused by data ambiguity and system misjudgment in existing adaptive control methods, effectively suppresses the acquisition performance degradation caused by the differences of the imaged object, thereby improving the diagnostic image quality, reducing artifacts and blurring, and enhancing the reliability and diagnostic accuracy of nuclear medicine imaging.

Description

Technical Field

[0001] This invention relates to the field of nuclear medicine imaging technology, and more specifically, to a nuclear medicine imaging adaptive control method and system based on a modular gamma probe. Background Technology

[0002] In the field of nuclear medicine imaging, we typically use modular gamma probe-based systems to precisely detect the distribution of radioactive tracers within the human body, thereby generating diagnostic images. These systems generally consist of multiple independent gamma probe modules, which can be flexibly configured and precisely positioned according to the actual imaging needs. To ensure imaging quality and efficiency, the system is equipped with a sophisticated positioning and actuation device for meticulous movement and angle adjustment of each probe module in three-dimensional space. Furthermore, to cope with various factors such as patient position, organ movement, and changes in probe characteristics, the system usually integrates an adaptive control method to optimize probe position and acquired parameters in real time, thereby improving image clarity and diagnostic accuracy.

[0003] However, in certain clinical scenarios, unusual physiological phenomena may arise, posing a significant challenge to the reliability of existing adaptive control methods. For example, consider this scenario: during a nuclear medicine examination of a patient with a liver tumor, we observe an abnormally high and diffuse uptake pattern of the radiotracer by the benign liver tissue surrounding the target organ, the liver tumor. This uptake is neither a background signal nor unique to the tumor itself, but rather a localized, diffuse accumulation of abnormality. This aberrant uptake blurs the actual boundary of the target organ in terms of tracer distribution, significantly reducing the contrast with surrounding tissue. The underlying cause of this physiological anomaly may be related to the patient's specific metabolic state, inflammatory response, or drug effects, but its external manifestation is the diffusion of the physical signal, resulting in spatial heterogeneity between the true contour of the target organ and the pre-defined focused scanning area.

[0004] Due to this non-specific, diffuse tracer uptake, the real-time counting data streams received by each module of the modular gamma probe array exhibit characteristics significantly different from the preset conditions during data acquisition. Specifically, relative to the preset counting distribution model corresponding to the target organ, the real-time signal intensity is unevenly distributed within the region, and the overall spatial distribution characteristics differ from the expected model. Under these abnormal conditions, the system may acquire and image these incomplete, partially misaligned, or completely mismatched regions, leading to an overestimation or underestimation of the target organ's apparent characteristics and further blurring its true boundaries.

[0005] Faced with such mismatched and noisy counting data, the core processing logic of adaptive control methods, especially its prior information on organ morphology and the adaptive adjustment rule set for probe position and angle, faces significant challenges. This prior information and rule set are typically trained and optimized based on a large amount of organ morphology data and tracer distribution patterns from typical patients. When encountering the aforementioned aberrant uptake patterns, the adaptive processing logic struggles to accurately adjust the target organ in real time. It attempts to continue optimizing imaging according to this non-ideal data acquisition and imaging method, but due to the deviation between the optimization target and the actual situation, the generated probe positioning and angle adjustment commands are not optimal, or even cause greater deviations. For example, the system might instruct the probe to move to a larger area than the actual tumor region for scanning, or linger in a certain area for too long, because it attempts to "cover" a target area it perceives as larger and with unclear boundaries in order to capture all possible signals. This "overfitting" deviates from precise focusing on the true target organ.

[0006] Before a formal scan begins, nuclear medicine imaging systems typically generate a preliminary reconnaissance image for the operator to review, ensuring correct probe positioning and scanning range. However, to improve visual clarity and user experience, the system's built-in image display optimization program automatically smooths and enhances the contrast of low-count-rate reconnaissance images. Under normal circumstances, this processing effectively suppresses noise, making the image appear "cleaner" and improving the technician's intuitive judgment of image quality. But in the aforementioned anomalous acquisition scenarios, this optimization program can mask the blurred features of signal boundaries. It might process anomalous diffuse areas as if they were normal low-signal backgrounds, or smooth blurred edges, making the reconnaissance image presented to the technician appear "acceptable," or even "clear." Lacking sufficient understanding of this specific physiological anomaly and overly relying on the system's "optimized" visual presentation, the technician fails to identify the potential signal distribution problem, thus approving the scan to continue. This misleading presentation of information by the human-computer interface in specific anomalous situations is a key factor leading to further escalation of the problem.

[0007] Due to the suboptimal probe positioning strategy guided by adaptive rules, and technicians approving scan initiation under system misguidance, the modular gamma probe spends an abnormally long time in suboptimal positions throughout the scan. When the probe remains far from the center of the actual target organ or faces the target at a suboptimal angle for an extended period, its field of view includes more non-target tissue. Although the tracer concentration in these tissues is low, they still generate a certain amount of radiation. Furthermore, due to the suboptimal geometry between the probe and the target organ, the detector receives an abnormally large amount of signals from non-target tissue and scattered signals. Scattered signals refer to the radiation from gamma photons that have undergone Compton scattering within the patient's body, changing their direction and energy, but are still captured by the detector. This non-target tissue and scattered signals further contaminate the real-time counting data stream of each module, as it increases background noise and blurs the true signal, leading to a further deterioration in the signal-to-noise ratio and exacerbated image blurring. For example, if the probe deviates from its optimal position, it may receive more scattered photons from deep tissues in the patient's body. These photons may be close to the direct photons in energy, but their spatial information has been distorted.

[0008] Ultimately, the cumulative effect of all the aforementioned factors—including abnormal uptake of tissues surrounding the patient's target organ, suboptimal probe positioning strategies based on inaccurate prior information in the adaptive control logic, masking of potential problems by the human-computer interface through image display optimization, and the resulting abnormal increase in noise—combined to cause serious problems in the core computational process of image reconstruction. The reconstructed nuclear medicine images of the target organ exhibit severe blurring, poor contrast, inaccurate volume representation, and may even contain artifacts. In this complex and multifaceted abnormal situation, the adaptive control method, which should have intelligently optimized image quality, inadvertently amplified the initial physiological challenges, rendering the final diagnostic images clinically unreliable. This not only requires time-consuming and specialized manual post-processing to attempt to salvage image information but may also necessitate repeated scans, thus negating the efficiency advantages of the modular gamma probe and adaptive control and increasing the patient's radiation exposure. Summary of the Invention

[0009] The purpose of this invention is to provide an adaptive control method and system for nuclear medicine imaging based on a modular gamma probe. This system aims to address the image quality degradation caused by data ambiguity and system misjudgment in existing adaptive control methods when the tissues surrounding the patient's target organ exhibit abnormal uptake patterns of the radiotracer during nuclear medicine imaging. The new method can accurately identify the tracer uptake pattern, dynamically optimize probe positioning decisions, and effectively suppress the degradation of acquisition efficiency caused by differences in the imaged object. This improves diagnostic image quality, reduces artifacts and blurring, and enhances the reliability and diagnostic accuracy of nuclear medicine imaging.

[0010] In a first aspect, the present invention provides an adaptive control method for nuclear medicine imaging based on a modular gamma probe, comprising the following steps: S1. Obtain multi-scale signal characteristics from the real-time counting data stream acquired by the modular gamma probe array; S2. Identify the uptake pattern of the current tracer distribution based on multi-scale signal characteristics; S3. Based on the characteristics of the identified acquisition pattern, dynamically adjust the weight of multi-scale signal features in the positioning decision of the modular gamma probe array; S4. After completing the weight adjustment, generate and execute adjustment instructions for the modular gamma probe array; the adjustment instructions are used to adjust the position and angle of the modular gamma probe. S5. After executing the adjustment command, control the modular gamma probe to acquire data and obtain image data; S6. Reconstruct the diagnostic image based on the image data.

[0011] The nuclear medicine imaging adaptive control method based on a modular gamma probe provided by this invention specifically solves the core problem of adaptive control failure under abnormal uptake modes, effectively overcomes the technical defects of image blurring, abnormal increase in scattered radiation, and suboptimal probe positioning, and ensures the clarity and morphological accuracy of diagnostic images.

[0012] Secondly, the present invention provides an adaptive control system for nuclear medicine imaging based on a modular gamma probe, comprising: The first acquisition module is used to acquire multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array; The identification module is used to identify the uptake pattern of the current tracer distribution based on multi-scale signal characteristics; The adjustment module is used to dynamically adjust the weight of multi-scale signal features in the positioning decision of the modular gamma probe array based on the characteristics of the identified acquisition pattern. The generation module is used to generate and execute adjustment instructions for the modular gamma probe array after the weight adjustment is completed; the adjustment instructions are used to adjust the position and angle of the modular gamma probe. The second acquisition module is used to control the modular gamma probe to acquire data and obtain image data after executing the adjustment command; The reconstruction module is used to reconstruct diagnostic images from image data.

[0013] As can be seen from the above, the nuclear medicine imaging adaptive control method based on a modular gamma probe provided by this invention achieves high-quality diagnostic image reconstruction by real-time acquisition of multi-scale signal features and accurate identification of acquisition patterns, combined with a dynamic weight adjustment mechanism to optimize probe positioning decisions and perform targeted probe position and angle adjustments. It effectively suppresses scattered radiation and achieves high-quality diagnostic image reconstruction. It has the advantages of accurately identifying abnormal diffuse acquisition patterns of tracers, dynamically optimizing probe positioning decisions, effectively suppressing the degradation of acquisition performance caused by differences in the imaged object, thereby improving diagnostic image quality, reducing artifacts and blurring, and enhancing the reliability and diagnostic accuracy of nuclear medicine imaging.

[0014] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0015] Figure 1 This is a flowchart of an adaptive control method for nuclear medicine imaging based on a modular gamma probe, provided in an embodiment of the present invention.

[0016] Figure 2 This is a schematic diagram of a nuclear medicine imaging adaptive control system based on a modular gamma probe, provided in an embodiment of the present invention.

[0017] Label Explanation: 100. First acquisition module; 200. Identification module; 300. Adjustment module; 400. Generation module; 500. Second acquisition module; 600. Reconstruction module. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0019] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0020] Reference Appendix Figure 1 This invention provides an adaptive control method for nuclear medicine imaging based on a modular gamma probe, comprising the following steps: S1. Obtain multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array; the multi-scale signal features include macro-scale features, meso-scale features and micro-scale features; S2. Based on the multi-scale signal characteristics, identify the uptake pattern of the current tracer distribution (identify whether it is an abnormal diffuse uptake pattern). S3. Based on the characteristics of the identified acquisition pattern, dynamically adjust the weights of multi-scale signal features in the positioning decision of the modular gamma probe array; the dynamic adjustment includes adjusting the first weight of macro-scale features related to the overall signal coverage in the positioning decision, the second weight of meso-scale features related to the diffusion center in the positioning decision, and the third weight of micro-scale features related to the sharp boundary in the positioning decision. S4. After completing the weight adjustment, generate and execute adjustment instructions for the modular gamma probe array; the adjustment instructions are used to adjust the position, angle and number of angles of the modular gamma probe; S5. After executing the adjustment command, control the modular gamma probe to acquire data and suppress scattered radiation to obtain image data; S6. Reconstruct the diagnostic image based on the image data.

[0021] This application provides an adaptive control method for nuclear medicine imaging based on a modular gamma probe. Obtaining multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array refers to extracting a set of parameters reflecting the signal's characteristics in different spatial or temporal dimensions. Specifically, macro-scale features can be achieved using global integral statistics methods, such as calculating the cumulative count rate distribution of the entire data stream; meso-scale features can be achieved using region clustering analysis methods, such as dividing signal intensity similar regions based on the K-means algorithm; and micro-scale features can be achieved using local gradient detection methods, such as using the Canny edge detection operator to identify signal abrupt changes, primarily to achieve hierarchical characterization of signal coverage, diffusion center distribution, and boundary sharpness. Furthermore, identifying whether the current tracer distribution exhibits an abnormal diffusion uptake pattern based on multi-scale signal features refers to determining whether the signal distribution deviates from a typical physiological pattern. In practical applications, this identification can be achieved using feature threshold comparison methods, such as comparing multi-scale feature values ​​with a preset reference range item by item; or using simple classification rules, such as a logic decision tree based on feature combinations, primarily to achieve accurate identification of non-specific diffusion signals. The dynamic adjustment of the weights of multi-scale signal features in the positioning decision of the modular gamma probe array refers to the redistribution of the contribution ratio of each scale feature to the probe positioning calculation. Specifically, the adjustment of the first weight can be based on signal coverage integrity assessment, such as calculating the integral area change rate of macroscopic scale features; the adjustment of the second weight can be based on diffusion center stability analysis, such as using centroid offset as the adjustment basis; and the adjustment of the third weight can be based on boundary clarity quantification, such as using the standard deviation of signal gradient amplitude for correction. This is mainly to achieve adaptive reconstruction of the positioning decision logic to abnormal diffusion characteristics. Furthermore, generating and executing adjustment instructions for the modular gamma probe array after weight adjustment means converting the optimized decision results into probe physical pose parameters. In practical applications, this step can be implemented using a geometric inverse mapping algorithm, such as determining the spatial position of each probe unit through coordinate system transformation; or using a parameter lookup table method, such as outputting angle adjustment values ​​and the number of angles to be adjusted according to pre-stored mapping relationships. This is mainly to achieve precise spatial focusing of the probe array on the real target organ. Among them, controlling the modular gamma probe to collect data and suppress scattered radiation after executing the adjustment command refers to reducing the interference of non-target radiation on the original data. In specific implementation, scattered radiation suppression can be achieved by energy window screening technology, such as setting a dynamic energy threshold to filter low-energy photon events; or by coincidence detection technology, such as using time correlation to exclude random coincidence events. The main purpose is to improve the signal-to-noise ratio of the original data.Finally, reconstructing diagnostic images from image data refers to generating visual images that can be used for clinical diagnosis. This process can be achieved using filtered back-projection algorithms, such as using a Shepp-Logan filter for reconstruction; or using algebraic reconstruction techniques, such as iteratively optimizing and solving linear equations, primarily to achieve accurate restoration of the target organ's morphology. This embodiment effectively distinguishes between normal tracer distribution and abnormal diffuse uptake patterns through multi-scale signal feature layered extraction and dynamic weight adjustment mechanisms, avoiding probe positioning deviations and scattered radiation accumulation problems caused by misjudgment in traditional methods. This allows the modular gamma probe array to focus more accurately on the real target organ, improving the reconstruction quality of the diagnostic images.

[0022] In nuclear medicine imaging, real-time counting data streams acquired by a modular gamma probe array are used to obtain multi-scale signal features, encompassing macro-scale, meso-scale, and micro-scale features. Macro-scale features reflect the overall signal coverage of the tracer distribution, meso-scale features characterize the spatial distribution of diffusion centers, and micro-scale features capture the sharpness of signal boundaries. Based on these multi-scale signal features, the system uses multi-dimensional correlation analysis to identify whether the current tracer distribution exhibits an abnormal diffusion uptake pattern. This identification process avoids the limitations of traditional single-threshold judgments and can effectively distinguish between normal background signals and pathological diffusion uptake. When an abnormal diffusion uptake pattern is identified, the system dynamically adjusts the weight allocation of multi-scale signal features in the localization decision. Specifically, this includes adjusting the first weight related to the overall signal coverage of macro-scale features, the second weight related to the diffusion centers of meso-scale features, and the third weight related to the sharp boundaries of micro-scale features. Weight adjustment is optimized based on the spatial heterogeneity distribution of anomalous patterns. For example, in areas with high diffusion and low signal-to-noise ratio, the first weight is increased to enhance the reliability of overall signal coverage, while the third weight is decreased to reduce over-reliance on blurred boundaries. After weight adjustment, the system converts the optimized decision parameters into specific position and angle parameters (including the number of angles) for each probe in the modular gamma probe array through geometric mapping, and generates corresponding adjustment instructions. Executing these instructions reconfigures the probe array to focus on the actual target organ region, thereby significantly reducing interference from non-target tissues during data acquisition and effectively suppressing the accumulation of scattered radiation. Ultimately, the diagnostic image reconstructed based on the optimized image data accurately restores the morphology of the target organ, avoiding volume distortion and artifacts.

[0023] As one specific implementation method: In step S1, the system uses a high-performance digital signal processor (DSP) to aggregate the raw gamma photon count data received from multiple modular gamma probes (each probe containing a scintillation crystal and a photomultiplier tube). At the macroscopic scale, the DSP calculates the ratio of the number of grid cells with count rates higher than the background threshold within a preset region of interest (ROI) to the total number of grid cells in the ROI as the diffusion range index, and calculates the average count rate of all grid cells within the ROI as the average intensity. At the mesoscopic scale, the DSP applies the Canny edge detection operator to identify signal edges and quantifies the degree of boundary blurring by calculating the gradient width of the edges. Simultaneously, it calculates local gray-level co-occurrence matrix (GLCM) features such as contrast, energy, and entropy to describe texture information. At the microscopic scale, each probe is equipped with a high-resolution analog-to-digital converter (ADC) and a time-to-digital converter (TDC). The DSP analyzes the Poisson distribution deviation of the gamma photon count rate within a 100-millisecond time window and within a single detector crystal space cell in real time, generates an energy spectrum histogram, and calculates the scattering ratio by analyzing the relative proportions of the photoelectric peak and the Compton scattering peak.

[0024] In step S2, the DSP normalizes the diffusion range index, average intensity, boundary blur index, GLCM texture features, Poisson distribution bias, and scattering ratio calculated in S1 to form a multi-dimensional feature vector. The system's built-in embedded controller (such as an ARM Cortex-M series microcontroller) runs a fuzzy logic judgment module. If the boundary blur index is higher than a preset threshold (e.g., 0.7) and the scattering ratio is higher than a preset threshold (e.g., 0.3), the controller determines that the current tracer distribution is an abnormal diffusion acquisition mode and generates a pattern recognition signal.

[0025] In step S3, when an abnormal diffusion mode signal is detected in S2, the weight adjustment unit in the embedded controller dynamically modifies the weight parameters in the probe positioning target function. By default, the first weight is 0.3, the second weight is 0.3, and the third weight is 0.4. In the abnormal diffusion mode, the weight adjustment unit reduces the third weight from 0.4 to 0.1, while increasing the first weight from 0.3 to 0.5 and the second weight from 0.3 to 0.4.

[0026] In step S4, the embedded controller calculates the optimal position, angle, and number of angles for the probe array based on the weights adjusted in S3 and in conjunction with real-time multi-scale features. For example, the calculation result may instruct the probe array to scan with a wider field of view and precisely align the geometric center of the probe array with the geometric center of the diffuse signal. Subsequently, the controller sends precise pulse and direction signals to the stepper motor or servo motor driver via a high-speed digital interface to control the modular gamma probe array to move and adjust its angle in three-dimensional space.

[0027] In step S5, after the modular gamma probe array moves to the position specified in command S4, it begins acquiring gamma photon data. During data acquisition, the system combines analysis of the energy spectrum based on microscopic characteristics to fine-tune the energy window of each probe module. For example, if the scattering ratio is high, the system tightens the lower limit of the energy window from 130 keV to 135 keV to exclude more low-energy scattered photons. After this optimized acquisition and scattering suppression, the probe array transmits the raw gamma photon count data to the image reconstruction unit to form image data.

[0028] In step S6, the image reconstruction unit (e.g., a computing server containing a high-performance GPU) receives the image data obtained in S5. This unit runs an iterative reconstruction algorithm, such as the Expectation-Maximization (EM) algorithm or the Ordered Subset Expectation-Maximization (OSEM) algorithm. Because the input image data has a higher signal-to-noise ratio, the reconstructed diagnostic image will exhibit accurate contrast, sharpness, and geometric and volumetric characteristics.

[0029] This method addresses the core issue of adaptive control failure under abnormal diffusion ingestion patterns through hierarchical analysis of multi-scale signal features and a dynamic weight adjustment mechanism. Specifically, the complementary identification of multi-scale features avoids misjudging abnormal diffusion as an expanded target region, while the dynamic weight adjustment mechanism reconstructs the positioning decision logic based on spatial heterogeneity distribution, ensuring the probe array always focuses on the actual target organ. Furthermore, the optimized probe geometry significantly reduces the proportion of non-target tissues in the field of view, thereby reducing the abnormal accumulation of scattered radiation and providing high-quality raw data for image reconstruction. Thus, this method effectively overcomes the technical shortcomings of image blurring, abnormally increased scattered radiation, and suboptimal probe positioning, ensuring the clarity and morphological accuracy of diagnostic images.

[0030] In some embodiments, the specific steps in step S1 include: S11. Evaluate the performance status parameters of each modular gamma probe unit in the modular gamma probe array; the performance status parameters include detection efficiency, system spatial resolution, energy resolution, and internal noise parameters; S12. Based on the performance status parameters, the raw counting data obtained from the real-time counting data stream acquired from the modular gamma probe array is calibrated to obtain calibrated counting data; S13. Based on the calibrated count data, extract macro-scale features, meso-scale features and micro-scale features, and perform spatial consistency verification on the extracted meso-scale features to obtain spatially consistent meso-scale features, and correct the extracted micro-scale features to obtain corrected micro-scale features.

[0031] Among them, performance status parameters refer to the key performance indicators of the modular gamma probe unit, which can specifically include detection efficiency, system spatial resolution, energy resolution, and internal noise parameters. Detection efficiency can be understood as the sensitivity of the probe in capturing radiated photons. It can be achieved through standard radiation source calibration or Monte Carlo simulation calculation, with the aim of quantifying signal capture capability and providing a benchmark for subsequent calibration. System spatial resolution can be understood as the system's ability to distinguish adjacent targets in space. It can be achieved by scanning a standard test sample with a fine structure, such as a lead slit array or a star target. After the signal distribution recorded by the probe array is processed by an image reconstruction algorithm, the half-width at half-maximum (WHM) of its point spread function or line spread function is analyzed to characterize the spatial resolution. Its purpose is to measure and characterize the system's ability to distinguish and identify two closely adjacent independent targets or details in space. Energy resolution can be understood as the probe's ability to distinguish photons of different energies. It can be achieved through energy spectrum peak analysis or Gaussian fitting algorithms, with the aim of optimizing the energy selection window to reduce scattering interference. Internal noise parameters can be understood as a measure of the probe's background interference level. It can be achieved through dark count statistics or time-domain noise modeling, with the aim of assessing the degree of noise contamination on the signal. Specifically, calibration refers to the systematic correction of the original counting data based on performance status parameters. This can be achieved using linear compensation models based on detection efficiency, dynamic energy window adjustment strategies based on energy resolution, and adaptive filtering algorithms based on internal noise. The aim is to eliminate signal weakening or distortion caused by individual differences in probe units, making the calibrated data more accurately reflect the tracer distribution. In practical applications, spatial consistency verification refers to checking the coherence of mesoscale features in adjacent regions. This can be achieved through correlation analysis of inter-regional eigenvalues ​​or sliding window consistency checks, aiming to avoid feature breaks caused by local noise or outliers. Correction of microscale features refers to enhancing boundary contrast or suppressing high-frequency noise. This can be achieved through nonlinear edge enhancement filters or local contrast adaptive algorithms, aiming to ensure that sharp boundary information remains clearly discernible even in anomalous diffusion modes.

[0032] Specifically, the proposed solution first evaluates the performance parameters of each modular gamma probe unit to provide a quantitative basis for calibration. Then, based on these parameters, the original counting data is calibrated to eliminate signal strength deviations caused by uneven detection efficiency, scattering interference caused by differences in energy resolution, and background contamination caused by abnormal internal noise parameters. When extracting multi-scale signal features, the calibrated data ensures that macro-scale features accurately reflect the overall signal coverage, meso-scale features reliably characterize the distribution of diffusion centers, and micro-scale features accurately capture boundary details. Furthermore, spatial consistency verification is performed on meso-scale features by dynamically comparing feature values ​​in adjacent regions to identify and correct local discontinuities, avoiding interference from random fluctuations in diffusion center identification. Simultaneously, micro-scale features are corrected by suppressing high-frequency noise and enhancing boundary contrast to prevent blurred boundaries in abnormal diffusion areas from being misjudged as normal background. These steps form a strict temporal logic chain: the evaluation stage provides input for calibration, the calibration stage ensures the accuracy of feature extraction, and the feature verification and correction stages ensure the reliability of multi-scale features in the spatial dimension, thereby maintaining the consistency of multi-scale signal features even under conditions of non-uniform probe unit performance.

[0033] As one specific implementation method: In step S11, during startup or routine maintenance of the nuclear medicine imaging system, a probe performance evaluation procedure is executed. For example, a uniformly distributed radiation source of known activity is placed at a predetermined position in front of the gamma probe array. Each modular gamma probe unit acquires gamma photon count data over a fixed time period. The main control unit analyzes the total count of each probe unit within a specific energy window to calculate its detection efficiency. Simultaneously, by analyzing the gamma photon energy spectrum acquired by each probe unit, the center energy and full width at half maximum (FWHM) of the photoelectric peak are identified, thereby calculating the energy resolution. Furthermore, in the absence of radiation source input, the background count rate of each probe unit is measured to obtain its internal noise parameters. These evaluation results, such as the detection efficiency coefficient, energy resolution value, and noise count rate, are stored in system memory and correlated with the spatial location of each probe unit.

[0034] In step S12, during the actual imaging scan, when the modular gamma probe array acquires the raw gamma photon counting data stream in real time, the main control unit calls the performance status parameters of each probe unit stored in S11. For example, for the raw count rate reported by each probe unit, the system normalizes it according to its detection efficiency coefficient to compensate for the differences in detection efficiency between different probe units. For energy spectrum data, the system applies an energy spectrum broadening or shrinking correction algorithm according to the energy resolution parameter of each probe unit to ensure the comparability of the energy spectra of different probe units. At the same time, background noise is subtracted from the raw counting data according to the internal noise parameter. After these calibration processes, a set of spatially more consistent and accurate calibrated counting data is obtained.

[0035] Step S13: Based on the calibrated counting data obtained in S12, the main control unit begins to extract multi-scale features: Macroscale characteristics: The overall diffusion range index and average intensity are calculated by summing or averaging the calibrated count data over the entire region of interest.

[0036] Mesoscale features: A gray-level co-occurrence matrix algorithm is applied to the calibrated count data of local regions to extract texture features such as contrast, energy, and entropy. Simultaneously, the Canny operator is used to identify potential edges, and the gradient width is calculated as a boundary blurring index. Subsequently, spatial consistency verification is performed on these mesoscale features. For example, the boundary blurring index and texture features of adjacent probe units or adjacent regions are compared. If a significant deviation is found between the feature values ​​of a local region and the surrounding region (e.g., exceeding a preset threshold), smoothing or interpolation correction is performed to ensure spatial consistency, resulting in spatially consistent verified mesoscale features.

[0037] Microscale characteristics: For each probe unit, the Poisson distribution bias is calculated based on post-calibration count data within a short time window (e.g., 20 milliseconds). Simultaneously, a real-time energy spectrum histogram is constructed, and the relative proportions of the photoelectric peak and the Compton scattering peak are calculated as the scattering ratio. These microscale characteristics are then corrected. For example, the shape of the Compton scattering peak in the energy spectrum is corrected based on the energy resolution parameters evaluated in S11 to more accurately quantify the scattered radiation. Statistical corrections are applied to the Poisson distribution bias to eliminate minor biases that may have been introduced during calibration, resulting in the corrected microscale characteristics.

[0038] Through the above technical solutions, this application effectively ensures the spatial consistency of multi-scale signal features when there is non-uniformity or degradation in detection efficiency, system spatial resolution, energy resolution, or internal noise in a single modular gamma probe unit, making the characterization of abnormal diffusion acquisition patterns more accurate. Specifically, the calibration process eliminates signal distortion caused by individual differences in probe units, ensuring that macro-scale features truly reflect the overall signal coverage, meso-scale features reliably identify the diffusion center distribution, and micro-scale features accurately capture boundary details. Spatial consistency verification avoids local noise interference in diffusion center identification, while micro-scale feature correction prevents boundary blurring from being misjudged as normal background, thereby preventing the adaptive control logic from generating suboptimal probe positioning commands due to feature distortion, ultimately ensuring the quality of image reconstruction.

[0039] In some embodiments, the specific steps in step S2 include: S21. Continuously acquire and store multi-scale signal feature sequences at multiple time points; S22. By performing time series analysis on multi-scale signal feature sequences, the evolution trends corresponding to macro-scale features, meso-scale features, and micro-scale features are obtained; S23. Based on the evolution trend, adjust the identification criteria of the ingestion pattern to obtain the adjusted identification criteria; S24. Based on the adjusted identification criteria, identify the uptake pattern of the current tracer distribution.

[0040] The process of continuously acquiring and storing multi-scale signal feature sequences at multiple time points refers to the system periodically collecting and saving historical records of signal features. This can be achieved using a circular buffer or a distributed time series database, aiming to establish a continuous data foundation in the time dimension and avoid misjudgments caused by instantaneous fluctuations in data at a single time point. Time series analysis can be understood as applying statistical models or machine learning algorithms to analyze the patterns of feature changes. This can be achieved using methods such as autoregressive models or wavelet transforms, aiming to quantify the overall signal coverage evolution reflected by macro-scale features, the diffusion center movement patterns revealed by meso-scale features, and the boundary sharpness changes reflected by micro-scale features. Adjusting the identification criteria for the acquisition mode refers to dynamically modifying the identification threshold or rules according to the evolution trend. This can automatically scale the criteria parameters based on the trend analysis results, aiming to make the identification standard respond to the signal evolution state in real time, for example, lowering the identification threshold when the diffusion degree increases rapidly. Pattern recognition based on the adjusted identification criteria refers to comparing the current features with dynamic criteria, which can be achieved using a classifier or threshold comparator, aiming to ensure that the identification results are synchronized with the real-time signal evolution and avoid the rigidity problem of fixed criteria in dynamic scenarios.

[0041] Specifically, the proposed solution first continuously acquires and stores multi-scale signal feature sequences at multiple time points, forming a time-dimensional dataset. Then, through time-series analysis, it extracts the evolution trends of macro-scale, meso-scale, and micro-scale features, such as identifying acceleration, deceleration, or stable states of feature changes. Next, it dynamically adjusts the identification criteria for anomalous diffusion patterns based on these trends; for example, it lowers the identification threshold when the trend shows an accelerated increase in diffusion to avoid misclassifying anomalous diffusion as normal background. Finally, it identifies the current tracer distribution based on the adjusted criteria, thereby achieving accurate capture of anomalous patterns. This process forms a closed-loop dynamic adjustment mechanism, ensuring that the identification criteria can respond to signal evolution in real time, avoiding the limitations of instantaneous snapshot-based identification methods in dynamic scenarios, and providing a reliable basis for subsequent probe array positioning decisions.

[0042] As one specific implementation method: Step S21: Continuously acquire and store multi-scale signal feature sequences at multiple time points: During nuclear medicine imaging scanning, the main control unit receives real-time counting data from the modular gamma probe array at a frequency of 10 times per second. After each reception, the digital signal processor (DSP) immediately calculates and extracts macroscopic scale features (e.g., diffusion range index, average intensity), mesoscopic scale features (e.g., boundary blur index, GLCM texture features), and microscopic scale features (e.g., Poisson distribution bias, scattering ratio) at the current time point. These feature values ​​are packaged into a feature vector and stored together with a timestamp in the system's cache or non-volatile memory, forming a continuous feature sequence. For example, a feature vector is stored every 100 milliseconds, continuing throughout the entire scan cycle.

[0043] Step S22 involves performing time series analysis on the multi-scale signal feature sequences to obtain the evolution trends of macro-scale, meso-scale, and micro-scale features. The time series analysis module in the main control unit processes the feature sequences stored in S21. For example, for the diffusion range index, the system can calculate its moving average and rate of change over the past 5 seconds; for the boundary ambiguity index, it can calculate its linear regression slope over the past 10 seconds; and for the scattering ratio, it can calculate its exponential smoothing value over the past 3 seconds. These calculation results, such as the average rate of change of diffusion range, the trend direction of boundary ambiguity, and the short-term fluctuation amplitude of scattering ratio, constitute the evolution trends of features at each scale.

[0044] Step S23: Based on the evolution trend, adjust the identification criteria for the ingestion pattern to obtain the adjusted identification criteria: The weight adjustment unit receives the evolution trend obtained in S22. For example, if the evolution trend of the diffusion range index shows that it is continuously expanding, and the evolution trend of the boundary blur index shows that it is continuously increasing, the weight adjustment unit will adjust the identification criteria for the abnormal diffusion pattern accordingly. Specifically, it may lower the discrimination threshold for "boundary sharpness" and increase the weights for "diffusion range" and "scattering ratio" to more leniently identify evolving diffusion patterns. These adjusted criteria are updated and stored.

[0045] Step S24: Based on the adjusted identification criteria, determine whether the current tracer distribution is an anomalous diffuse uptake pattern. The main control unit uses the updated identification criteria from S23 to judge the multi-scale signal features acquired in real time. For example, it compares the feature values ​​such as the diffusion range index, boundary blur index, and scattering ratio at the current time point with the adjusted criteria. If the current feature values ​​meet the conditions of the adjusted criteria (e.g., the diffusion range index is higher than the new threshold, and the scattering ratio is higher than the new threshold), the system identifies the current tracer distribution as an anomalous diffuse uptake pattern.

[0046] Through the above technical solutions, the system can effectively capture the dynamic evolution of abnormal diffuse uptake patterns and dynamically adjust the identification criteria, thereby avoiding the identification results being lagging or inconsistent with the actual situation, ensuring the accurate identification of tracer distribution, and providing a reliable basis for the positioning decision of the subsequent modular gamma probe array.

[0047] In some embodiments, the specific steps in step S22 include: S221. Within a continuous time window, calculate the changes (e.g., rate of change) of each scale feature in the multi-scale signal feature sequence, and obtain the statistical analysis results by performing statistical analysis on the rate of change; S222. Based on the statistical analysis results, by identifying the changes (acceleration or deceleration) or stable states of features at each scale, the evolution trends corresponding to macro-scale features, meso-scale features, and micro-scale features are obtained.

[0048] Among them, the continuous time window refers to the time period used for analyzing the dynamic behavior of the signal within a local time range. It can be implemented using a fixed-length sliding window or a mechanism that adaptively adjusts the window length according to the dynamic characteristics of the signal. The purpose is to avoid the loss of details caused by global time series analysis and focus on the signal change characteristics within a local time range. The rate of change refers to the degree of change of multi-scale signal features per unit time. It can be implemented using first-order difference, moving average difference, or differential calculation methods based on digital filters. The purpose is to quantify the instantaneous change speed of signal features and provide basic data for subsequent statistical analysis. The statistical analysis result refers to the feature quantity obtained after statistical processing of the rate of change. It can be implemented by calculating the mean, variance, kurtosis of the rate of change, or parameter estimation based on probability distribution models. The purpose is to distinguish between short-term sharp fluctuations and long-term stable trends of the signal and provide a more reliable basis for trend identification. Acceleration, deceleration, or steady state refers to the change state of the signal feature change speed. It can be implemented using simple judgment based on thresholds, state machine models, or lightweight machine learning classifiers. The purpose is to dynamically distinguish whether the signal is in a rapid change, slow change, or steady stage, thereby accurately capturing the evolution trend.

[0049] Specifically, the proposed solution calculates the rate of change of multi-scale signal features within a continuous time window, focusing time series analysis on a local time range and avoiding the loss of details caused by global analysis. Subsequently, statistical analysis is performed on the rate of change, distinguishing between short-term sharp fluctuations and long-term stable trends based on the instantaneous fluctuation characteristics of the rate of change. Finally, the acceleration, deceleration, or stable state of each scale feature is identified based on the statistical analysis results, dynamically determining the stage of signal change. This step-by-step processing mechanism enables the system to effectively filter transient non-physiological signal fluctuations while preserving true physiological evolutionary trend characteristics. It is particularly suitable for capturing the nonlinear, multi-stage, or irregular dynamic characteristics exhibited in the evolution of diffuse uptake patterns, thus providing a reliable basis for obtaining the evolutionary trends corresponding to macro-scale, meso-scale, and micro-scale features.

[0050] As one specific implementation method: In step S221, the system's main control unit receives the macroscopic, mesoscopic, and microscopic multi-scale signal feature sequences from step S21. Assume one macroscopic feature is the "diffusion range index," one mesoscopic feature is the "boundary ambiguity index," and one microscopic feature is the "scattering ratio." The system sets a continuous time window, for example, 5 seconds. Within each 5-second time window, the system calculates the average rate of change for these features. For example, for the diffusion range index, the difference between the current window's end value and the window's beginning value is calculated divided by the window duration. Subsequently, the system performs statistical analysis on these rates of change, for example, calculating the moving average and standard deviation of the diffusion range index's rate of change over the past 30 seconds, obtaining a statistical analysis result.

[0051] In step S222, the main control unit determines the evolution status of each scale feature based on the statistical analysis results obtained in step S221. For example, if the moving average rate of change of the diffusion range index is consistently higher than a certain positive threshold (e.g., 0.05 units / second) and the standard deviation is lower than a certain threshold, it is identified as an "accelerated expansion" state; if the rate of change is close to zero and the standard deviation is small, it is identified as a "stable" state; if the rate of change is consistently lower than a certain negative threshold, it is identified as an "accelerated contraction" state. In this way, the system obtains the evolution trends corresponding to the diffusion range index, boundary blurring index, and scattering ratio, such as "accelerated expansion of diffusion range," "stable boundary blurring," and "slow decrease in scattering ratio."

[0052] Through the above scheme, this application can accurately distinguish between transient non-physiological signal fluctuations and real physiological evolution trends in nuclear medicine imaging, effectively capture the nonlinear, multi-stage or irregular dynamic characteristics presented by the evolution of diffuse uptake patterns, thereby improving the accuracy of obtaining multi-scale signal feature evolution trends and providing a reliable basis for the identification of abnormal diffuse uptake patterns.

[0053] In some embodiments, step S221, which involves obtaining statistical analysis results by performing statistical analysis on the rate of change, includes: S2211. Real-time monitoring of the instantaneous fluctuation amplitude and duration of the rate of change; S2212. Identify the local dynamic activity level of the rate of change based on the instantaneous fluctuation amplitude and duration; S2213. Adjust the length of the continuous time window and the overlap of the time windows used for statistical analysis according to the local dynamic activity level; S2214. Using the adjusted continuous time window length and time window overlap, perform statistical analysis on the rate of change to obtain the statistical analysis results.

[0054] In practical applications, real-time monitoring of the instantaneous fluctuation amplitude and duration of the rate of change refers to the quantitative characterization of the instantaneous dynamic characteristics of the rate of change of multi-scale signals. This can be achieved using a sliding window algorithm combined with peak detection technology. By calculating the extreme values ​​and duration periods of signal changes per unit time, the aim is to capture the sudden fluctuation characteristics of the signal in the local time domain. Identifying the local dynamic activity level of the rate of change can be understood as classifying the dynamic state of the signal based on the correlation between fluctuation amplitude and duration. This can be implemented using a threshold comparator or a fuzzy logic classifier. By setting two-dimensional amplitude-time judgment rules, the aim is to quantify the dynamic characteristics of the signal into operable activity levels. The activity index; specifically, adjusting the continuous time window length and time window overlap refers to dynamically optimizing the time parameter configuration for statistical analysis. This can be achieved using a parameter adaptive adjustment module combined with a preset rule base. By establishing a mapping relationship between the activity index and the window parameters, the goal is to make the statistical window match the actual dynamic characteristics of the signal. In practical applications, using the adjusted continuous time window length and time window overlap for statistical analysis refers to performing statistical operations based on dynamically optimized parameters. This can be achieved using recursive filtering algorithms or weighted average calculations. By applying window parameters that are adjusted in real time, the goal is to generate statistical results that have a balanced sensitivity to transient fluctuations and long-term trends.

[0055] Specifically, the proposed solution provides fundamental data support for identifying the level of local dynamic activity by real-time monitoring of the instantaneous fluctuation amplitude and duration of the rate of change. Since the fluctuation amplitude directly reflects the severity of signal change and the duration characterizes the stability of change, the correlation analysis based on the two can accurately quantify the local dynamic characteristics of the signal. Furthermore, the continuous time window length and time window overlap are dynamically adjusted according to the identified level of local dynamic activity. When the activity level is high, the window length is shortened to improve the response speed to transient fluctuations and the overlap is increased to ensure temporal continuity. When the activity level is low, the window length is extended to enhance the ability to capture long-term trends and reduce overlap to avoid data redundancy. Based on this, the adjusted window parameters are used to perform statistical analysis on the rate of change, enabling the statistical process to adaptively balance noise suppression and feature preservation, thereby ensuring that the acquisition of evolution trends is not affected by instantaneous noise and can accurately reflect real physiological changes. This technical solution establishes a closed-loop mechanism of "monitoring-identification-adjustment-analysis," which dynamically matches the statistical analysis process with the dynamic characteristics of the signal, effectively solving the sensitivity imbalance problem caused by fixed window parameters under abnormal diffuse ingestion mode.

[0056] As one specific implementation method: In step S2211, the digital signal processor (DSP) in the system samples the rate of change of macroscopic, mesoscopic, and microscopic scale features reported by each modular gamma probe unit at 10-millisecond intervals. For each sampling point, the DSP calculates the deviation from the average of the previous 5 sampling points as the instantaneous fluctuation amplitude, and records the number of consecutive sampling points where the amplitude exceeds a preset threshold (e.g., 10% of the average rate of change) as the duration.

[0057] In step S2212, the DSP determines the local dynamic activity level of the rate of change based on the data monitored in S2211. For example, if the instantaneous fluctuation amplitude exceeds the threshold for 100 consecutive milliseconds, it is identified as "high activity"; if the fluctuation amplitude is consistently below the threshold for more than 500 milliseconds, it is identified as "low activity".

[0058] In step S2213, when the DSP identifies "high activity", it adjusts the length of the continuous time window used to calculate the statistical analysis results from the default 500 milliseconds to 200 milliseconds, and adjusts the overlap of the time window from the default 50% to 80%. When it identifies "low activity", it adjusts the length of the time window to 1000 milliseconds and the overlap to 30%.

[0059] In step S2214, the DSP uses the adjusted time window parameters from S2213 to calculate a moving average on the rate of change data, obtaining a smoothed statistical analysis result. For example, during the "high activity" period, a 200-millisecond window and 80% overlap are used to calculate the moving average; during the "low activity" period, a 1000-millisecond window and 30% overlap are used to calculate the moving average.

[0060] Through the above technical solution, this application enables the statistical analysis process to adaptively match the dynamic characteristics of the evolution of abnormal diffuse uptake patterns. When facing abnormal diffuse uptake in benign liver tissue during liver tumor examination, the system can accurately identify the transient fluctuations and long-term trends of tracer distribution, avoiding misjudging diffuse signals as larger areas of interest. This ensures that the positioning decision of the modular gamma probe array is based on the real physiological evolution trend, effectively improving the accuracy of target organ boundary identification.

[0061] In some embodiments, the specific steps in step S3 include: S31. Identify the degree of spatiotemporal heterogeneity distribution within the ingestion pattern; the degree of spatiotemporal heterogeneity distribution includes the signal-to-noise ratio or structural clarity of local regions within the region of interest; S32. Based on the degree of spatiotemporal heterogeneity, the weights of multi-scale signal features in the localization decision-making of the modular gamma probe array are locally adjusted. This local adjustment includes: S321. In local regions where the degree of spatiotemporal heterogeneity is higher than the first preset value and the signal-to-noise ratio is lower than the second preset value, increase the first weight and decrease the third weight; S322. In local regions where the degree of spatiotemporal heterogeneity distribution is lower than the third preset value or the structural clarity is lower than the fourth preset value (i.e., weak structure), keep the second and third weights unchanged and reduce the first weight.

[0062] In practical applications, the degree of spatiotemporal heterogeneity distribution refers to the temporal and spatial variation characteristics of signal properties within an abnormal diffusion area. It can be achieved using image segmentation algorithms or local statistical analysis methods. By performing cluster analysis on the pixel-level signal intensity distribution of each local region within the diffusion area, the aim is to accurately perceive the signal quality differences in different regions. Localized adjustment can be understood as a regionalized weight allocation strategy, which can be achieved using a spatial weight mapping table or a dynamic region division mechanism. By dividing the coverage area of ​​the gamma probe array into multiple sub-regions and independently setting weight parameters, the aim is to dynamically match the weight adjustment with the local signal characteristics, avoiding the problem of insufficient regional adaptability caused by global adjustment.

[0063] Specifically, the proposed solution first identifies the degree of spatiotemporal heterogeneity within the anomalous diffusion acquisition pattern, obtaining signal-to-noise ratio (SNR) or structural clarity information for each local region within the diffusion area. Then, based on this distribution information, it differentiates the weights of multi-scale signal features in the localization decision based on temporal and spatial dimensions. In local regions with high diffusion and low SNR, the system increases the first weight of macro-scale features related to overall signal coverage in the localization decision to enhance the overall capture capability of the diffusion signal, while decreasing the third weight of micro-scale features related to sharp boundaries to suppress noise interference introduced by boundary ambiguity. In local regions with low diffusion or low structural clarity, the system maintains the second weight of meso-scale features related to the diffusion center and the third weight of micro-scale features related to sharp boundaries to maintain sensitivity to key structures, while decreasing the first weight of macro-scale features related to overall signal coverage to reduce background interference. This regional weight adjustment mechanism ensures that the localization decision can dynamically respond to spatial changes in signal distribution, forming an adaptive optimization path for different regional characteristics.

[0064] As a specific implementation, suppose that during liver nuclear medicine imaging of a patient, the system identifies an abnormal diffuse uptake pattern around a liver tumor. Within this diffuse region, spatiotemporal heterogeneity is identified through step S31. Specifically, in a local region A at the tumor edge, the diffusion level is higher than a first preset value (e.g., diffusion range index higher than 0.8) and the signal-to-noise ratio is lower than a second preset value (e.g., scattering ratio higher than 0.4), indicating that this region is highly blurred and noisy. In another local region B within the tumor, although it also exhibits a diffuse pattern, its diffusion level is lower than a third preset value (e.g., diffusion range index lower than 0.6) and its structural clarity is higher than a fourth preset value (e.g., boundary blurring index lower than 0.5), indicating that this region still contains some weak structural information.

[0065] According to the localization adjustment logic of step S32: For a local region A (high diffusion, low signal-to-noise ratio), the system executes sub-step S321: increasing the weight of overall signal coverage (e.g., adjusting the weight of macro-scale features from the default 0.3 to 0.55) and reducing the dependence on sharp boundaries (e.g., adjusting the weight of sharp boundary features at the micro-scale from the default 0.4 to 0.05). This means that when making probe positioning decisions, the system instructs the probe array to scan the region with a wider field of view, prioritizing the capture of the overall signal in the blurred region rather than trying to find non-existent sharp edges in the noise.

[0066] For local region B (with low diffusion or weak structure), the system executes sub-step S322: maintaining the same level of attention to the diffusion center (second weight) and sharp boundaries (third weight) (e.g., the second weight remains at 0.4, and the third weight remains at 0.1), while appropriately reducing the weight of overall signal coverage (e.g., adjusting the weight of macroscopic features from the default 0.3 to 0.25). This means that when making probe positioning decisions, the system instructs the probe array to attempt to focus on local weak structures in this region while ensuring basic coverage, in order to obtain more refined structural information and avoid losing potential diagnostic details due to over-smoothing.

[0067] Through this localized adjustment, the system can generate probe positioning and angle adjustment instructions for different local areas. For example, in area A, the probe may be instructed to scan with a large field of view and a long dwell time, while in area B, the probe may be instructed to detect with a smaller field of view and a finer scanning path.

[0068] Through the above scheme, this application can implement a refined weight adjustment strategy for the non-uniformity within the abnormal diffuse uptake pattern, optimize the balance between signal coverage and noise control in low signal-to-noise ratio regions, and maintain sensitivity to key anatomical features in weak structural regions, thereby effectively avoiding excessive smoothing in local areas, improving the accuracy of target organ boundary recognition and the reconstruction quality of the final diagnostic image.

[0069] In some embodiments, the specific steps in step S4 include: S41. After completing the weight adjustment, determine the desired spatial sensitivity distribution or target coverage area of ​​the modular gamma probe array; S42. The desired spatial sensitivity distribution or target coverage area is geometrically mapped to the position and angle parameters (including the number of angles) of each modular gamma probe in the modular gamma probe array. S43. Generate adjustment instructions based on position and angle parameters.

[0070] The desired spatial sensitivity distribution refers to the distribution of the modular gamma probe array's response capability to radioactive signals in three-dimensional space. It can be achieved using a sensitivity heatmap generated based on weighted allocation results, aiming to transform the abstract decision of multi-scale signal characteristics into a quantifiable spatial imaging target. The target coverage area can be understood as the boundary range of the anatomical region requiring high-precision imaging, which can be represented using a three-dimensional surface model or voxel mesh. The purpose is to clearly define the focusing range of the probe array to avoid misjudgment caused by signal boundary ambiguity. Geometric mapping specifically refers to the mathematical transformation process of converting spatial targets into probe physical parameters, which can be achieved using coordinate transformation matrices or ray tracing algorithms. The purpose is to ensure that the probe layout strictly matches the geometric characteristics of the target area. Position parameters and angle parameters refer to the coordinate position and orientation of each modular gamma probe in space, which can be represented using position vectors and rotation matrices in a Cartesian coordinate system. The purpose is to accurately describe the physical pose of the probe. Adjustment commands can be understood as control signals that drive the probe to perform positioning actions. They can be achieved using digital communication protocols or pulse signal sequences, aiming to reliably transmit position and angle adjustment information.

[0071] Specifically, the solution of this application first determines the desired spatial sensitivity distribution or target coverage area after weight adjustment, transforming the weight allocation results of multi-scale signal features into a clear spatial imaging target, thereby accurately defining the key areas requiring high-sensitivity imaging. Subsequently, through geometric mapping, this spatial target is precisely converted into specific position and angle parameters for each probe. This process fully considers the spatial geometric constraints of the probe array and the morphological characteristics of the target area, ensuring the mathematical accuracy of the parameter conversion. Finally, adjustment instructions are generated based on the generated position and angle parameters, driving the probes to perform physical adjustment actions consistent with the geometric characteristics of the target area, enabling the probe array to dynamically focus on the true contour of the target organ, effectively eliminating the mismatch between abstract decision-making and physical layout, and reducing the ineffective dwell time of the probes in non-target areas.

[0072] As a preferred embodiment: In nuclear medicine imaging systems, it is assumed that a modular gamma probe array consists of N independent probe modules, each with an adjustable field of view and three-dimensional motion capability.

[0073] In step S41, after receiving the adjusted weight parameters (e.g., first weight = 0.5, second weight = 0.4, third weight = 0.1) output by the weight adjustment unit, the system's main control unit calculates a desired spatial sensitivity distribution map based on these weights and the preliminary geometric information of the currently identified diffusion region (e.g., the center coordinates and approximate range of the diffusion region). This distribution map is a three-dimensional grid, where each grid point represents the desired detection sensitivity at that location. For example, the desired sensitivity value is highest at the geometric center of the diffusion region and gradually decreases outwards, but the rate of decrease is more gradual than in the conventional mode, reflecting enhanced attention to "overall signal coverage" and "diffusion center," while avoiding pursuing excessively high local sensitivity at the diffusion edge.

[0074] In step S42, the main control unit runs a geometric mapping algorithm. This algorithm takes the desired spatial sensitivity distribution map defined in S41 as input and considers the physical characteristics (e.g., detector crystal size, collimator type, intrinsic sensitivity function) and kinematic constraints (e.g., maximum range of motion, minimum safe distance, maximum rotation angle) of each probe module. The goal of the algorithm is to calculate a set of optimal position coordinates (X_i, Y_i, Z_i) and angle parameters (pitch angle_i, yaw angle_i) for each probe module (e.g., probe module 1 to probe module N) such that the actual spatial sensitivity distribution of the entire probe array matches the desired spatial sensitivity distribution to the maximum. For example, the algorithm may iteratively optimize and adjust the position and angle of each probe module until the mean square error between the array's overall detection response function and the desired distribution is minimized.

[0075] In step S43, the main control unit takes the position coordinates (X_i, Y_i, Z_i) and angle parameters (pitch angle_i, yaw angle_i) of each probe module calculated in S42 as input to generate a series of adjustment commands for the probe array motion controller. These commands are standardized motion control commands; for example, for a stepper motor driver, the commands may include the number of pulses at the target position, the movement speed, the acceleration curve, etc.; for a servo motor driver, the commands may include the encoder value at the target position, PID control parameters, etc. These commands are packaged into data packets and sent to the positioning and driving device of the probe array through a high-speed communication interface (e.g., EtherCAT or CANopen).

[0076] Through the above scheme, this application can accurately convert the weight adjustment results of multi-scale signal features into physical adjustment instructions for modular gamma probe arrays, avoid misjudgment of coverage caused by signal boundary ambiguity in abnormal diffusion acquisition mode, reduce the ineffective dwell time of probe in non-target area, significantly reduce the capture probability of scattered radiation, and thus provide a high-quality data foundation for image reconstruction.

[0077] Reference Appendix Figure 2 This invention provides an adaptive control system for nuclear medicine imaging based on a modular gamma probe, comprising: The first acquisition module 100 is used to acquire multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array; The identification module 200 is used to identify the uptake pattern of the current tracer distribution based on multi-scale signal characteristics; The adjustment module 300 is used to dynamically adjust the weight of multi-scale signal features in the positioning decision of the modular gamma probe array based on the characteristics of the identified acquisition pattern. The generation module 400 is used to generate and execute adjustment instructions for the modular gamma probe array after the weight adjustment is completed; the adjustment instructions are used to adjust the position, angle and number of angles of the modular gamma probe. The second acquisition module 500 is used to control the modular gamma probe to acquire data and obtain image data after executing the adjustment command; Reconstruction module 600 is used to reconstruct diagnostic images based on image data.

[0078] It should be noted that, to achieve the above control, the nuclear medicine imaging adaptive control system based on modular gamma probes includes an adaptive adjustment hardware structure. Specifically, it includes multiple modular gamma probes and a SPECT host with multiple rotatable frames. Each modular gamma probe includes an adaptive multi-pinhole collimator and a high-resolution gamma detector. The adaptive multi-pinhole collimator is a movable structure with multiple switching structures interconnected by drive motors. The high-resolution gamma detector is fixed to one end of one of the switching structures. The rotatable frame includes a stator and multiple rotors. Each rotor is fixedly connected to the modular gamma probe. The SPECT host includes a motion control module and an imaging control module. The motion control module is used to drive and control the drive motors to realize the movement of the modular gamma probe around the rotors, and to drive and control the rotors to realize the movement of the modular gamma probe around the stator. The imaging control module is used to control the high-resolution gamma detector to achieve nuclear medicine imaging (the specific structures described in this embodiment are all prior art and will not be repeated here).

[0079] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0080] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An adaptive control method for nuclear medicine imaging based on a modular gamma probe, characterized in that, Includes the following steps: S1. Obtain multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array; the multi-scale signal features include macro-scale features, meso-scale features and micro-scale features; S2. Identify the uptake pattern of the current tracer distribution based on multi-scale signal characteristics; S3. Based on the characteristics of the identified acquisition pattern, dynamically adjust the weight of multi-scale signal features in the positioning decision of the modular gamma probe array; S4. After completing the weight adjustment, generate and execute adjustment instructions for the modular gamma probe array; the adjustment instructions are used to adjust the position and angle of the modular gamma probe. S5. After executing the adjustment command, control the modular gamma probe to acquire data and obtain image data; S6. Reconstruct the diagnostic image based on the image data; The specific steps in step S1 include: S11. Evaluate the performance status parameters of each modular gamma probe unit in the modular gamma probe array; S12. Based on the performance status parameters, the raw counting data obtained from the real-time counting data stream acquired from the modular gamma probe array is calibrated to obtain calibrated counting data; S13. Based on the calibrated count data, extract macro-scale features, meso-scale features and micro-scale features, and perform spatial consistency verification on the extracted meso-scale features to obtain spatially consistent meso-scale features, and correct the extracted micro-scale features to obtain corrected micro-scale features.

2. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 1, characterized in that, Performance parameters include detection efficiency, system spatial resolution, energy resolution, and internal noise parameters.

3. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 1, characterized in that, The specific steps in step S2 include: S21. Continuously acquire and store multi-scale signal feature sequences at multiple time points; S22. By performing time series analysis on multi-scale signal feature sequences, the evolution trends corresponding to macro-scale features, meso-scale features, and micro-scale features are obtained; S23. Based on the evolution trend, adjust the identification criteria of the ingestion pattern to obtain the adjusted identification criteria; S24. Based on the adjusted identification criteria, identify the uptake pattern of the current tracer distribution.

4. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 3, characterized in that, The specific steps in step S22 include: S221. Within a continuous time window, calculate the changes in features at each scale in a multi-scale signal feature sequence, and obtain statistical analysis results by performing statistical analysis on the changes; S222. Based on the statistical analysis results, by identifying the changes or stable states of features at each scale, the evolution trends corresponding to macro-scale features, meso-scale features, and micro-scale features are obtained.

5. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 4, characterized in that, In step S221, the steps of obtaining statistical analysis results by performing statistical analysis on the changes include: S2211. Real-time monitoring of the instantaneous fluctuation amplitude and duration of changes; S2212. Identify the degree of local dynamic activity based on the instantaneous fluctuation amplitude and duration; S2213. Adjust the length of the continuous time window and the overlap of the time windows used for statistical analysis according to the local dynamic activity level; S2214. Using the adjusted continuous time window length and time window overlap, perform statistical analysis on the changes and obtain the statistical analysis results.

6. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 1, characterized in that, Dynamic adjustment includes adjusting the first weight of macro-scale features related to overall signal coverage in positioning decisions, the second weight of meso-scale features related to diffusion centers in positioning decisions, and the third weight of micro-scale features related to sharp boundaries in positioning decisions.

7. The adaptive control method for nuclear medicine imaging based on a modular gamma probe according to claim 6, characterized in that, The specific steps in step S3 include: S31. Identify the degree of spatiotemporal heterogeneity distribution within the ingestion pattern; the degree of spatiotemporal heterogeneity distribution includes the signal-to-noise ratio or structural clarity of local regions within the region of interest; S32. Based on the degree of spatiotemporal heterogeneity, the weights of multi-scale signal features in the localization decision-making of the modular gamma probe array are locally adjusted. This local adjustment includes: S321. In local regions where the degree of spatiotemporal heterogeneity is higher than the first preset value and the signal-to-noise ratio is lower than the second preset value, increase the first weight and decrease the third weight; S322. In local regions where the degree of spatiotemporal heterogeneity distribution is lower than the third preset value or the structural clarity is lower than the fourth preset value, keep the second and third weights unchanged and reduce the first weight.

8. A nuclear medicine imaging adaptive control system based on a modular gamma probe, employing the nuclear medicine imaging adaptive control method based on a modular gamma probe as described in any one of claims 1-7, characterized in that, include: The first acquisition module is used to acquire multi-scale signal features from the real-time counting data stream acquired by the modular gamma probe array; The identification module is used to identify the uptake pattern of the current tracer distribution based on multi-scale signal characteristics; The adjustment module is used to dynamically adjust the weight of multi-scale signal features in the positioning decision of the modular gamma probe array based on the characteristics of the identified acquisition pattern. The generation module is used to generate and execute adjustment instructions for the modular gamma probe array after the weight adjustment is completed; the adjustment instructions are used to adjust the position and angle of the modular gamma probe. The second acquisition module is used to control the modular gamma probe to acquire data and obtain image data after executing the adjustment command; The reconstruction module is used to reconstruct diagnostic images from image data.

Images