Low-field magnetic resonance head imaging radio frequency coil system

Abstract

The invention relates to the technical field of magnetic resonance imaging, in particular to a low-field magnetic resonance head imaging radio-frequency coil system, which comprises a flexible array coil sensing module for acquiring head magnetic resonance original radio-frequency signals, coil coupling state data and environment interference parameters; the dynamic field intensity calibration module constructs a dynamic magnetic field distribution model, corrects low-field magnetic field uniformity deviation in real time through a multi-source magnetic field compensation algorithm, and outputs magnetic field calibration data; the intelligent imaging optimization center identifies low-field imaging artifacts by using an improved CNN-Transform model according to the original radio frequency signal and the magnetic field calibration data, locates a distortion source and generates an optimization scheme; the three-dimensional imaging reconstruction unit builds a head three-dimensional imaging model, fuses an anatomical structure and signal distribution, and outputs a preliminary imaging result; the self-adaptive radio frequency regulation and control unit predicts coil performance attenuation and imaging risks and dynamically adjusts radio frequency parameters. Therefore, the problems of unbalanced transceiving efficiency, poor universality and the like in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of magnetic resonance imaging technology, specifically to a low-field magnetic resonance head imaging radio frequency coil system. Background Technology

[0002] The radio frequency coil system for low-field magnetic resonance imaging (MRI) of the head is the core of low-field MRI equipment. Its performance directly determines the quality of head imaging and transmission / reception efficiency, and is crucial for the detection of brain lesions and the assessment of neurological function. During long-term operation, factors such as the low-field environment, electromagnetic interference, and individual anatomical differences can easily lead to imbalances in transmission / reception efficiency and poor versatility, resulting in increased imaging artifacts, decreased lesion detection sensitivity, and even misdiagnosis or missed diagnosis. As medical technology advances towards precision and portability, higher demands are placed on the dynamic sensing, parameter control, and versatility of the coil system, necessitating a high-performance system that integrates physical and simulation models to ensure reliable imaging.

[0003] However, traditional coil systems have significant shortcomings: fixed design leads to poor versatility; unreasonable transmit / receive links cause imbalance in transmit / receive efficiency and large fluctuations in imaging quality; the electromagnetic control method is singular and cannot dynamically sense changes in multiple factors, resulting in insufficient imaging stability; optimization strategies are lagging behind and do not take into account real-time operating status, which can easily lead to cost waste or poor imaging; and there is a lack of efficient monitoring methods, making it difficult for operators to quickly grasp the equipment status. Overall, they face multiple challenges such as poor versatility and imbalance in transmit / receive efficiency. Summary of the Invention

[0004] This application provides a low-field magnetic resonance head imaging radio frequency coil system to solve the problems of poor versatility and imbalance of transmission and reception efficiency in the prior art.

[0005] The first aspect of this application provides a low-field magnetic resonance imaging (MRI) head imaging radio frequency coil system, comprising: a flexible array coil sensing module for acquiring raw MRI radio frequency signals, coil coupling state data, and environmental interference parameters; a dynamic field strength calibration module for constructing a dynamic magnetic field distribution model of the head imaging region, and based on the coil coupling state data, environmental interference parameters, and coil coupling characteristics, using a multi-source magnetic field compensation algorithm to correct the magnetic field uniformity deviation in the low-field environment in real time, thereby obtaining magnetic field calibration data; an intelligent imaging optimization center for identifying low-field-specific imaging artifact types, locating signal distortion sources, and generating an imaging quality optimization scheme by combining the raw MRI radio frequency signals and the magnetic field calibration data with an improved CNN-Transformer hybrid model; a three-dimensional imaging reconstruction unit for constructing a three-dimensional head imaging model, jointly rendering the anatomical structure and functional signal intensity distribution of brain tissue, and outputting preliminary imaging results after artifact pre-labeling and signal fidelity calibration; and an adaptive radio frequency control unit for predicting coil performance degradation trends and imaging risks based on the imaging quality optimization scheme, the preliminary imaging results, and real-time coil monitoring data, and dynamically adjusting the radio frequency transmission power, frequency, and receiver gain parameters.

[0006] Preferably, the flexible array coil sensing module includes a flexible array coil array, a coupling monitoring unit, and an environmental interference sensor group. The flexible array coil array is used to acquire raw radio frequency signals from head magnetic resonance imaging in real time. The coupling monitoring unit is used to acquire coil coupling state data such as inter-coupling coefficient and impedance matching state. The environmental interference sensor group is used to acquire environmental interference parameters such as magnetic field drift, electromagnetic radiation, and temperature.

[0007] Preferably, the dynamic field strength calibration module includes a magnetic field modeling unit and a multi-source compensation unit. The magnetic field modeling unit is used to construct a dynamic magnetic field distribution model of the head imaging area based on head anatomical structure data and real-time magnetic field detection data. The multi-source compensation unit is used to use a multi-source magnetic field compensation algorithm, combined with coil coupling state data, environmental interference parameters and coil coupling characteristics, to correct the magnetic field uniformity deviation in low field environment in real time and generate magnetic field calibration data.

[0008] Preferably, the intelligent imaging optimization center includes an artifact recognition unit and a distortion tracing module. The artifact recognition unit is used to identify low-field-specific imaging artifacts such as chemical shift, magnetic field inhomogeneity, and motion-related artifacts by utilizing an improved CNN-Transformer hybrid model and combining the original RF signal of magnetic resonance imaging with magnetic field calibration data. The distortion tracing module is used to locate the physical location and influence range of the signal distortion source and generate an imaging quality optimization scheme based on the spatial mapping relationship between artifact features and the magnetic field distribution model.

[0009] Preferably, the three-dimensional imaging reconstruction unit includes an imaging modeling engine, a joint rendering unit, and an artifact calibration module. The imaging modeling engine is used to construct a three-dimensional imaging model of the head based on the calibrated radio frequency signals, restoring the anatomical structure and functional signal distribution characteristics of brain tissue. The joint rendering unit is used to render the morphological and functional signal intensity distribution cloud map of brain tissue in real time, fusing and displaying anatomical and functional imaging. The artifact calibration module is used to pre-mark artifact regions in the imaging results and complete the calibration processing of radio frequency signals through a signal fidelity algorithm.

[0010] Preferably, the adaptive radio frequency control unit includes a performance prediction module and a radio frequency parameter adjustment unit. The performance prediction module is used to predict the signal-to-noise ratio attenuation trend of the coil based on the Weibull distribution and coil performance degradation data, and to assess the imaging risk in combination with real-time imaging data. The radio frequency parameter adjustment unit is used to dynamically adjust the radio frequency transmit power, frequency and receive gain parameters according to the imaging quality optimization scheme, preliminary imaging results and real-time coil monitoring data.

[0011] A second aspect of this application provides a method for low-field magnetic resonance imaging (MRI) of the head using a radio frequency (RF) coil, comprising: acquiring raw MRI signals of the head, coil coupling state data, and environmental interference parameters; constructing a dynamic magnetic field distribution model of the head imaging region based on the coil coupling state data, the environmental interference parameters, and coil coupling characteristics; using a multi-source magnetic field compensation algorithm to correct the magnetic field uniformity deviation in the low-field environment in real time, thereby obtaining magnetic field calibration data; based on the raw MRI RF signals and the magnetic field calibration data, combined with an improved CNN-Transformer model, identifying low-field-specific imaging artifact types, locating signal distortion sources, generating an imaging quality optimization scheme; combining the dynamic magnetic field distribution model and the raw MRI RF signals of the head, constructing a three-dimensional imaging model of the head; jointly rendering the anatomical structure and functional signal intensity distribution of brain tissue; and outputting preliminary imaging results after artifact pre-labeling and signal fidelity calibration; and dynamically adjusting the RF transmit power, frequency, and receive gain parameters based on the imaging quality optimization scheme, the preliminary imaging results, and real-time coil monitoring data, combined with Weibull distribution prediction of coil performance degradation trends and imaging risks.

[0012] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a low-field magnetic resonance head imaging radio frequency coil method as described in the above embodiments.

[0013] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement a low-field magnetic resonance head imaging radio frequency coil method as described in the above embodiments.

[0014] A fifth aspect of this application provides a computer program product, including a computer program or instructions for implementing a low-field magnetic resonance head imaging radio frequency coil method as described in the above embodiments.

[0015] Therefore, this application has the following beneficial effects: This application's embodiments comprehensively acquire raw radio frequency signals from head MRI, coil coupling state data, and environmental interference parameters through a flexible array coil sensing module, providing multi-dimensional foundational data for imaging calibration and optimization. A dynamic field strength calibration module, combining a multi-source magnetic field compensation algorithm and a dynamic magnetic field distribution model, accurately corrects magnetic field uniformity deviations in low-field environments, overcoming the limitations of insufficient magnetic field stability in traditional low-field imaging. An intelligent imaging optimization center utilizes an improved CNN-Transformer hybrid model to fuse multi-source data, enhancing the accuracy of identifying low-field-specific imaging artifacts and locating signal distortion sources, thus strengthening the targeted nature of imaging quality optimization. A three-dimensional imaging reconstruction unit achieves fusion rendering of brain tissue anatomical structures and functional signals, completing artifact pre-labeling and signal fidelity calibration, accurately presenting head imaging details. An adaptive radio frequency control unit dynamically adjusts radio frequency-related parameters based on coil performance attenuation prediction and imaging risk assessment, effectively improving the clarity, fidelity, and stability of low-field MRI head imaging, and reducing the impact of imaging artifact interference and coil performance attenuation. This solves the problems of unbalanced transmission and reception efficiency and poor versatility in existing technologies.

[0016] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0017] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of a low-field magnetic resonance head imaging radio frequency coil system provided according to an embodiment of this application; Figure 2 This is a schematic diagram of a flexible array coil sensing module according to an embodiment of this application; Figure 3 This is a schematic diagram of a dynamic field strength calibration module according to an embodiment of this application; Figure 4 This is a schematic diagram of an intelligent imaging optimization center according to an embodiment of this application; Figure 5 This is a schematic diagram of a three-dimensional imaging reconstruction unit provided according to an embodiment of this application; Figure 6 This is a schematic diagram of an adaptive radio frequency control unit according to an embodiment of this application; Figure 7 This is a schematic diagram of a low-field magnetic resonance head imaging radio frequency coil system provided according to an embodiment of this application; Figure 8 This is a flowchart of a low-field magnetic resonance head imaging radio frequency coil method according to an embodiment of this application; Figure 9 This is a schematic diagram of a low-field magnetic resonance head imaging radio frequency coil method according to an embodiment of this application; Figure 10 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The following description, with reference to the accompanying drawings, illustrates a low-field magnetic resonance imaging (MRI) head imaging radio frequency coil system according to an embodiment of this application. Addressing the issue of transmit / receive efficiency imbalance mentioned in the background section, this application provides a low-field MRI head imaging radio frequency coil system. In this system, a flexible array coil sensing module comprehensively acquires raw MRI radio frequency signals, coil coupling state data, and environmental interference parameters, providing multi-dimensional foundational data for imaging calibration and optimization. A dynamic field strength calibration module, combining a multi-source magnetic field compensation algorithm and a dynamic magnetic field distribution model, accurately corrects magnetic field uniformity deviations in low-field environments, overcoming the limitations of insufficient magnetic field stability in traditional low-field imaging. An intelligent imaging optimization center utilizes an improved CNN-Transformer hybrid model to fuse multi-source data, enhancing the accuracy of identifying low-field-specific imaging artifacts and locating signal distortion sources, thereby improving the targeted nature of imaging quality optimization. A three-dimensional imaging reconstruction unit achieves fusion rendering of brain tissue anatomical structures and functional signals, completing artifact pre-labeling and signal fidelity calibration, accurately presenting head imaging details. The adaptive RF control unit dynamically adjusts RF-related parameters based on coil performance degradation prediction and imaging risk assessment, effectively improving the clarity, fidelity, and stability of low-field MRI head imaging, while reducing the impact of imaging artifacts and coil performance degradation. This solves the problems of unbalanced transmission and reception efficiency and poor versatility in existing technologies.

[0020] Figure 1 This is a schematic diagram of a low-field magnetic resonance head imaging radio frequency coil system provided in an embodiment of this application.

[0021] This application provides a low-field magnetic resonance imaging (MRI) head radio frequency coil system, the system 10 comprising: Flexible array coil sensing module 100, dynamic field strength calibration module 200, intelligent imaging optimization center 300, three-dimensional imaging reconstruction unit 400, and adaptive radio frequency control unit 500.

[0022] The system comprises the following components: a flexible array coil sensing module 100 for acquiring raw radio frequency signals from head MRI, coil coupling status data, and environmental interference parameters; a dynamic field strength calibration module 200 for constructing a dynamic magnetic field distribution model of the head imaging region, using coil coupling status data, environmental interference parameters, and coil coupling characteristics, and employing a multi-source magnetic field compensation algorithm to correct magnetic field uniformity deviations in low-field environments in real time, thereby obtaining magnetic field calibration data; an intelligent imaging optimization center 300 for identifying low-field-specific imaging artifact types, locating signal distortion sources, and generating imaging quality optimization schemes by combining raw MRI radio frequency signals and magnetic field calibration data with an improved CNN-Transformer hybrid model; a three-dimensional imaging reconstruction unit 400 for constructing a three-dimensional head imaging model, jointly rendering the anatomical structure and functional signal intensity distribution of brain tissue, and outputting preliminary imaging results after artifact pre-labeling and signal fidelity calibration; and an adaptive radio frequency control unit 500 for predicting coil performance degradation trends and imaging risks based on the imaging quality optimization scheme, preliminary imaging results, and real-time coil monitoring data, and dynamically adjusting radio frequency transmission power, frequency, and receiver gain parameters.

[0023] It is understood that in this embodiment, the flexible array coil sensing module comprehensively collects the original radio frequency signals, coil coupling state data, and environmental interference parameters of head MRI, providing multi-dimensional basic data for imaging calibration and optimization. The dynamic field strength calibration module, combined with a multi-source magnetic field compensation algorithm and a dynamic magnetic field distribution model, accurately corrects the magnetic field uniformity deviation in low-field environments, overcoming the limitations of insufficient magnetic field stability in traditional low-field imaging. The intelligent imaging optimization center uses an improved CNN-Transformer hybrid model to fuse multi-source data, improving the accuracy of low-field-specific imaging artifact identification and signal distortion source localization, and enhancing the targeted nature of imaging quality optimization. The three-dimensional imaging reconstruction unit realizes the fusion rendering of brain tissue anatomical structure and functional signals, completes artifact pre-labeling and signal fidelity calibration, and accurately presents head imaging details. The adaptive radio frequency control unit dynamically adjusts radio frequency related parameters based on coil performance attenuation prediction and imaging risk assessment, effectively improving the clarity, fidelity, and stability of low-field MRI head imaging, and reducing the impact of imaging artifact interference and coil performance attenuation. Thus, it solves the problems of unbalanced transmission and reception efficiency and poor versatility in existing technologies.

[0024] In this embodiment of the application, the flexible array coil sensing module 100 includes: as follows Figure 2As shown, the system includes a flexible array coil array, a coupling monitoring unit, and an environmental interference sensor group.

[0025] Among them, the flexible array coil array is used to acquire the raw radio frequency signal of head magnetic resonance in real time; the coupling monitoring unit is used to acquire coil coupling state data such as the mutual coupling coefficient and impedance matching state between coils; and the environmental interference sensor group is used to acquire environmental interference parameters such as magnetic field drift, electromagnetic radiation and temperature.

[0026] It is understood that the embodiments of this application acquire raw radio frequency signals of head magnetic resonance imaging in real time through a flexible array coil array, accurately obtaining basic data for brain magnetic resonance imaging; the coupling monitoring unit acquires data on the mutual coupling coefficient and impedance matching status between coils in real time, ensuring the stability of coil operation and signal acquisition quality; the environmental interference sensor group acquires parameters of magnetic field drift, electromagnetic radiation and temperature interference, weakening the interference of environmental factors on magnetic resonance signals, providing raw data for magnetic resonance signal analysis, accurate imaging and subsequent intelligent sensing and control, and improving accuracy and stability.

[0027] In this embodiment of the application, the dynamic field strength calibration module 200 includes: as follows Figure 3 As shown, the magnetic field modeling unit and the multi-source compensation unit are shown.

[0028] The magnetic field modeling unit is used to construct a dynamic magnetic field distribution model of the head imaging area based on head anatomical structure data and real-time magnetic field detection data; the multi-source compensation unit is used to use a multi-source magnetic field compensation algorithm, combined with coil coupling state data, environmental interference parameters and coil coupling characteristics, to correct the magnetic field uniformity deviation in low field environment in real time and generate magnetic field calibration data.

[0029] It is understood that the magnetic field modeling unit in this application integrates head anatomical structure data and real-time magnetic field detection data to construct a dynamic magnetic field distribution model of the head imaging area, accurately restoring the true distribution characteristics of the magnetic field under low field conditions; the multi-source compensation unit uses a multi-source magnetic field compensation algorithm, combined with coil coupling state, environmental interference parameters and coil coupling characteristics, to correct the magnetic field uniformity deviation in real time and generate calibration data, providing a reliable benchmark and accurate compensation basis for magnetic resonance signal correction and imaging quality optimization, thereby improving the uniformity and clarity of head imaging.

[0030] It should be noted that a dynamic magnetic field distribution model for the head imaging region is constructed based on head anatomical data and real-time magnetic field detection data. First, based on the head anatomical data, the morphology, layered structure, spatial location, and geometric distribution characteristics of brain tissue are extracted to clarify the key boundaries and target range of the imaging region, forming the basic spatial framework for magnetic field modeling. Simultaneously, real-time magnetic field detection data is used to collect information on the magnitude, distribution gradient, and instantaneous changes of the field strength at different locations on the head, obtaining measured parameters that reflect the current true state of the magnetic field. Then, the spatial features of the anatomical structure are matched, fitted, and modeled with the real-time magnetic field detection results. Combining the influence of the head's physiological structure on the magnetic field distribution, the field strength distribution characteristics are dynamically updated. Finally, a dynamic magnetic field model that can accurately reflect the real-time changes, spatial distribution, and structural correlation of the magnetic field in the head imaging region is constructed, providing a true and reliable field strength benchmark and data support for subsequent magnetic field uniformity compensation, signal correction, and imaging quality optimization.

[0031] Dynamic magnetic field model formula: in, Spatial location The static magnetic field vector at that location; This represents the amplitude of the static magnetic field strength. It is the unit vector in the z-direction; It is a spatial position vector; Spatial location The magnetic field shift caused by tissue magnetization; Spatial location The magnetic susceptibility of the tissue; Spatial location Magnetic field distortion function at the location; For time Spatial location The magnetic field shift caused by environmental factors; This is the temperature-to-magnetic field conversion coefficient; For time The change in temperature over time; The environmental noise-magnetic field conversion coefficient; For time The ambient noise intensity at that time; The motion-to-magnetic field conversion coefficient; For time Motion state parameters at time; It is a time variable; For time Spatial location The total magnetic field vector at that location.

[0032] Multi-source magnetic field compensation algorithm formula: in, For time Spatial location The total magnetic field deviation at the location; It is a time variable; It is a spatial position vector; For time Spatial location The measured magnetic field strength at the location; Spatial location The target magnetic field strength at the location; This represents the total magnetic field deviation. This refers to the magnetic field deviation caused by the coil coupling effect. The magnetic field shift is caused by environmental factors; The magnetic field shift caused by tissue magnetization; For time Spatial location Magnetic field compensation at the location; This is the coil coupling compensation coefficient; The parameters represent the coil coupling strength. This is the environmental disturbance compensation coefficient; Comprehensive parameters for environmental interference; This is the noise compensation coefficient; For environmental noise parameters; This is the tissue magnetic susceptibility compensation coefficient; Spatial location The magnetic susceptibility of the tissue; For time Spatial location The final calibration magnetic field strength at the location; For time Spatial location The measured magnetic field strength at the location.

[0033] For example, taking low-field head MRI as an example, the multi-source compensation unit first uses a multi-source magnetic field compensation algorithm to accurately decouple the magnetic field distortion components caused by coil mutual coupling and impedance mismatch, based on a dynamic magnetic field distribution model, coil coupling characteristics, and prior information about the head's anatomical structure. Simultaneously, it identifies the influence of magnetic field drift, electromagnetic radiation, and temperature changes on the uniformity of the main magnetic field. By combining coil coupling state data and environmental interference parameters, it performs weighted fitting and deviation quantification on different interference sources, clearly locating abnormal positions such as field strength shift and gradient distortion within the imaging area. Subsequently, the fitted multi-source interference components are iteratively calibrated with the target uniform magnetic field. Real-time compensation calculations correct the magnetic field uniformity deviation in the low-field environment, dynamically restoring a uniform magnetic field distribution that meets imaging requirements. The final output is magnetic field calibration data that can be directly used for field intensity control and signal correction. This data provides quantitative support for optimizing the working state of the flexible array coil, suppressing artifacts and noise interference, and improving the clarity and signal-to-noise ratio of head MRI.

[0034] In this embodiment of the application, the intelligent imaging optimization center 300 includes: as follows Figure 4 As shown, the artifact recognition unit and the distortion tracing module are shown.

[0035] The artifact recognition unit uses an improved CNN-Transformer hybrid model, combined with the original RF signal and magnetic field calibration data from magnetic resonance imaging, to identify low-field-specific imaging artifacts such as chemical shift, magnetic field inhomogeneity, and motion. The distortion tracing module is used to locate the physical location and influence range of signal distortion sources and generate imaging quality optimization schemes based on the spatial mapping relationship between artifact features and magnetic field distribution models.

[0036] It is understood that the artifact recognition unit in this application adopts an improved CNN-Transformer hybrid model, which combines the original radio frequency signal of magnetic resonance imaging with magnetic field calibration data to accurately identify the unique imaging artifact types of chemical shift, magnetic field inhomogeneity, and motion low field, effectively improving the comprehensiveness and accuracy of artifact detection. The distortion tracing module, based on the spatial mapping relationship between artifact features and magnetic field distribution model, accurately locates the physical location and influence range of signal distortion sources and generates imaging quality optimization schemes, providing targeted basis for artifact suppression, signal correction, and imaging parameter adjustment, thereby improving the clarity, fidelity, and imaging quality stability of low field head magnetic resonance imaging.

[0037] It should be noted that the improved CNN-Transformer hybrid model formula is as follows: in, The output features of the convolutional layer; To improve convolution operations; For time Spatial location Magnetic resonance imaging data at the location; It is a time variable; It is a spatial position vector; These are the trainable parameters for the convolutional layer; The flattened feature vector; For feature flattening operation; This is the feature vector after adding positional encoding; For position encoding operations; Output features for the Transformer layer; This is a multi-head self-attention mechanism; This is the location encoding feature of the previous layer; It is a feedforward neural network; This is the result of multi-head self-attention calculation; For query vector; The key vector; It is a value vector; Use the Softmax activation function; The dimension of the key vector; This is a matrix transpose operation; This is the feature vector after adding positional encoding; To query the vector projection matrix; The key vector projection matrix; The projection matrix of the value vector; The result of feature fusion; For feature splicing operations; Characteristics of magnetic field calibration data; Predict the probability of artifact type; It is a fully connected layer.

[0038] Based on the spatial mapping relationship between artifact features and the magnetic field distribution model, the physical location and influence range of the signal distortion source are located, and an imaging quality optimization scheme is generated. First, based on the chemical shift, magnetic field inhomogeneity, and motion artifact features output by the artifact recognition unit, key feature information such as the position, shape, intensity, and distribution pattern of artifacts is extracted to form an imaging distortion feature vector. Simultaneously, a dynamic magnetic field distribution model is invoked to obtain three-dimensional field strength data such as the magnitude of the field strength, spatial gradient, real-time drift, and uniformity deviation within the head imaging region, constructing a magnetic field distortion baseline map for the corresponding spatial location. Then, the coordinate information of the artifact features in the imaging domain is compared with the magnetic field distribution model. The model is registered, mapped, and correlated in the spatial coordinates of the physical field to establish the correspondence between artifact performance and magnetic field anomalies, coil coupling, and environmental interference. Through spatial matching and source tracing calculations, the physical source causing signal distortion and its influence range within the imaging area are accurately located. Finally, combined with magnetic field calibration data, coil operating status, and environmental interference parameters, an imaging quality optimization scheme is generated, including field strength compensation, signal correction coefficients, and imaging parameter adjustment strategies. This provides a targeted solution for artifact suppression, imaging correction, and system adaptive control, thereby improving the fidelity and reliability of low-field head magnetic resonance imaging from the source.

[0039] In this embodiment of the application, the three-dimensional imaging reconstruction unit 400 includes, as follows: Figure 5 As shown, the imaging modeling engine, joint rendering unit, and artifact calibration module are included.

[0040] The imaging modeling engine is used to construct a three-dimensional imaging model of the head based on the calibrated radio frequency signals, restoring the anatomical structure and functional signal distribution characteristics of brain tissue; the joint rendering unit is used to render the anatomical structure morphology and functional signal intensity distribution cloud map of brain tissue in real time, and to integrate and display the anatomical and functional imaging; the artifact calibration module is used to pre-label the artifact regions in the imaging results and complete the calibration processing of radio frequency signals through a signal fidelity algorithm.

[0041] It is understood that the imaging modeling engine in this application constructs a three-dimensional imaging model of the head based on the calibrated radio frequency signals, accurately restoring the anatomical structure and functional signal distribution characteristics of brain tissue, providing a complete data foundation for brain visualization analysis; the joint rendering unit renders the morphological and functional signal intensity distribution cloud map of brain tissue anatomical structure in real time, efficiently integrating and displaying anatomical imaging and functional imaging, improving the intuitiveness and information content of three-dimensional imaging; the artifact calibration module pre-marks the artifact regions in the imaging results, completes the radio frequency signal calibration processing through the signal fidelity algorithm, effectively suppresses residual artifacts and ensures signal authenticity, enhancing the reliability and intuitiveness of clinical diagnosis.

[0042] It should be noted that the signal fidelity algorithm formula is as follows: in, pixel coordinates The calibrated image signal at the location; These are the row coordinates of the image; For image column coordinates; pixel coordinates The original distorted image signal at the location; A set of pre-labeled artifact regions; For artifact pixels The corresponding set of normal tissue regions in the neighborhood; The row coordinates of the neighboring pixels; The column coordinates of the neighboring pixels; For neighboring pixels The corresponding spatial weighting coefficients.

[0043] A three-dimensional head imaging model is constructed based on calibrated radio frequency signals to reconstruct the anatomical structure and functional signal distribution characteristics of brain tissue. First, pure radio frequency signals, acquired by a flexible array coil and calibrated by field strength and coupling correction, are used as the data source. Signal amplitude, phase, and spectral characteristics at different layers and locations are extracted and arranged in an orderly manner according to head spatial coordinates for three-dimensional mesh reconstruction. Based on the correspondence between signal intensity and brain tissue relaxation characteristics, signal differences between different tissues such as gray matter, white matter, and cerebrospinal fluid are distinguished, restoring the true geometric morphology, layered structure, spatial location, and boundary contours of brain tissue. Simultaneously, combining the signal response patterns corresponding to functional regions, the strength and distribution characteristics of functional signals in each brain region are analyzed and mapped, fusing anatomical structural information and functional signal information into a model. Finally, a three-dimensional head imaging model that completely and accurately reflects the morphological structure and functional distribution of brain tissue is formed, providing a high-fidelity, high-precision three-dimensional data carrier for subsequent joint rendering, distortion recognition, and imaging optimization.

[0044] Radio frequency (RF) signal calibration is performed based on artifact features in the imaging results. First, the reconstructed 3D head imaging data is traversed globally and inspected pixel by pixel. Combining the artifact type, location, and distribution information output by the artifact recognition unit, residual artifact regions such as chemical shift, magnetic field inhomogeneity, and motion are accurately pre-marked to determine the signal anomaly range to be corrected. Then, using the real RF signals of surrounding normal tissue as a reference, a signal fidelity algorithm is used to fit and correct the distorted signal amplitude, phase, and spectral characteristics within the marked area. While preserving the original anatomical structure and functional signals of the brain tissue, signal distortion and noise interference caused by artifacts are suppressed. Through iterative calibration and fidelity constraints, the signals in the abnormal areas are restored to a reasonable range that conforms to real physiological characteristics. Finally, the fine calibration of the RF signal is completed, providing a reliable data foundation with no artifacts and high fidelity for subsequent joint rendering and 3D imaging output, improving the accuracy and readability of head MRI results.

[0045] In this embodiment of the application, the adaptive radio frequency control unit 500 includes, as follows: Figure 6 As shown, the performance prediction module and the radio frequency parameter adjustment unit are included.

[0046] The performance prediction module is used to predict the signal-to-noise ratio attenuation trend of the coil based on the Weibull distribution and coil performance degradation data, and to assess the imaging risk in combination with real-time imaging data; the radio frequency parameter adjustment unit is used to dynamically adjust the radio frequency transmit power, frequency and receive gain parameters according to the imaging quality optimization scheme, preliminary imaging results and real-time coil monitoring data.

[0047] It is understood that the performance prediction module in this application is based on Weibull distribution and coil performance degradation data to predict the coil signal-to-noise ratio attenuation trend, and comprehensively assesses imaging quality risks in combination with real-time imaging data, providing a forward-looking basis for coil status prediction and parameter adjustment; the radio frequency parameter adjustment unit dynamically adjusts the radio frequency transmit power, frequency and receive gain parameters according to the imaging quality optimization scheme, preliminary imaging results and real-time coil monitoring data, to improve the working stability and signal consistency of the flexible array coil, and suppress imaging drift and noise rise.

[0048] It should be noted that the Weibull distribution formula is: in, The cumulative distribution function is the Weibull distribution. It is a time variable; For position parameters; For scale parameters; For shape parameters; It is an exponential function; Let Weibull be the probability density function. Let be the reliability function of the Weibull distribution.

[0049] Based on the imaging quality optimization scheme, preliminary imaging results, and real-time coil monitoring data, the RF transmit power, frequency, and receive gain parameters are dynamically adjusted. Real-time comparison and analysis are performed based on the imaging quality optimization scheme, preliminary imaging results, and real-time coil monitoring data to accurately identify deviations between the actual signal state and ideal imaging conditions. Based on the deviation analysis results, a parameter adjustment mechanism is automatically activated to specifically adjust the RF transmit power, stabilize signal excitation efficiency, and avoid interference caused by weak signals or abnormal power. The RF frequency is adjusted to precisely calibrate the resonance state, ensuring high synchronization between signal transmission and reception and the magnetic field environment. The receive gain is adjusted to enhance the effective signal and suppress environmental interference and coil noise, improving the adaptability to low-field environments, coil coupling changes, and imaging state fluctuations.

[0050] This application proposes a low-field magnetic resonance imaging (MRI) head imaging radio frequency coil system. A flexible array coil sensing module comprehensively acquires raw MRI radio frequency signals, coil coupling state data, and environmental interference parameters, providing multi-dimensional foundational data for imaging calibration and optimization. A dynamic field strength calibration module, combining a multi-source magnetic field compensation algorithm and a dynamic magnetic field distribution model, accurately corrects magnetic field uniformity deviations in low-field environments, overcoming the limitations of insufficient magnetic field stability in traditional low-field imaging. An intelligent imaging optimization center utilizes an improved CNN-Transformer hybrid model to fuse multi-source data, enhancing the accuracy of identifying low-field-specific imaging artifacts and locating signal distortion sources, thus improving the targeted nature of imaging quality optimization. A three-dimensional imaging reconstruction unit achieves fusion rendering of brain tissue anatomical structures and functional signals, completing artifact pre-labeling and signal fidelity calibration to accurately present head imaging details. An adaptive radio frequency control unit dynamically adjusts radio frequency-related parameters based on coil performance attenuation prediction and imaging risk assessment, effectively improving the clarity, fidelity, and stability of low-field MRI head imaging and reducing the impact of imaging artifact interference and coil performance attenuation. This solves the problems of unbalanced transmission and reception efficiency and poor versatility in existing technologies.

[0051] The following will illustrate a low-field magnetic resonance head imaging radio frequency coil system through a specific embodiment, such as... Figure 7 As shown, it includes: A clinical medical imaging center is equipped with multiple low-field magnetic resonance imaging (MRI) devices, primarily used for head disease screening and routine neurological diagnoses in primary healthcare institutions. In practical use, traditional fixed radiofrequency coils suffer from poor fit and uneven signal acquisition. They are also susceptible to environmental magnetic field drift, coil mutual coupling, and human movement, leading to image artifacts, distortion, and low signal-to-noise ratios, making it difficult to meet the requirements for continuous, stable, and high-quality head imaging. Therefore, the imaging center adopts the low-field MRI head imaging radiofrequency coil system provided in this embodiment, ensuring stable operation throughout the entire head MRI process and continuously improving image quality.

[0052] The flexible array coil sensing module, serving as the front-end data source for the entire system, comprises a flexible array coil array, a coupling monitoring unit, and an environmental interference sensor group, forming a multi-dimensional signal acquisition network covering the entire head. The flexible array coil array, made of an ultra-thin flexible substrate, closely conforms to the contours of the human head and is evenly distributed across the forehead, parietal lobe, temporal lobe, and occipital lobe regions. During scanning, it continuously acquires raw radio frequency signals from head magnetic resonance imaging, ensuring complete signal coverage and uniform spatial distribution. The coupling monitoring unit monitors the coil array's operating status in real time, obtaining the mutual coupling coefficient and impedance matching status between coils through impedance detection and signal coupling analysis, providing a basis for subsequent magnetic field compensation and parameter adjustment. The environmental interference sensor group is deployed inside the scanning room and around the coil module, continuously acquiring environmental magnetic field drift, spatial electromagnetic radiation, and environmental temperature changes, comprehensively recording external interference factors that may affect image quality. All acquired data is uploaded to the system processing unit in real time via a high-speed transmission interface. After filtering, noise reduction, and standardization, a unified data stream is formed, providing reliable raw information for dynamic field strength calibration, intelligent imaging optimization, and 3D reconstruction, ensuring the integrity and stability of the system's input data.

[0053] The dynamic field strength calibration module continuously performs magnetic field modeling and multi-source compensation operations during system operation, which is completed collaboratively by the magnetic field modeling unit and the multi-source compensation unit. Based on prior data of the head's anatomical structure and real-time magnetic field detection results, the magnetic field modeling unit establishes a three-dimensional dynamic magnetic field distribution model of the head imaging area, accurately reflecting the changes in magnetic field strength and gradient at different locations and times, and clearly presenting the non-uniform magnetic field areas and distortion locations under low-field conditions. The multi-source compensation unit receives coil coupling state data, environmental interference parameters, and the coil's own coupling characteristics, and uses a multi-source magnetic field compensation algorithm to decouple various interferences, eliminating magnetic field deviations caused by coil mutual coupling, impedance mismatch, environmental temperature drift, and spatial electromagnetic radiation. Through real-time iterative correction, the system can quickly compensate for the degradation of magnetic field uniformity under low-field conditions, generating high-precision magnetic field calibration data and transmitting it back to the imaging link, ensuring a stable and uniform magnetic field within the scanning area. This provides an ideal field strength environment for subsequent RF signal reception and image reconstruction, physically reducing the probability of image distortion and artifacts.

[0054] The intelligent imaging optimization center undertakes the core tasks of artifact recognition and distortion source tracing. It consists of an artifact recognition unit and a distortion source tracing module, enabling automatic judgment and source localization of imaging quality problems. The artifact recognition unit receives calibrated raw RF signals and magnetic field calibration data from the magnetic resonance imaging system. It employs an improved CNN-Transformer hybrid model for feature extraction and classification, fully utilizing local spatial feature extraction and long-range dependency modeling capabilities to accurately identify common low-field imaging types such as chemical shift artifacts, magnetic field inhomogeneity artifacts, and motion artifacts, marking their location, size, and severity. The distortion source tracing module spatially maps the identified artifact features to a dynamic magnetic field distribution model. Through coordinate matching and correlation analysis, it reverse-engineers the true physical source of signal distortion, determining whether the distortion is caused by magnetic field inhomogeneity, abnormal coil coupling, or environmental interference, and assessing its specific impact range within the head region. After locating the distortion source, the system automatically generates targeted imaging quality optimization schemes, including field strength compensation strategies, signal correction coefficients, and coil parameter adjustment suggestions, providing clear guidance for 3D imaging reconstruction and adaptive adjustment of RF parameters.

[0055] The 3D imaging reconstruction unit is responsible for converting the calibrated signal into a clear and complete 3D image of the head. This is achieved through the collaborative work of the imaging modeling engine, the joint rendering unit, and the artifact calibration module. The imaging modeling engine uses radio frequency signals that have undergone field strength calibration and interference suppression as a basis, and performs 3D mesh reconstruction according to the spatial structure of the head. It distinguishes between gray matter, white matter, and cerebrospinal fluid based on signal differences in different brain tissues, fully restoring the anatomical structure and functional signal distribution characteristics of the brain, and constructing a high-fidelity 3D head imaging model. The joint rendering unit simultaneously renders anatomical and functional information, displaying brain tissue layers, boundaries, and spatial locations in grayscale, and presenting functional signal intensity changes in the form of distribution cloud maps. This integrates anatomical and functional imaging in the same 3D space, supporting multi-angle observation and arbitrary cross-sectional browsing, improving image intuitiveness and diagnostic information. The artifact calibration module performs global detection on the preliminary reconstructed image, pre-marks abnormal areas based on artifact information provided by the intelligent imaging optimization center, and corrects distorted radio frequency signals using a signal fidelity algorithm. While preserving real tissue information, it eliminates residual artifacts, ensuring clear image details and accurate tissue boundaries, ultimately outputting 3D imaging results that meet clinical diagnostic requirements.

[0056] The adaptive radio frequency (RF) control unit achieves closed-loop optimization and long-term stable operation of the system, comprising a performance prediction module and an RF parameter adjustment unit. The performance prediction module, based on the Weibull distribution model and historical coil performance degradation data, continuously analyzes the coil's signal-to-noise ratio (SNR) variation, predicts signal attenuation trends during long-term use, and assesses imaging risks potentially caused by coil aging, environmental changes, or parameter deviations, issuing early warnings. The RF parameter adjustment unit dynamically adjusts the system's RF operating parameters, including RF transmit power, operating frequency, and signal receiving gain, based on the imaging quality optimization scheme, preliminary imaging results output by the 3D imaging reconstruction unit, and real-time coil monitoring data. It appropriately optimizes the transmit power and frequency matching relationship when magnetic field homogeneity is insufficient, reasonably increases the receiving gain and suppresses noise when the signal is weak, and adjusts the signal acquisition strategy in advance when the risk of artifacts is high, ensuring the RF link is always in its optimal operating range. The entire adjustment process is real-time responsive, continuous, and smooth, requiring no manual intervention. It effectively adapts to low-field environmental fluctuations, coil state changes, and differences among subjects, continuously ensuring stable and reliable imaging quality, extending system lifespan, and improving the efficiency and consistency of head MRI scans.

[0057] In summary, the embodiments of this application acquire multi-dimensional signals through a flexible array coil sensing module, providing stable and reliable raw data. The dynamic field strength calibration module constructs a precise magnetic field model and completes multi-source compensation, laying a physical foundation for high-quality imaging. The intelligent imaging optimization center accurately identifies artifacts and traces the source of distortion based on the improved model, effectively suppressing image distortion and artifact interference, and improving imaging reliability. The three-dimensional imaging reconstruction unit realizes the fusion and reconstruction of anatomical and functional information, and, together with artifact calibration, ensures clear and faithful images, providing intuitive and accurate image support for clinical diagnosis. The adaptive radio frequency control unit dynamically adjusts radio frequency parameters based on performance prediction, realizing closed-loop optimization and long-term stable operation of the system, adapting to low-field environments and individual differences, improving imaging consistency and equipment lifespan, and solving problems such as poor coil fit, uneven signal, numerous artifacts, and low signal-to-noise ratio in traditional systems.

[0058] Next, referring to the accompanying drawings, a low-field magnetic resonance head imaging radio frequency coil method is described according to an embodiment of this application.

[0059] like Figure 8 As shown, this low-field magnetic resonance imaging head radio frequency coil method includes the following steps: In step S101, the original radio frequency signal of the head magnetic resonance imaging, coil coupling status data, and environmental interference parameters are acquired.

[0060] It is understood that by acquiring the original radio frequency signal of the head magnetic resonance imaging, coil coupling status data and environmental interference parameters, the embodiments of this application can perceive the head signal distribution, coil working status and external environment changes in real time, accurately capture the impact of signal anomalies, coil mutual coupling interference and environmental factors on imaging quality, provide high-quality data basis for subsequent dynamic magnetic field modeling, multi-source field strength compensation, intelligent artifact recognition and imaging optimization, and improve imaging stability and image signal-to-noise ratio.

[0061] In step S102, a dynamic magnetic field distribution model of the head imaging area is constructed based on coil coupling state data, environmental interference parameters and coil coupling characteristics. A multi-source magnetic field compensation algorithm is used to correct the magnetic field uniformity deviation in low field environment in real time to obtain magnetic field calibration data.

[0062] Among them, the dynamic magnetic field distribution model is a three-dimensional model built based on head anatomical structure data and real-time magnetic field detection data. It is used to accurately reflect the distribution law of magnetic field intensity and gradient changes with time and space in the head imaging area, as well as the non-uniform magnetic field areas and distortion locations.

[0063] It is understood that the embodiments of this application, by employing a dynamic magnetic field distribution model and combining coil coupling state data, environmental interference parameters, and coil coupling characteristics, accurately present the spatiotemporal variation patterns of magnetic field strength and gradient in the head imaging region, as well as the magnetic field non-uniformity areas and distortion locations. This provides a precise magnetic field distribution basis for multi-source magnetic field compensation algorithms, efficiently decouples various interferences, corrects magnetic field uniformity deviations in low-field environments in real time, generates high-precision magnetic field calibration data, suppresses imaging artifacts caused by magnetic field distortion from the source, improves magnetic field stability and uniformity, lays a solid foundation for subsequent radio frequency signal calibration, artifact recognition, and high-quality three-dimensional imaging, and ensures the clarity and reliability of low-field head magnetic resonance imaging.

[0064] In step S103, based on the original radio frequency signal and magnetic field calibration data of magnetic resonance imaging, combined with the improved CNN-Transformer model, the types of imaging artifacts specific to low field are identified, the source of signal distortion is located, and an imaging quality optimization scheme is generated. Combined with the dynamic magnetic field distribution model and the original radio frequency signal of head magnetic resonance imaging, a three-dimensional imaging model of the head is constructed. The anatomical structure and functional signal intensity distribution of brain tissue are jointly rendered, and the preliminary imaging results of artifact pre-labeling and signal fidelity calibration are output.

[0065] Among them, the improved CNN-Transformer model is a hybrid network that integrates the local feature extraction of convolutional neural networks with the global dependency modeling capabilities of Transformer, and is used to accurately identify various artifact types in low-field magnetic resonance imaging from magnetic resonance radio frequency signals and magnetic field calibration data.

[0066] It is understood that the embodiments of this application utilize an improved CNN-Transformer model to accurately identify low-field-specific imaging artifact types such as chemical shift, magnetic field inhomogeneity, and motion artifacts, providing a reliable basis for locating signal distortion sources, generating targeted imaging quality optimization schemes, effectively reducing the probability of artifact omission and misjudgment, improving the accuracy and intelligence level of imaging quality judgment, providing a clear and reliable processing foundation for subsequent 3D imaging reconstruction and artifact calibration, and improving the clarity and fidelity of low-field head magnetic resonance imaging.

[0067] For example, in the 3D monitoring of a servo press, when using finite element analysis on the core transmission system (including bearings and lead screw assemblies), it is first discretized into 15,000 high-precision tetrahedral elements. Based on material parameters (elastic modulus of bearing steel 210 GPa, thermal conductivity of lead screw 45 W / (m·K)), a multiphysics coupling model is constructed. During the simulation, the force field (contact stress between the slider and the die, peak value up to 350 MPa), thermal field (temperature conduction generated by motor operation, maximum temperature of 58℃ in the bearing area), and deformation field (axial displacement of 0.025 mm due to load on the lead screw) during stamping are simulated simultaneously. The stress cloud map and temperature gradient curve obtained from the simulation are compared with the data from the physical layer sensors: the vibration amplitude (0.08 g) captured by the vibration sensor in the 1.2 kHz frequency band corresponds to the stress fluctuation of the bearing raceway in the simulation; the temperature difference between the actual measured temperature of 56℃ by the infrared thermal imager and the simulated temperature is ≤3%; and the deviation between the workpiece size deviation (0.023 mm) from the industrial CT scan and the simulation result of the lead screw deformation is ≤0.002 mm. By calibrating model parameters in real time using a digital twin, early wear of bearings caused by stress concentration (local stress on the raceway exceeding 380MPa) and the risk of minor deviation of the lead screw can be accurately identified. This provides a quantitative correction basis for the adaptive control unit (such as fine-tuning the slider pressure by ±2%), effectively avoiding workpiece deviation caused by abnormal transmission system, and realizing three-dimensional dynamic monitoring and accurate early warning of equipment status.

[0068] In step S104, based on the imaging quality optimization scheme, preliminary imaging results, and real-time coil monitoring data, and combined with the Weibull distribution to predict the coil performance degradation trend and imaging risk, the RF transmit power, frequency, and receive gain parameters are dynamically adjusted.

[0069] Among them, the Weibull distribution is a probability distribution used to describe the performance degradation law of equipment and components. It is used to combine coil performance degradation data to predict the coil signal-to-noise ratio degradation trend and assess imaging risk.

[0070] It is understood that the embodiments of this application use the Weibull distribution to accurately describe the coil performance degradation and signal-to-noise ratio change law, combine historical coil degradation data with real-time monitoring status to reliably predict the coil performance degradation trend, identify potential imaging risks in advance, provide a scientific basis for dynamic adjustment of radio frequency parameters, complete adaptive adjustment before the imaging quality deteriorates, avoid image distortion and signal-to-noise ratio reduction caused by coil performance degradation, improve operational stability and imaging consistency, and extend the effective service life of the coil.

[0071] For example, in low-field MRI head imaging monitoring, the adaptive radio frequency control unit (RFC) collects data such as coil signal-to-noise ratio (SNR), coupling state, signal attenuation amplitude, and imaging quality indicators in real time. It then establishes a performance degradation model synchronized with the actual coil state using a Weibull distribution. This model can dynamically fit key parameters such as coil degradation rate and lifetime threshold, predict SNR trends under different usage durations, and identify early degradation by comparing historical degradation data with real-time monitoring. For instance, it can anticipate the risk of SNR decline and simultaneously push adjustment data to the RF parameter adjustment unit, optimizing transmit power, operating frequency, and receive gain in real time. This ensures the imaging SNR remains stable within diagnostic requirements while avoiding image failure and rescanning due to sudden changes in coil performance, thus improving imaging stability.

[0072] According to the embodiments of this application, a low-field magnetic resonance imaging (MRI) head imaging radio frequency coil method is proposed. A flexible array coil sensing module comprehensively collects the original MRI radio frequency signal, coil coupling state data, and environmental interference parameters, providing multi-dimensional basic data for imaging calibration and optimization. A dynamic field strength calibration module, combining a multi-source magnetic field compensation algorithm and a dynamic magnetic field distribution model, accurately corrects magnetic field uniformity deviations in low-field environments, overcoming the limitations of insufficient magnetic field stability in traditional low-field imaging. An intelligent imaging optimization center utilizes an improved CNN-Transformer hybrid model to fuse multi-source data, improving the accuracy of identifying low-field-specific imaging artifacts and locating signal distortion sources, enhancing the targeted nature of imaging quality optimization. A three-dimensional imaging reconstruction unit achieves fusion rendering of brain tissue anatomical structures and functional signals, completing artifact pre-labeling and signal fidelity calibration, accurately presenting head imaging details. An adaptive radio frequency control unit dynamically adjusts radio frequency-related parameters based on coil performance attenuation prediction and imaging risk assessment, effectively improving the clarity, fidelity, and stability of low-field MRI head imaging, reducing the impact of imaging artifact interference and coil performance attenuation. This solves the problems of unbalanced transmission and reception efficiency and poor versatility in existing technologies.

[0073] The following will illustrate a low-field magnetic resonance imaging (MRI) head radio frequency coil method through a specific embodiment, such as... Figure 9 As shown, it includes: A 0.35T low-field magnetic resonance scanner of a certain model was selected as the control object. This device is equipped with a head-specific 8-channel RF coil. First, multiple types of sensors were deployed to achieve full data acquisition. At the RF signal acquisition level, a directional coupler and a high-precision power sensor were installed at the output of the RF power amplifier, with a range covering 0 to 5kW and a measurement accuracy of ±0.5W, used for real-time monitoring of forward and reflected power changes. The eight-channel receiving coil array integrates a low-noise preamplifier and a PIN diode switch. Each coil unit has a built-in S-parameter monitoring circuit with a measurement accuracy of ±0.1dB, acquiring the impedance matching status between the coil and the human body load at a 1kHz sampling rate. The data acquisition card uses a 24-bit analog-to-digital converter with a sampling frequency of 10MHz, directly capturing the analog RF echo signal generated by hydrogen proton precession. Environmental interference monitoring needs to cover the interior of the magnet compartment and the RF shielding room. A triaxial magnetic field probe was arranged around the gradient coil chamber, with a range of ±10 millitalas, a measurement accuracy of ±1 nanotesla, and a sampling frequency of 500Hz, used to monitor the eddy current field introduced by gradient switching. A fiber optic electric field sensor is installed on a hexahedron in the shielded room to capture 50Hz power frequency leakage and radio frequency pulse noise. Distributed temperature sensors are deployed on the coil array support, with a measurement accuracy of ±0.5 degrees Celsius, to monitor the temperature rise of the coil conductors in contact with the human body in real time. All data is aggregated to the receiving unit of the spectrometer host via shielded twisted-pair cables, and after digital down-conversion processing, is transmitted to the image reconstruction workstation via gigabit Ethernet, with transmission latency controlled within 10ms. The raw K-space data and coil status logs are stored in the InfluxDB time-series database and retained for 7 days. A MySQL database stores patient information and scanning protocol parameters, constructing a complete traceability system for the scanning process.

[0074] During the data cleaning phase, a differentiated strategy was adopted based on the characteristics of the magnetic resonance signal. Gaussian white noise in the radio frequency echo signal was reduced by 4-level decomposition using a db6 wavelet basis, and high-frequency coefficient soft thresholding was applied, improving the signal-to-noise ratio to over 40dB. The 3-sigma criterion was used to identify abrupt changes in data caused by motion artifacts. K-space outliers exceeding the mean plus or minus 3 standard deviations were corrected by weighted median filtering of the nearest 12 points. When the occasional missing value of coil sensitivity data was less than 3%, K-space conjugate symmetric filling was used to complete the data; continuous missing values ​​were marked as invalid and triggered a coil fault warning. Feature extraction needed to consider both frequency domain characteristics and anatomical structural features. Eight frequency domain statistics, such as the energy proportion of the central region and the high-frequency attenuation coefficient, were extracted from the K-space data to reflect the signal attenuation characteristics. Six spatial features, such as the uniformity of signal-to-noise ratio distribution and the location of the maximum signal intensity, were extracted from the coil sensitivity spectrum. Combined with the TE / TR parameters of the spin echo sequence, the T1 / T2 weighted contrast features were calculated. Standardization eliminates dimensional differences between different coil channels. The Z-score formula maps all features to the range of -1 to 1, constructing a structured dataset containing timestamps, 24-dimensional feature vectors, and artifact labels. The dataset is divided in an 8:2 ratio into a training set of 100,000 labeled scan data and a test set of 25,000 data. It is stored in HDF5 format with a compression rate of 40%. A real-time data pipeline is built using ZeroMQ to provide standardized input for the improved CNN-Transformer model.

[0075] The model input layer receives a 24-dimensional normalized feature vector and a 256x256 K-space amplitude image matrix. A dual-stream network structure processes numerical and image features separately. The CNN branch uses a ResNet34 backbone network and extracts local artifact texture features in K-space through four residual blocks, such as phase stripes caused by motion or signal holes caused by metal implants. The Transformer branch uses a ViT-Base architecture, flattening the K-space data into a 16x16 image patch sequence. A 12-layer multi-head self-attention mechanism captures global spatial correlation and analyzes the propagation patterns of artifacts in the frequency and phase encoding directions. The feature fusion module injects the local features extracted by the CNN into the Transformer's query vector through a cross-attention mechanism, outputting an artifact probability map and distortion source coordinates. Combined with a dynamic magnetic field distribution model, a back-projection algorithm is used to reconstruct a three-dimensional voxel model of the head, rendering the brain tissue anatomy in grayscale and using a heatmap overlay function to activate the signal intensity of the region. During this process, K-space data repair is automatically performed on high-confidence artifact regions, and pre-labeled metadata containing artifact location, type and severity is generated. Finally, preliminary imaging slices calibrated for signal fidelity are output with slice thickness controlled within 3mm for physicians to view in real time.

[0076] The structural similarity index, contrast-to-noise ratio, and signal-to-noise ratio of the preliminary imaging results are continuously monitored and comprehensively evaluated in conjunction with the Q-value change trend in real-time coil monitoring data. The Weibull distribution model uses historical coil tuning data as prior information and employs maximum likelihood estimation to update shape and scale parameters in real time, predicting the coil detuning probability density function within the next 100 scanning cycles. When the predicted risk exceeds a set threshold, an automatic parameter adjustment strategy is triggered. If the decrease in signal-to-noise ratio is mainly due to an increase in the noise figure of the receiving channel, the combined receiving weights of the coil array are dynamically adjusted to suppress the gain of channels with higher noise. If an increase in reflected power due to load changes in RF transmission power is detected, the transmission frequency is adjusted in reverse based on the gradient descent algorithm, with the offset controlled within ±1kHz. If it is predicted that the coil is about to enter a high-fault period, the redundant coil channel switching mechanism is activated, and the RF pulse flip angle is reduced to decrease heat loss. Furthermore, scanning sequence parameters are automatically adjusted based on artifact characteristics; for example, the navigation echo acquisition frequency is increased when motion artifacts are severe, and the echo interval time is shortened when there are metal artifacts, maximizing the suppression of artifact amplification while ensuring scanning efficiency.

[0077] After all parameters are dynamically adjusted, the reconstructed image undergoes a final quality check. A full-reference image quality assessment algorithm is used to calculate the similarity between the current image and the standard template. When the similarity is below 95%, a secondary optimization process is triggered, readjusting the magnetic field compensation weights and artifact correction coefficients. Once the image is confirmed to be acceptable, the artifact pre-labeling information is overlaid on the final image as a semi-transparent layer, generating a diagnostic report containing coil performance attenuation trend charts, cumulative values ​​of specific absorption rates, and key sequence parameters. For hardware protection, the junction temperature and VSWR of the RF power amplifier are monitored in real time. If abnormal overheating or reflected power exceeding 50W is detected, a hardware-level interrupt is immediately executed, cutting off RF transmission and activating the water cooling system at full speed. Simultaneously, all operation logs, original K-space data fragments, and model decision paths are encrypted and stored on a local server, supporting time-stamped backtracking of the entire data chain for any scan node, providing complete data traceability support for equipment maintenance and algorithm iteration.

[0078] In summary, the embodiments of this application significantly improve the accuracy of MRI artifact recognition by acquiring and standardizing full-volume multi-source data, combined with precise dynamic magnetic field modeling and improved CNN-Transformer model evaluation. This allows for early warning of potential faults such as coil detuning, and the anomaly localization accuracy covers single components and channels. Three-dimensional visualization rendering and dynamic scanning parameter optimization reduce unplanned downtime, lower equipment maintenance costs and inspection scrap rates, and ensure the continuity of clinical examinations. Dual-end collaborative monitoring and adaptive control mechanisms improve image pass rate and reduce rework rates due to excessive artifacts. Weibull distribution lifetime prediction and adaptive adjustment of radio frequency parameters accurately predict the remaining lifetime of critical coils, control the response time to sudden faults, and improve the imaging efficiency, diagnostic quality, and operational safety of low-field MRI.

[0079] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 1001, the processor 1002, and the computer program stored on the memory 1001 and capable of running on the processor 1002.

[0080] When the processor 1002 executes the program, it implements a low-field magnetic resonance head imaging radio frequency coil method provided in the above embodiments.

[0081] Furthermore, electronic devices also include: Communication interface 1003 is used for communication between memory 1001 and processor 1002.

[0082] The memory 1001 is used to store computer programs that can run on the processor 1002.

[0083] The memory 1001 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0084] If the memory 1001, processor 1002, and communication interface 1003 are implemented independently, then the communication interface 1003, memory 1001, and processor 1002 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 10The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0085] Optionally, in a specific implementation, if the memory 1001, processor 1002, and communication interface 1003 are integrated on a single chip, then the memory 1001, processor 1002, and communication interface 1003 can communicate with each other through an internal interface.

[0086] The processor 1002 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0087] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described low-field magnetic resonance head imaging radio frequency coil method.

[0088] Furthermore, embodiments of this application also provide a computer program product, including a computer program or instructions, which, when executed, implement the aforementioned low-field magnetic resonance head imaging radio frequency coil method.

[0089] In the description of this specification, the references to "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, 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.

[0090] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0091] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0092] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0093] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0094] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A low-field magnetic resonance imaging head radio frequency coil system, characterized in that, include: The system comprises a flexible array coil sensing module, a dynamic field strength calibration module, an intelligent imaging optimization center, a three-dimensional imaging reconstruction unit, and an adaptive radio frequency control unit; among which... The flexible array coil sensing module is used to collect raw radio frequency signals from head magnetic resonance imaging, coil coupling status data, and environmental interference parameters. The dynamic field strength calibration module is used to construct a dynamic magnetic field distribution model of the head imaging area. Based on the coil coupling state data, environmental interference parameters and coil coupling characteristics, a multi-source magnetic field compensation algorithm is used to correct the magnetic field uniformity deviation in low field environment in real time to obtain magnetic field calibration data. The intelligent imaging optimization center is used to identify low-field-specific imaging artifact types, locate signal distortion sources, and generate imaging quality optimization schemes by combining the original radio frequency signal of the magnetic resonance and the magnetic field calibration data with the improved CNN-Transformer hybrid model. The three-dimensional imaging reconstruction unit is used to construct a three-dimensional imaging model of the head, jointly render the anatomical structure and functional signal intensity distribution of brain tissue, and output preliminary imaging results with artifact pre-labeling and signal fidelity calibration. The adaptive radio frequency control unit is used to predict the coil performance degradation trend and imaging risk based on the imaging quality optimization scheme, the preliminary imaging results and real-time coil monitoring data, and dynamically adjust the radio frequency transmission power, frequency and receiving gain parameters.

2. The low-field magnetic resonance head imaging radio frequency coil system according to claim 1, characterized in that, The flexible array coil sensing module includes a flexible array coil array, a coupling monitoring unit, and an environmental interference sensor group. The flexible array coil array is used to acquire raw radio frequency signals from head magnetic resonance imaging in real time. The coupling monitoring unit is used to acquire coil coupling state data such as inter-coupling coefficient and impedance matching state. The environmental interference sensor group is used to acquire environmental interference parameters such as magnetic field drift, electromagnetic radiation, and temperature.

3. The low-field magnetic resonance head imaging radio frequency coil system according to claim 1, characterized in that, The dynamic field strength calibration module includes a magnetic field modeling unit and a multi-source compensation unit. The magnetic field modeling unit is used to construct a dynamic magnetic field distribution model of the head imaging area based on head anatomical structure data and real-time magnetic field detection data. The multi-source compensation unit is used to use a multi-source magnetic field compensation algorithm, combined with coil coupling state data, environmental interference parameters and coil coupling characteristics, to correct the magnetic field uniformity deviation in low field environment in real time and generate magnetic field calibration data.

4. The low-field magnetic resonance head imaging radio frequency coil system according to claim 1, characterized in that, The intelligent imaging optimization center includes an artifact recognition unit and a distortion tracing module. The artifact recognition unit uses an improved CNN-Transformer hybrid model, combined with the original RF signal from magnetic resonance imaging and magnetic field calibration data, to identify low-field-specific imaging artifacts such as chemical shift, magnetic field inhomogeneity, and motion. The distortion tracing module is used to locate the physical location and influence range of signal distortion sources based on the spatial mapping relationship between artifact features and the magnetic field distribution model, and to generate an imaging quality optimization scheme.

5. A low-field magnetic resonance head imaging radio frequency coil system according to claim 1, characterized in that, The 3D imaging reconstruction unit includes an imaging modeling engine, a joint rendering unit, and an artifact calibration module. The imaging modeling engine is used to construct a 3D head imaging model based on the calibrated radio frequency signals, restoring the anatomical structure and functional signal distribution characteristics of brain tissue. The joint rendering unit is used to render the morphological structure and functional signal intensity distribution cloud map of brain tissue in real time, fusing and displaying anatomical and functional imaging. The artifact calibration module is used to pre-mark artifact regions in the imaging results and complete the calibration processing of radio frequency signals through a signal fidelity algorithm.

6. A low-field magnetic resonance head imaging radio frequency coil system according to claim 1, characterized in that, The adaptive radio frequency control unit includes a performance prediction module and a radio frequency parameter adjustment unit. The performance prediction module is used to predict the signal-to-noise ratio attenuation trend of the coil based on the Weibull distribution and coil performance degradation data, and to assess the imaging risk in combination with real-time imaging data. The radio frequency parameter adjustment unit is used to dynamically adjust the radio frequency transmit power, frequency and receive gain parameters according to the imaging quality optimization scheme, preliminary imaging results and real-time coil monitoring data.

7. A method applied to a low-field magnetic resonance head imaging radio frequency coil system according to any one of claims 1-6, characterized in that, The method includes: Acquire raw radio frequency signals, coil coupling status data, and environmental interference parameters from head magnetic resonance imaging; Based on the coil coupling state data, the environmental interference parameters, and the coil coupling characteristics, a dynamic magnetic field distribution model of the head imaging area is constructed. A multi-source magnetic field compensation algorithm is used to correct the magnetic field uniformity deviation in low-field environments in real time, and magnetic field calibration data is obtained. Based on the original radio frequency signal from the magnetic resonance imaging and the magnetic field calibration data, combined with the improved CNN-Transformer model, the unique imaging artifact types of low field are identified, the source of signal distortion is located, and an imaging quality optimization scheme is generated. Combined with the dynamic magnetic field distribution model and the original radio frequency signal from the head magnetic resonance imaging, a three-dimensional imaging model of the head is constructed. The anatomical structure and functional signal intensity distribution of brain tissue are jointly rendered, and the preliminary imaging results of artifact pre-labeling and signal fidelity calibration are output. Based on the imaging quality optimization scheme, the preliminary imaging results, and real-time coil monitoring data, and combined with the Weibull distribution to predict the coil performance degradation trend and imaging risk, the RF transmit power, frequency, and receive gain parameters are dynamically adjusted.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the low-field magnetic resonance head imaging radio frequency coil method of claim 7.

9. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When a computer program or instruction is executed, it implements a low-field magnetic resonance head imaging radio frequency coil method of claim 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When a computer program or instruction is executed, it implements a low-field magnetic resonance head imaging radio frequency coil method of claim 7.

Images