Data processing method and system applied to magnetic resonance image

Abstract

The invention provides a data processing method and system applied to a magnetic resonance image, and relates to the technical field of image processing, and the method comprises the steps: obtaining a velocity field time sequence of each voxel in a three-dimensional space in four-dimensional flow nuclear magnetic resonance imaging data; performing multi-scale frequency domain processing on the velocity field time sequence of each voxel to obtain a frequency domain characteristic matrix and a frequency domain energy ratio; based on the frequency domain characteristic matrix and the frequency domain energy ratio, performing weighted fusion to obtain a comprehensive entropy; constructing a target optimization function in combination with the comprehensive entropy and a preset regularity index of the frequency domain feature, solving the target optimization function, and screening out a voxel set meeting a set condition; the voxel set meeting the set condition is subjected to post-processing, image data of the blood vessel area in the four-dimensional flow magnetic resonance imaging data are determined, and effective recognition of the blood vessel area in the 4DFlowMRI data is achieved by combining the local and global frequency domain features and the regularity index of the frequency domain features.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a data processing method and system for magnetic resonance imaging. Background Art

[0002] Magnetic resonance imaging (MRI), as a non-invasive medical imaging technology, plays a vital role in clinical diagnosis. Among them, four-dimensional flow magnetic resonance imaging (4DFlowMRI) technology provides an unprecedented perspective for the diagnosis and research of cardiovascular diseases by virtue of its ability to capture the dynamics of blood flow during the cardiac cycle. However, accurately extracting information about vascular regions from massive 4DFlowMRI data has always been a major challenge in this field.

[0003] Traditional 4D Flow MRI data processing methods mainly rely on image segmentation algorithms. Although these algorithms can distinguish between blood vessels and non-vascular areas to a certain extent, they often find it difficult to achieve ideal accuracy when processing complex vascular structures or low signal-to-noise ratio data. In addition, due to the high-dimensional characteristics of 4D Flow MRI data (including time dimension and three-dimensional space dimension), traditional methods also have bottlenecks in processing speed and computational efficiency.

[0004] Therefore, it is necessary to provide a data processing method and system for magnetic resonance imaging to solve the above technical problems. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a data processing method and system for magnetic resonance imaging, which realizes the effective identification of vascular areas in 4DFlowMRI data by combining local and global frequency domain features and regularity indicators of frequency domain features.

[0006] The present invention provides a data processing method for magnetic resonance imaging, the method comprising the following steps: Expand the four-dimensional flow MRI data according to the time dimension to obtain the velocity field time series of each voxel; Performing multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio; Calculating local entropy and global entropy based on the frequency domain feature matrix and calculating the weights of the local entropy and the global entropy in real time based on the frequency domain energy proportion, and performing weighted fusion on the local entropy and the global entropy according to their respective weights to obtain a comprehensive entropy; A target optimization function is constructed by combining the comprehensive entropy and a preset regularity index of frequency domain characteristics, and the target optimization function is solved by using a gradient descent method to screen out a set of voxels that meet the set conditions; The voxel set that meets the set conditions is post-processed to determine image data of the blood vessel region in the four-dimensional flow magnetic resonance imaging data.

[0007] Preferably, the four-dimensional flow magnetic resonance imaging data is expanded according to the time dimension to obtain the velocity field time series of each voxel, including: Decomposing four-dimensional flow MRI data into three-dimensional velocity field data at multiple moments; For each voxel of the three-dimensional velocity field data, velocity values ​​at different time points are extracted to form a corresponding velocity field time series, and the velocity field time series is normalized.

[0008] Preferably, performing multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio includes: Performing wavelet transform on the velocity field time series to obtain wavelet coefficients jointly representing time and frequency; Extracting low-frequency, medium-frequency and high-frequency energy distribution characteristics according to the wavelet coefficients, and constructing a frequency domain feature matrix; A reference frequency range related to blood vessel flow is defined, and a frequency domain energy proportion of each voxel is calculated based on the reference frequency range.

[0009] Preferably, the calculation process of the comprehensive entropy includes: The frequency domain features of each voxel of the frequency domain feature matrix are normalized to obtain a probability distribution, and local entropy is calculated based on the probability distribution, wherein the calculation formula of the local entropy is: in, represents the local entropy, Representing voxels At different frequencies The probability distribution on ; The entire frequency domain feature matrix is ​​statistically analyzed to obtain a global probability distribution, and a global entropy is calculated based on the global probability distribution, wherein the calculation formula of the global entropy is: in, represents the global entropy; represents the global probability distribution; The weights of the local entropy and the global entropy are calculated according to the frequency domain energy proportion, wherein the weight calculation formula is: in, represents the weight of local entropy, represents the weight of global entropy, represents the energy ratio in the frequency domain, Represents the adjustment parameter, which is used to control the sensitivity of the weight to the frequency domain energy ratio; The local entropy and the global entropy are combined with their respective weights for weighted fusion to obtain the comprehensive entropy, wherein the calculation formula of the comprehensive entropy is: .

[0010] Preferably, the calculation method of the preset regularity index of the frequency domain feature includes: Extracting the energy standard deviation of each frequency subband in the frequency domain feature matrix; A reference interval is defined according to the physiological frequency range of blood vessel pulsation, the energy distribution ratio of the energy standard deviation within the reference interval is calculated, and the energy distribution ratio is used as a regularity indicator.

[0011] Preferably, the process of determining the voxel set includes: Linearly weighting the comprehensive entropy and the regularity index to obtain a target optimization function; The target optimization function is iteratively optimized using the gradient descent method with adaptive learning rate, and the optimization is terminated when the function value change is less than the preset threshold; All voxels are sorted according to the optimized objective function values, and a set of voxels whose objective function values ​​are higher than a dynamic threshold is screened out. The dynamic threshold is determined according to the percentile of the objective function value distribution of all voxels.

[0012] Preferably, the method for determining the dynamic threshold comprises the following steps: Obtain the optimization objective function values ​​of all voxels and calculate their numerical distribution; Based on a preset percentile range, extracting a corresponding objective function value from the numerical distribution as a candidate dynamic threshold; According to the expected spatial continuity of the vascular region, by counting the clustering density of the voxels corresponding to the candidate dynamic threshold in the three-dimensional space, the candidate threshold with the largest clustering density is selected as the final dynamic threshold, wherein the clustering density is calculated by counting the number of connected areas in the three-dimensional space in units of voxel neighborhoods, and the density is inversely proportional to the number of connected areas.

[0013] Preferably, the post-processing includes performing morphological processing on the set of voxels that meet the set conditions.

[0014] The present invention also provides a data processing system for magnetic resonance imaging, which is used to execute a data processing method for magnetic resonance imaging, and the system comprises: A sequence acquisition module is used to expand the four-dimensional flow MRI data according to the time dimension to obtain the velocity field time series of each voxel; A multi-scale processing module, used for performing multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio; A comprehensive entropy calculation module, used to calculate the local entropy and the global entropy based on the frequency domain feature matrix and to calculate the weights of the local entropy and the global entropy in real time based on the frequency domain energy proportion, and to perform weighted fusion of the local entropy and the global entropy according to their respective weights to obtain a comprehensive entropy; A voxel set screening module is used to construct a target optimization function by combining the comprehensive entropy and the regularity index of the preset frequency domain characteristics, and solve the target optimization function by using the gradient descent method to screen out the voxel set that meets the set conditions; The post-processing module is used to post-process the voxel set that meets the set conditions to determine the image data of the blood vessel area in the four-dimensional flow magnetic resonance imaging data.

[0015] Compared with the related art, the data processing method and system for magnetic resonance imaging provided by the present invention have the following beneficial effects: The present invention first expands 4DFlowMRI data according to the time dimension to obtain the velocity field time series of each voxel; then performs multi-scale frequency domain processing on the velocity field time series to extract the frequency domain feature matrix and the frequency domain energy ratio; then calculates the local entropy and the global entropy based on the frequency domain feature matrix, and adjusts the weights of the two in real time according to the frequency domain energy ratio to obtain the comprehensive entropy; then combines the comprehensive entropy and the regularity index of the preset frequency domain characteristics to construct a target optimization function, and solves it using the gradient descent method; finally, post-processes the voxel set that meets the set conditions to determine the image data of the vascular area.

[0016] The invention not only overcomes the limitations of traditional 4D Flow MRI data processing methods in identifying vascular regions, but also improves processing speed and computational efficiency, providing a more accurate and efficient tool for the diagnosis and research of cardiovascular diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a data processing method for magnetic resonance imaging provided by the present invention; Figure 2 A module structure diagram of a data processing system for magnetic resonance imaging provided by the present invention. DETAILED DESCRIPTION

[0018] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only the parts related to the present invention, rather than all structures, are shown in the accompanying drawings. In addition, the embodiments of the present invention and the features in the embodiments may be combined with each other without conflict.

[0019] It should also be noted that, for ease of description, only the parts related to the present invention, but not all of the contents, are shown in the accompanying drawings. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations therein can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it can also have additional steps not included in the accompanying drawings. The process can correspond to methods, functions, procedures, subroutines, subprograms, etc.

[0020] Embodiment 1

[0021] Four-dimensional flow magnetic resonance imaging data contains blood flow velocity information in three-dimensional space coordinates and time dimensions, and its essence is the quantitative representation of dynamic vascular flow in time and space. Since the pulsating characteristics of blood flow in blood vessels are closely related to physiological activities, the velocity changes in the time dimension can reflect the hemodynamic characteristics, so it is necessary to extract the velocity evolution law of each voxel (the smallest unit in three-dimensional space) over time from the original four-dimensional data. In traditional methods, three-dimensional static velocity field analysis is difficult to capture dynamic characteristics, and the direct processing of four-dimensional data has problems such as high computational complexity and noise sensitivity. It needs to be decomposed into a time series through structured decomposition to adapt to subsequent frequency domain analysis and feature extraction. Therefore, the present invention provides a data processing method for magnetic resonance imaging.

[0022] refer to Figure 1 As shown, the present invention provides a data processing method for magnetic resonance imaging, comprising the following steps: S1: Expand the 4D flow MRI data according to the time dimension to obtain the velocity field time series of each voxel.

[0023] Wherein, step S1 specifically includes the following steps: S11: Decomposing the four-dimensional flow magnetic resonance imaging data into three-dimensional velocity field data at multiple moments.

[0024] In this embodiment, the four-dimensional stream MRI data file (usually in NIfTI or DICOM format) is first read and its data dimension structure is parsed. The four-dimensional data is composed of X, Y, Z and T, where X, Y and Z are three-dimensional spatial coordinates and T is the number of time points (e.g., 20-30 phases in a cardiac cycle). After slicing by time dimension, each moment Corresponds to a 3D velocity field data block, which contains the velocity components of each voxel in the X, Y and Z directions , and To simplify the calculation, the velocity vector modulus is usually used as the velocity scalar value of the voxel, and T independent three-dimensional velocity field data files are finally generated.

[0025] S12: For each voxel of the three-dimensional velocity field data, extract velocity values ​​at different time points to form a corresponding velocity field time series, and perform normalization processing on the velocity field time series.

[0026] In this embodiment, step S12 further extracts the The velocity values ​​of voxels are used to form a one-dimensional time series with a length of T. In order to improve the stability of subsequent frequency domain analysis, the time series is normalized by Z-score to eliminate the differences in velocity magnitudes between different voxels.

[0027] Through the above steps, the spatiotemporal characteristics of the four-dimensional data are decoupled into separate representations of spatial distribution and temporal evolution, which not only retains the dynamic information of vascular pulsation, but also reduces the data dimension. The normalization process effectively suppresses the velocity fluctuation deviation caused by magnetic field inhomogeneity or noise, making the frequency domain characteristics of voxels at different positions comparable, providing standardized input for subsequent multi-scale frequency domain analysis (such as wavelet transform) to avoid high-frequency noise from submerging the real blood flow signal. This process takes into account both computational efficiency and feature fidelity, and is the basis for subsequent entropy fusion and vascular region screening.

[0028] S2: Perform multi-scale frequency domain processing on the velocity field time series of each voxel to obtain the frequency domain feature matrix and frequency domain energy ratio.

[0029] The process of obtaining the frequency domain feature matrix and the frequency domain energy ratio includes the following steps: S21: Perform wavelet transform on the velocity field time series to obtain wavelet coefficients that jointly represent time and frequency.

[0030] In this embodiment, the Daubechies wavelet basis is selected to perform discrete wavelet transform on the normalized velocity field time series, and the decomposition layer number is set to 4-6 layers to cover the physiologically relevant frequency band of 0.5-20 Hz. Multi-level decomposition is performed on the time series of each voxel to obtain approximate coefficients (low-frequency components) and detail coefficients (high-frequency components). For example, the first-layer detail coefficients correspond to the highest frequency band (10-20 Hz, which may contain noise), and the fourth-layer approximate coefficients correspond to the lowest frequency band (0.5-1.25 Hz, reflecting cardiac cycle fluctuations). The time-frequency matrix is ​​generated by reconstructing the wavelet coefficients of each frequency band. To enhance the feature stability, the absolute value of the wavelet coefficients of each frequency band is taken and the sliding average (window length 3 to 5 time points) is taken to suppress instantaneous noise fluctuations.

[0031] S22: Extract the energy distribution characteristics of low frequency, medium frequency and high frequency according to the wavelet coefficients, and construct the frequency domain feature matrix.

[0032] In this embodiment, based on the number of wavelet decomposition layers (such as 4 layers), the detail coefficients (D1-D4) and the approximate coefficient (A4) of each layer are divided into three sub-bands: low frequency, medium frequency, and high frequency: Low-frequency subband (0.5-1.25 Hz): corresponds to the 4th layer approximation coefficient (A4), representing the slow changes in blood flow dominated by the cardiac cycle; Mid-frequency subband (1.25-5 Hz): reconstructed from the third and second layer detail coefficients (D3, D2), reflecting the local pulsation and turbulence of blood vessels; High-frequency subband (5-20 Hz): reconstructed from the first layer detail coefficients (D1), which may contain noise or rapid blood flow fluctuations.

[0033] For each subband, calculate its energy value, which is: ; ; .

[0034] Finally, the three-band energy values ​​of each voxel are normalized into a probability distribution, and the probability distributions of the three-band energy values ​​of each voxel are combined into a three-dimensional frequency domain feature matrix according to the spatial coordinates. Each channel of the matrix corresponds to a frequency band energy ratio.

[0035] S23: defining a reference frequency range related to vascular flow, and calculating a frequency domain energy proportion of each voxel based on the reference frequency range.

[0036] In this embodiment, the reference frequency range is defined based on the physiological characteristics of human vascular pulsation. The main frequency of arterial blood flow affected by cardiac contraction is 0.8-2Hz (corresponding to a heart rate of 60-120 beats / minute), and the main frequency of venous return modulated by respiration is about 0.1-0.3Hz. Taking into account the clinical data, the reference frequency range is set to 0.5-5Hz to cover the main blood flow signals. For each voxel, the energy proportion of its wavelet coefficient in the reference frequency band is calculated: Extract all wavelet coefficients belonging to the reference frequency band (0.5-5 Hz) in the time-frequency matrix to form a subset; Calculate total energy and reference band energy; The frequency domain energy ratio is defined as the reference frequency band energy divided by the total energy, and is smoothed to reduce spatial mutations.

[0037] S3: Calculate the local entropy and global entropy based on the frequency domain feature matrix, and calculate the weights of the local entropy and global entropy in real time based on the frequency domain energy ratio, and perform weighted fusion of the local entropy and global entropy according to their respective weights to obtain the comprehensive entropy.

[0038] In magnetic resonance imaging data processing, entropy analysis of frequency domain features is often used to quantify the complexity and regularity of signals, but traditional methods often use local entropy (reflecting the uncertainty of the frequency domain distribution of a single voxel) or global entropy (reflecting the statistical characteristics of the entire data set) alone, ignoring the complementarity of the two in vascular area detection. Local entropy is sensitive to noise, which may cause isolated noise voxels to be misjudged as blood vessels; while global entropy is overly dependent on the overall distribution, which may weaken the dynamic characteristics of local blood flow. In addition, the frequency domain energy of blood flow in vascular areas is usually concentrated in frequency bands related to physiological activities (such as heart beat frequency), but existing methods do not dynamically associate the frequency domain energy distribution characteristics with entropy calculations, resulting in insufficient differentiation between blood vessels and surrounding tissues. Therefore, this application adopts a combined approach for processing.

[0039] The calculation process of comprehensive entropy includes: S31: Normalize the frequency domain features of each voxel of the frequency domain feature matrix to obtain a probability distribution, and calculate the local entropy based on the probability distribution, wherein the calculation formula of the local entropy is: in, represents the local entropy, Representing voxels At different frequencies The probability distribution on .

[0040] In this embodiment, when the energy proportion of a certain frequency sub-band is significant (such as the main frequency of blood vessel pulsation), the local entropy value is low, indicating that the frequency domain distribution is concentrated; if the energy is dispersed (such as noise interference), the entropy value increases.

[0041] S32: Perform statistical analysis on the entire frequency domain feature matrix to obtain a global probability distribution, and calculate a global entropy based on the global probability distribution, wherein the calculation formula of the global entropy is: in, represents the global entropy; represents the global probability distribution.

[0042] In this embodiment, the energy values ​​of all frequency sub-bands of all voxels are firstly summarized to generate a global energy distribution, and then the global entropy is calculated. The global entropy reflects the frequency domain consistency of the entire data set. For example, if the energy of most voxels is concentrated in the blood vessel-related frequency band, the global entropy is low.

[0043] S33: Calculate the weights of local entropy and global entropy according to the frequency domain energy ratio, where the weight calculation formula is: in, represents the weight of local entropy, represents the weight of global entropy, represents the energy ratio in the frequency domain, Represents an adjustment parameter in the range of [0, 1], which is used to control the sensitivity of the weight to the frequency domain energy ratio.

[0044] In this embodiment, when When it is larger, the weight depends more on the frequency domain energy ratio, that is, it tends to adjust the weight according to the frequency domain energy distribution of the voxel. When it is smaller, the weight is closer to a fixed value, reducing the dependence on the frequency domain energy distribution.

[0045] S34: Combine the local entropy and the global entropy and their respective weights for weighted fusion to obtain the comprehensive entropy, where the calculation formula of the comprehensive entropy is: .

[0046] In this embodiment, the comprehensive entropy effectively balances local details and global statistical characteristics: in the vascular area, the higher frequency domain energy ratio drives the weight to be biased towards local entropy, highlighting the concentrated frequency domain distribution of blood flow pulsation; in the non-vascular area, the lower frequency domain energy ratio makes the weight biased towards global entropy, suppressing the local entropy fluctuations caused by random noise.

[0047] S4: The target optimization function is constructed by combining the comprehensive entropy and the regularity index of the preset frequency domain characteristics, and the target optimization function is solved by the gradient descent method to screen out the voxel set that meets the set conditions.

[0048] In step S4, the calculation method of the preset regularity index of the frequency domain feature includes: First, the energy standard deviation of each frequency subband in the frequency domain feature matrix is ​​extracted.

[0049] In this embodiment, each voxel in the frequency domain feature matrix contains the energy proportion of the three sub-bands of low frequency, medium frequency and high frequency (as defined in step S22). For each sub-band (low frequency, medium frequency, high frequency), the standard deviation of its energy value over time is calculated, which is specifically expressed as follows: Low frequency subband (0.5-1.25 Hz): Energy standard deviation .

[0050] IF subband (1.25-5 Hz): Energy standard deviation .

[0051] High frequency sub-band (5-20Hz): Energy standard deviation .

[0052] The standard deviation reflects the severity of the fluctuation of the frequency energy over time. Since the vascular area is driven by periodic heart beats, the standard deviation of its main frequency sub-band (such as intermediate frequency 1.25-5Hz) should be low and stable, while the standard deviation of non-vascular areas (such as noise) may show irregular high fluctuations.

[0053] Then, the reference interval is defined according to the physiological frequency range of vascular pulsation, the energy distribution ratio of the energy standard deviation within the reference interval is calculated, and the energy distribution ratio is used as a regularity indicator.

[0054] In this embodiment, according to the physiological characteristics of blood vessel pulsation, the reference frequency range is set to 0.5-5 Hz (covering low-frequency and medium-frequency sub-bands). The ratio of the sum of the standard deviations of the sub-bands (i.e., low-frequency and medium-frequency) within this range to the sum of the standard deviations of the full frequency bands (low+medium+high) is calculated. , the specific formula is as follows: The ratio The closer it is to 1, the more the energy fluctuation is concentrated in the blood vessel-related frequency band and the stronger the regularity.

[0055] Comparative Example Three-dimensional Gaussian filtering (kernel size 3 × 3 × 3, standard deviation 1.0) was performed to suppress the influence of isolated noise points and enhance the spatial continuity of the vascular region.

[0056] After determining the regularity index of the frequency domain feature, all voxels are screened to obtain the required voxel set. The determination process includes: First, the comprehensive entropy and regularity index are linearly weighted to obtain the target optimization function.

[0057] In this embodiment, and are the weight coefficients of comprehensive entropy and regularity index respectively ( + =1, default =0.6, = 0.4), determined by cross-validation optimization.

[0058] Secondly, the target optimization function is iteratively optimized using the gradient descent method with adaptive learning rate, and the optimization is terminated when the change in the function value is less than the preset threshold.

[0059] In this embodiment, the objective function is optimized using the gradient descent method with adaptive learning rate: Initialize the learning rate is 0.1, and the objective function gradient is calculated at each iteration , if three consecutive iterations If it decreases, the learning rate will increase ( ← ×1.2), otherwise it decreases ( ← ×0.5). The termination condition is the adjacent iteration Change .

[0060] After the iterative optimization is completed, a dynamic threshold needs to be determined to facilitate the subsequent voxel screening. The specific steps are: Get the optimized objective function values ​​of all voxels and calculate their numerical distribution.

[0061] In this embodiment, the objective function values ​​after optimization of all voxels (obtained by weighting the comprehensive entropy and regularity index) are obtained, and their numerical distribution is statistically analyzed. To avoid interference from extreme values, the objective function values ​​are truncated (e.g., outliers in the first 1% and last 1% are eliminated) to generate a smooth probability density function.

[0062] Based on the preset percentile range, the corresponding objective function value is extracted from the numerical distribution as a candidate dynamic threshold.

[0063] In this embodiment, based on the preset percentile range , extract multiple candidate thresholds from the distribution, each candidate threshold corresponds to Quantile.

[0064] According to the expected spatial continuity of the vascular region, the candidate threshold with the largest clustering density is selected as the final dynamic threshold by counting the clustering density of the voxels corresponding to the candidate dynamic threshold in three-dimensional space. The clustering density is calculated by counting the number of connected regions in three-dimensional space in units of voxel neighborhoods, and the density is inversely proportional to the number of connected regions.

[0065] In this embodiment, next, for each candidate threshold, a set of voxels with an objective function value ≥ the candidate threshold is screened out, and its clustering density in three-dimensional space is calculated. The clustering density is defined as: within a 3×3×3 neighborhood centered on each voxel, the number of connected regions is counted (using 26 neighborhood connectivity judgment), and the density is inversely proportional to the number of connected regions. The higher the density, the stronger the spatial continuity of the voxel set, and the more consistent with the tubular structure characteristics of the blood vessel. The higher the density, the stronger the spatial continuity of the voxel set, and the more consistent with the tubular structure characteristics of the blood vessel. Finally, the candidate threshold that maximizes the density is selected as the dynamic threshold.

[0066] Finally, all voxels are sorted according to the optimized objective function value, and the voxel set with objective function values ​​higher than the dynamic threshold is screened out.

[0067] S5: Post-processing the voxel set that meets the set conditions to determine the image data of the blood vessel region in the four-dimensional flow magnetic resonance imaging data.

[0068] In this embodiment, the post-processing is to perform morphological processing on the voxel set that meets the set conditions. The specific process is as follows: Three-dimensional closing operation fills the interior holes: A 3D closing operation (dilation followed by erosion) is performed on a set of voxels using a spherical structural element (radius 2-3 voxels). The dilation operation marks the area with at least one vascular voxel in the voxel neighborhood (3×3×3) as a vessel, filling in small breaks caused by noise or threshold deviation; the erosion operation then removes the edge protrusions introduced by the dilation and restores the original size of the vessel. For example, for the fracture area at the branch of the intracranial artery, the closing operation can connect the discontinuities with a spacing of less than or equal to 2 voxels.

[0069] Three-dimensional opening operation removes isolated noise: The same structural element is used to perform a 3D opening operation (erosion followed by dilation). The erosion operation removes isolated noise points (single voxels without adjacent vascular voxels) and retains continuous areas; the dilation operation restores the eroded vascular edges. For example, the scattered noise voxels generated by respiratory motion in liver images can be removed while retaining the main structure of the portal vein.

[0070] Multi-scale connected region screening: Calculate the volume (number of voxels) of all connected regions and set a dynamic volume threshold: Extract the distribution histogram of the volume of all connected regions and take the 90th percentile as the reference threshold; The final volume threshold ensured that at least 10 voxels of small vessel branches were retained.

[0071] Connected regions with a volume smaller than a threshold are deleted to eliminate residual noise. For example, at the carotid bifurcation, false plaque signals with a volume smaller than 15 voxels can be removed.

[0072] Anisotropic diffusion smooths edges: A three-dimensional anisotropic diffusion filter (Perona-Malik model) was applied, with the number of iterations set to 3-5 and the conductivity coefficient k=0.05. This operation smoothes the edges along the long axis of the vessel while preserving the sharpness of the branches. For example, it can correct the blurring of the vessel wall caused by the closing operation and improve the clarity of the edges of the coronary arteries.

[0073] Embodiment 2 The present invention also provides a data processing system for magnetic resonance imaging, which is used to execute a data processing method for magnetic resonance imaging, referring to Figure 2 As shown, the system includes: The sequence acquisition module 100 is used to expand the four-dimensional flow magnetic resonance imaging data according to the time dimension to obtain the velocity field time series of each voxel.

[0074] The multi-scale processing module 200 is used to perform multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio.

[0075] The comprehensive entropy calculation module 300 is used to calculate the local entropy and the global entropy based on the frequency domain feature matrix and to calculate the weights of the local entropy and the global entropy in real time based on the frequency domain energy ratio, and to perform weighted fusion of the local entropy and the global entropy according to their respective weights to obtain the comprehensive entropy.

[0076] The voxel set screening module 400 is used to construct a target optimization function by combining comprehensive entropy and a preset regularity index of frequency domain characteristics, and solve the target optimization function by using a gradient descent method to screen out a voxel set that meets the set conditions.

[0077] The post-processing module 500 is used to perform post-processing on the voxel set that meets the set conditions to determine the image data of the blood vessel area in the four-dimensional flow magnetic resonance imaging data.

[0078] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0079] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0080] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. A data processing method for magnetic resonance imaging, characterized in that: The method comprises the following steps: Expand the four-dimensional flow MRI data according to the time dimension to obtain the velocity field time series of each voxel; Performing multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio; Calculating local entropy and global entropy based on the frequency domain feature matrix and calculating the weights of the local entropy and the global entropy in real time based on the frequency domain energy proportion, and weighted fusion of the local entropy and the global entropy according to their respective weights to obtain a comprehensive entropy; A target optimization function is constructed by combining the comprehensive entropy and a preset regularity index of frequency domain characteristics, and the target optimization function is solved by using a gradient descent method to screen out a set of voxels that meet the set conditions; The voxel set that meets the set conditions is post-processed to determine image data of the blood vessel region in the four-dimensional flow magnetic resonance imaging data.

2. A data processing method for magnetic resonance imaging according to claim 1, characterized in that: The four-dimensional flow magnetic resonance imaging data is expanded according to the time dimension to obtain the velocity field time series of each voxel, including: Decomposing four-dimensional flow MRI data into three-dimensional velocity field data at multiple moments; For each voxel of the three-dimensional velocity field data, velocity values ​​at different time points are extracted to form a corresponding velocity field time series, and the velocity field time series is normalized.

3. A data processing method for magnetic resonance imaging according to claim 2, characterized in that: The multi-scale frequency domain processing is performed on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio, including: Performing wavelet transform on the velocity field time series to obtain wavelet coefficients jointly representing time and frequency; Extracting low-frequency, medium-frequency and high-frequency energy distribution characteristics according to the wavelet coefficients, and constructing a frequency domain feature matrix; A reference frequency range related to blood vessel flow is defined, and a frequency domain energy proportion of each voxel is calculated based on the reference frequency range.

4. A data processing method for magnetic resonance imaging according to claim 3, characterized in that: The calculation process of the comprehensive entropy includes: The frequency domain features of each voxel of the frequency domain feature matrix are normalized to obtain a probability distribution, and local entropy is calculated based on the probability distribution, wherein the calculation formula of the local entropy is: in, represents the local entropy, Representing voxels At different frequencies The probability distribution on ; The entire frequency domain feature matrix is ​​statistically analyzed to obtain a global probability distribution, and a global entropy is calculated based on the global probability distribution, wherein the calculation formula of the global entropy is: in, represents the global entropy; represents the global probability distribution; The weights of the local entropy and the global entropy are calculated according to the frequency domain energy proportion, wherein the weight calculation formula is: in, represents the weight of local entropy, represents the weight of global entropy, represents the energy ratio in the frequency domain, Represents the adjustment parameter, which is used to control the sensitivity of the weight to the frequency domain energy ratio; The local entropy and the global entropy are combined with their respective weights for weighted fusion to obtain the comprehensive entropy, wherein the calculation formula of the comprehensive entropy is: 。 5. The data processing method for magnetic resonance imaging according to claim 3, characterized in that: The calculation method of the preset regularity index of the frequency domain feature includes: Extracting the energy standard deviation of each frequency subband in the frequency domain feature matrix; A reference interval is defined according to the physiological frequency range of blood vessel pulsation, the energy distribution ratio of the energy standard deviation within the reference interval is calculated, and the energy distribution ratio is used as a regularity indicator.

6. A data processing method for magnetic resonance imaging according to claim 5, characterized in that: The process of determining the voxel set includes: Linearly weighting the comprehensive entropy and the regularity index to obtain a target optimization function; The target optimization function is iteratively optimized using the gradient descent method with adaptive learning rate, and the optimization is terminated when the function value change is less than the preset threshold; All voxels are sorted according to the optimized objective function values, and a set of voxels whose objective function values ​​are higher than a dynamic threshold is screened out. The dynamic threshold is determined according to the percentile of the objective function value distribution of all voxels.

7. A data processing method for magnetic resonance imaging according to claim 6, characterized in that: The method for determining the dynamic threshold comprises the following steps: Obtain the optimization objective function values ​​of all voxels and calculate their numerical distribution; Based on a preset percentile range, extracting a corresponding objective function value from the numerical distribution as a candidate dynamic threshold; According to the expected spatial continuity of the vascular region, by counting the clustering density of the voxels corresponding to the candidate dynamic threshold in the three-dimensional space, the candidate threshold with the largest clustering density is selected as the final dynamic threshold, wherein the clustering density is calculated by counting the number of connected areas in the three-dimensional space in units of voxel neighborhoods, and the density is inversely proportional to the number of connected areas.

8. The data processing method for magnetic resonance imaging according to claim 1, characterized in that: The post-processing includes performing morphological processing on the voxel set that meets the set conditions.

9. A data processing system for magnetic resonance imaging, used for executing a data processing method for magnetic resonance imaging according to any one of claims 1 to 8, characterized in that: The system comprises: A sequence acquisition module is used to expand the four-dimensional flow MRI data according to the time dimension to obtain the velocity field time series of each voxel; A multi-scale processing module, used for performing multi-scale frequency domain processing on the velocity field time series of each voxel to obtain a frequency domain feature matrix and a frequency domain energy ratio; A comprehensive entropy calculation module, used to calculate the local entropy and the global entropy based on the frequency domain feature matrix and to calculate the weights of the local entropy and the global entropy in real time based on the frequency domain energy proportion, and to perform weighted fusion of the local entropy and the global entropy according to their respective weights to obtain a comprehensive entropy; A voxel set screening module is used to construct a target optimization function by combining the comprehensive entropy and the regularity index of the preset frequency domain characteristics, and solve the target optimization function by using the gradient descent method to screen out the voxel set that meets the set conditions; The post-processing module is used to post-process the voxel set that meets the set conditions to determine the image data of the blood vessel area in the four-dimensional flow magnetic resonance imaging data.

Images