Magnetic resonance method for evaluating rehabilitation curative effect of child cerebral palsy

Abstract

The invention discloses a magnetic resonance method for evaluating the rehabilitation curative effect of child cerebral palsy, and relates to the technical field of medical image processing. The method comprises: obtaining structural image data (T1 weighted imaging and diffusion tensor imaging) and resting state image data of each child patient; and extracting brain and tissue boundaries from the T1 weighted imaging by using a 3D-Unet segmentation model. The resting state and structure image data are preprocessed and aligned to the brain boundary, and then the data are registered to a child standard brain template; based on a child standard brain template, obtaining and normalizing various clinical indexes, splicing the clinical indexes into multi-index vector data, inputting the multi-index vector data into a pre-trained support vector machine model, and outputting clinical index differences; and determining an abnormal brain region of each child according to the difference, finally providing a multi-modal fusion index analysis report, and observing changes of brain functions and white matter integrity before and after rehabilitation.

Description

Technical Field

[0001] This application relates to the technical field of medical image processing, and particularly to a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy. Background Art

[0002] Cerebral palsy is a high-incidence birth defect disease, ranking first among children's disabling diseases, bringing a heavy burden to the families of children with the disease and society. The rehabilitation of children with cerebral palsy has always focused on improving motor function. Remodeling the neural circuits corresponding to motor function in the brains of children through rehabilitation treatment is the core of rehabilitation. At present, the evaluation of the rehabilitation efficacy of children with cerebral palsy mainly relies on clinical observation and neuropsychological tests. Although these methods can provide certain information on the rehabilitation effect, it usually takes a long time to observe obvious changes, and the results are easily affected by the state of the children and the experience of the evaluators, with a certain degree of subjectivity.

[0003] In the prior art, researchers have been exploring more accurate and sensitive methods to evaluate the rehabilitation effect of cerebral palsy. With the development of imaging technology, the application of advanced technologies such as resting-state functional magnetic resonance imaging and diffusion imaging helps to evaluate the dynamic changes of neuronal activity and white matter integrity, and has been widely used in the research of other nervous system diseases. Resting-state functional magnetic resonance imaging technology reflects the changes in the brain activation pattern by measuring the changes in hemodynamics caused by neuronal activity; diffusion imaging technology effectively evaluates the changes in the integrity of white matter tracts by measuring the diffusion movement of water molecules, and can provide evidence for rehabilitation treatment.

[0004] However, most of the existing studies on resting-state functional magnetic resonance and diffusion imaging are group studies of conventional technical methods, without fully considering the true level of individual child patients, resulting in an incomplete understanding of the brain development and pathological changes in children. Summary of the Invention

[0005] Based on this, it is necessary to provide a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy in view of the above technical problems, so as to solve the problem that the prior art does not fully consider the true level of individual child patients, resulting in an incomplete understanding of the brain development and pathological changes in children.

[0006] The present invention adopts the following technical solutions: The present invention provides a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy, including: Obtaining the structural image data and resting-state image data of an individual child patient to be evaluated; Preprocessing the resting-state image data and the structural image data; and registering the preprocessed resting-state image data and structural image data to a children's standard brain template respectively; Obtain multiple clinical indicators of the registered resting-state imaging data and structural imaging data in the pediatric standard brain template respectively and normalize them; splice the normalized multiple clinical indicators into multi-index vector data and input it into a pre-trained support vector machine model to output the clinical indicator differences; based on the clinical indicator differences, obtain the abnormal brain regions of the individual pediatric patients to be evaluated; Determine a multi-modal fusion index analysis report on the rehabilitation efficacy of pediatric cerebral palsy based on the abnormal brain regions of the individual pediatric patients to be evaluated.

[0007] Preferably, preprocess the resting-state imaging data, specifically including: Convert the resting-state imaging data from DICOM format to NIFTI format; Perform temporal slice correction on the NIFTI-format resting-state imaging data; and perform head motion correction on the corrected resting-state imaging data; Smooth the aligned resting-state imaging data using a Gaussian smoothing kernel; Remove the artifacts related to head motion in the smoothed resting-state imaging data; Adopt temporal filtering to filter out the physiological signal noise of the resting-state imaging data after removing artifacts, and obtain the preprocessed resting-state imaging data.

[0008] Preferably, the structural imaging data includes: diffusion imaging data and T1-weighted imaging data; Preprocess the diffusion imaging data, specifically including: Convert the resting-state imaging data from DICOM format to NIFTI format; Perform slice outlier detection on the diffusion imaging data, and remove the Gibbs-ringing artifacts and noise of the diffusion imaging data to obtain the denoised diffusion imaging data; Synthesize an undistorted non-diffusion weighted image from the denoised diffusion imaging data and combine it with the original non-diffusion weighted image, and then correct the diffusion imaging data; Perform bias field correction on the corrected diffusion imaging data; Adopt a self-supervised learning algorithm to filter the bias field corrected diffusion imaging data to obtain the preprocessed diffusion imaging data; Preprocess the T1-weighted imaging data, including: Resample the T1-weighted imaging data to a specified size space; Use the trained 3DUnet segmentation model to extract the brain tissue of the individual pediatric patients to be evaluated from the space; Resample the brain tissue to another specified size space again.

[0009] Preferably, the clinical indicators include functional indicators and structural indicators; multiple clinical indicators of the registered resting-state imaging data and structural imaging data in the pediatric standard brain template are obtained respectively, specifically including: Based on the registered resting-state imaging data in the pediatric standard brain template, functional indicators are calculated, and the functional indicators include the whole-brain functional connectivity matrix, regional homogeneity and its average value, and low-frequency signal amplitude and its average value; Based on the registered structural imaging data in the pediatric standard brain template, structural indicators are calculated, and the structural indicators include the whole-brain structural connectivity matrix, fractional anisotropy and its average value, and mean diffusivity and its average value.

[0010] Preferably, the whole-brain functional connectivity matrix is constructed by dividing the pediatric standard brain template into multiple regions of interest, calculating the functional connectivity strength between each region of interest, and based on the functional connectivity strength between the regions of interest; Regional homogeneity is obtained by calculating the coordination coefficient between each voxel in the pediatric standard brain template and its adjacent voxels; The low-frequency signal amplitude is obtained by converting the registered resting-state imaging data into a frequency-domain signal through Fourier transform and calculating the power spectral density of each voxel in the pediatric standard brain template where the frequency-domain signal is in the low-frequency band.

[0011] Preferably, the whole-brain structural connectivity matrix is constructed by reconstructing the white matter fiber bundles of the registered structural imaging data through fiber tracking technology and calculating the connection strength of each fiber bundle, and based on the connection strength of each fiber bundle.

[0012] Preferably, the multimodal fusion index analysis report for determining the rehabilitation efficacy of pediatric cerebral palsy specifically includes: Extracting the multimodal fusion indexes of the abnormal brain regions of the patient child to be evaluated; the multimodal fusion indexes include abnormal brain region functional connectivity, abnormal brain region average regional homogeneity, abnormal brain region average low-frequency signal amplitude, abnormal brain region structural connectivity, abnormal brain region average fractional anisotropy, and abnormal brain region average diffusivity; According to the multimodal fusion indexes, a multimodal fusion index analysis report is obtained through automated software.

[0013] The above at least one technical solution adopted by the present invention can achieve the following beneficial effects: In a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention, preprocessing is performed on resting-state image data and diffusion imaging data, and they are aligned to the brain boundary of the children's structural images; the preprocessed magnetic resonance image data and the children's brain T1-weighted imaging data are registered to the children's standard brain template by using the Riemannian manifold registration method; based on the children's standard brain template, clinical indicators of the registered magnetic resonance image data are calculated; based on different anatomical regions of the template, tools such as FSL are used to calculate the registered clinical indicators; through canonical correlation analysis to process the structural indicators and functional indicators, and then the relevant components of the structural and functional indicators are output and normalized, the normalized indicators are spliced into multi-index vector data, and are input into a pre-trained support vector machine model for indicator analysis, and abnormal brain regions of each child patient individual are obtained according to the indicator differences; according to the abnormal brain regions of each child individual, a multi-modal fusion indicator analysis report on the rehabilitation efficacy of children with cerebral palsy is issued to observe the changes in brain function and white matter integrity of each child patient individual before and after rehabilitation.

[0014] The present invention realizes the registration of the magnetic resonance image data of children individuals with the children's standard brain template through the preprocessing of the resting-state image data and diffusion imaging data of each child individual and the use of the Riemannian manifold brain registration algorithm. Through the calculation of clinical indicators and canonical correlation analysis, and by using a support vector machine model for data analysis, considering the individual differences in the brain structure of patients, the post-processing process of magnetic resonance image data is made more concise, and data analysis can be carried out according to the specific conditions of each patient individual, improving the accuracy of diagnosis and the pertinence of rehabilitation efficacy evaluation, realizing precise image analysis for each child patient individual, avoiding the problem that individual differences are averaged in group studies, ensuring that the evaluation results are closer to the real conditions of each child with cerebral palsy, thereby helping clinicians optimize treatment plans, realizing precise and personalized rehabilitation treatment, improving the symptom remission rate of children with cerebral palsy, and effectively reducing the family and social economic burden of children. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings: Figure 1 It is a schematic flow chart of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 2 It is the T1-weighted imaging, diffusion tensor imaging and resting-state functional image data of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 3 It is the segmentation and resampling effect diagram of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 4 Effect display of registering the resting-state image data of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention to the standard brain template of children; Figure 5 Effect display diagram of registering T1WI of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention to the standard brain template of children; Figure 6 Effect display diagram of registering T1WI of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention to the standard brain template of children; Figure 7 Flow chart for calculating functional MRI indexes of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 8 Whole-brain functional connectivity matrix diagram of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 9 Diagram of calculating FA parameter map and MD parameter map from diffusion imaging data of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 10 Flow chart for calculating structural indexes of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 11 Whole-brain structural connectivity matrix diagram of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention; Figure 12 Analysis report diagram of the rehabilitation efficacy of children with cerebral palsy issued by the automated software of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy provided by the present invention. Detailed implementation manners

[0016] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0017] The following will, with reference to the drawings, detail the technical solutions provided by the embodiments of the present application.

[0018] Figure 1 Schematic diagram of the flow of a magnetic resonance method for evaluating the rehabilitation efficacy of children with cerebral palsy in the present invention, specifically including the following steps: S101: Obtain the structural image data and resting-state image data of the individual children patients to be evaluated.

[0019] Specifically, scanning is performed on a magnetic resonance of at least 3T to obtain high-quality structural image data of children, including T1 weighted imaging data (T1WI), diffusion tensor imaging (DTI), and a resting-state functional magnetic resonance sequence with a scanning time of at least more than 8 minutes. The T1WI slice thickness is ≤1mm. The diffusion imaging sequence evaluates the integrity of white matter, and the resting-state functional magnetic resonance sequence evaluates changes in brain function activity and connectivity. For the scanning parameters and some of the images obtained after scanning, see Figure 2 and Table 1, the list of T1 weighted imaging, diffusion tensor imaging, and resting-state functional image data and scanning parameters.

[0020] Table 1 List of Scanning Parameters

[0021] Specifically, the present invention follows the diagnostic suggestions of senior radiologists, and all data are discriminated by 2 doctors with 10 years of experience in brain MRI diagnosis. Among them, the example is a 6-year-old boy with a clinical diagnosis of spastic diplegia. Table 2 shows the clinical data of the MRI dataset of children with cerebral palsy in the present invention. The first column represents the serial number, the second column represents the age, the third column represents the gender, the fourth column represents the clinical diagnosis, the fifth column represents the gross motor score, and the sixth column represents the hand function score.

[0022] Table 2 Clinical Data of MRI Dataset of Children with Cerebral Palsy

[0023] S102: Preprocess the resting-state image data and the structural image data; and register the preprocessed resting-state image data and structural image data to the pediatric standard brain template respectively.

[0024] Optionally, preprocessing the resting-state image data specifically includes: converting the resting-state image data from DICOM format to NIFTI format; performing time slice correction on the NIFTI-format resting-state image data; performing head motion correction on the corrected resting-state image data; smoothing the aligned resting-state image data using a Gaussian smoothing kernel; removing artifacts related to head motion in the smoothed resting-state image data; and filtering out the physiological signal noise of the resting-state image data after removing artifacts using temporal filtering to obtain the preprocessed resting-state image data.

[0025] Specifically, the resting-state imaging data was converted from DICOM format to NIFTI format using dcm2niix software (version v1.0.20241211); the time slice correction was performed on the NIFTI-format resting-state imaging data using the nipy.algorithms.slicetiming.timefuncs function of fmriprep software, and the correction parameters were determined according to the individual resting-state imaging slicetiming and TR; the resting-state imaging data was smoothed using a Gaussian smoothing kernel of [6*6*6]mm in fmriprep software; the ICA-AROMA algorithm was used to remove the artifacts related to head movement in the smoothed resting-state functional magnetic resonance sequence data; time filtering was used to remove physiological signal noise to obtain the preprocessed resting-state imaging data; the preprocessed resting-state imaging data was aligned to the child brain boundary using the Riemannian manifold brain registration algorithm.

[0026] Optionally, the diffusion imaging data was preprocessed, specifically including: performing slice outlier detection on the diffusion imaging data to remove Gibbs-ringing artifacts; removing the noise of the diffusion imaging data; for the diffusion imaging data after noise removal, an undistorted non-diffusion weighted image was synthesized from the structural image and combined with the original non-diffusion weighted image, and then the diffusion imaging data was corrected; performing bias field correction on the corrected diffusion imaging data; using a self-supervised learning algorithm to filter the diffusion imaging data after bias field correction to obtain the preprocessed diffusion imaging data.

[0027] Specifically, the data format of the diffusion imaging sequence was converted from DICOM to NIFTI format; slice outlier detection was performed on the NIFTI-format diffusion imaging data, and mrdegibbs in the mrtrix3 toolbox was used to remove Gibbs-ringing artifacts; and dwidenoise was used to remove the noise of the diffusion imaging data; for the diffusion imaging data, the Synb0-DisCo algorithm was used to synthesize an undistorted non-diffusion weighted image from the structural image and combine it with the original b0 image, and then the diffusion imaging data was corrected using dwifslpreproc in mrtrix3 software to remove head movement, eddy current correction, and artifacts caused by susceptibility; performing bias field correction on the corrected diffusion imaging data; using the self-supervised learning algorithm patch2self to filter the diffusion imaging data after bias field correction to remove noise and obtain the preprocessed diffusion imaging data; The preprocessed child brain structural data (T1WI) and resting-state functional MRI data were registered to the child standard brain template using the Riemannian manifold brain registration algorithm. For the effect of registering the resting-state imaging data to the child standard brain template, see Figure 4 。

[0028] Optionally, preprocess the T1-weighted imaging data, including: resampling the T1-weighted imaging data to a space of a specified size; using the trained 3D Unet segmentation model to extract the brain tissue of the individual child patient to be evaluated from the space; and resampling the brain tissue to another space of a specified size.

[0029] Specifically, referring to Figure 3 , for the segmentation and resampling effect diagram of the T1-weighted imaging data, preprocessing the T1-weighted imaging data includes: resampling the T1-weighted imaging data to a space of 1mm×1mm×1m and 256×256×256 layers, and then using a 3D-Unet segmentation model trained based on pediatric structural images to extract pediatric brain tissue from the T1WI image, and resampling it to a space of 3mm×3mm×3mm.

[0030] Register the preprocessed diffusion imaging data and T1-weighted imaging data to the pediatric standard brain template using the Riemannian manifold brain registration algorithm. For the effect diagram, refer to Figure 5 and Figure 6 .

[0031] S103: Obtain and normalize multiple clinical indicators of the registered resting-state imaging data and structural imaging data in the pediatric standard brain template respectively; splice the normalized multiple clinical indicators into multi-index vector data, and input it into the pre-trained support vector machine model to output the clinical indicator differences; according to the clinical indicator differences, obtain the abnormal brain regions of the individual child patient to be evaluated.

[0032] Optionally, the clinical indicators include functional indicators and structural indicators; the specific steps of respectively obtaining multiple clinical indicators of the registered resting-state imaging data and diffusion imaging data in the pediatric standard brain template are as follows: based on the registered resting-state imaging data in the pediatric standard brain template, calculate the functional indicators, and the functional indicators include the whole-brain functional connectivity matrix, regional homogeneity and its average value, and low-frequency signal amplitude and its average value; based on the registered diffusion imaging data in the pediatric standard brain template, calculate the structural indicators, and the structural indicators include the whole-brain structural connectivity matrix, fractional anisotropy and its average value, and mean diffusivity and its average value.

[0033] Optionally, the whole-brain functional connectivity matrix is constructed by dividing the pediatric standard brain template into multiple regions of interest, calculating the functional connectivity strength between each region of interest, and based on the functional connectivity strength between the regions of interest; the regional homogeneity is obtained by calculating the coordination coefficient between each voxel in the pediatric standard brain template and its adjacent voxels; the low-frequency signal amplitude is obtained by converting the registered resting-state imaging data into a frequency-domain signal through Fourier transform and calculating the power spectral density of each voxel in the low-frequency band of the frequency-domain signal in the pediatric standard brain template.

[0034] Among them, the whole-brain functional connectivity matrix, regional homogeneity, and low-frequency signal amplitude are calculated based on the registered resting-state imaging data. The calculation process is shown in Figure 7 where the functional MRI connectivity matrix diagram is shown in Figure 8 .

[0035] Specifically, through the registered resting-state magnetic resonance sequence data, based on canonical correlation analysis, the resting-state magnetic resonance metrics (low-frequency signal amplitude and regional homogeneity) are calculated. After standard normalization respectively, the average values of regional homogeneity and low-frequency signal amplitude in brain regions are calculated based on a brain template. The fusion analysis results of the resting-state MRI metrics without normalization are shown in Table 3; Table 3 Fusion analysis results of resting-state MRI metrics without normalization

[0036] Through the preprocessed diffusion imaging data, based on canonical correlation analysis, the clinical metrics FA and MD of diffusion tensor imaging data are calculated. After standard normalization respectively, the average values of FA and MD in brain regions are calculated based on a brain template. See Table 4.

[0037] Table 4 Fusion analysis results of diffusion tensor sequence metrics without normalization

[0038] The average values of regional homogeneity, low-frequency signal amplitude, FA, and MD in brain regions of the resting-state metrics are combined with the whole-brain functional connectivity matrix and the whole-brain structural connectivity matrix into a multi-index vector, and a pre-trained support vector machine is used to identify abnormal brain regions.

[0039] Using the cerebral palsy dataset and the normal children dataset with the same above preprocessing process, the multi-index vector dataset established after calculating the resting-state metrics and DTI metrics is used to process the data with the trained support vector machine model. The multi-index vector of the fused multi-modal imaging metrics of the patient is used as the original input matrix, and the features with statistical significance in the algorithm output are selected as abnormal brain regions.

[0040] Specifically, the FA parameter map and MD parameter map are calculated from the preprocessed diffusion imaging data. See Figure 9 ; The fiber orientation density (FOD) is calculated by the CSD (spherical deconvolution) method. Based on the gray and white matter tissues extracted by the 3D-Unet segmentation model, the whole-brain fiber bundles are deterministically traced and the whole-brain connectivity matrix is calculated. The calculation process of the structural MRI metrics is shown in Figure 10 where the whole-brain structural connectivity matrix diagram is shown in Figure 11 .

[0041] S104: Determine a multimodal fusion index analysis report on the rehabilitation efficacy of childhood cerebral palsy based on the abnormal brain regions of the individual child patient to be evaluated.

[0042] Optionally, determining a multimodal fusion index analysis report on the rehabilitation efficacy of childhood cerebral palsy specifically includes: Extract the multimodal fusion indices of the abnormal brain regions of the individual child patient to be evaluated; the multimodal fusion indices include functional connectivity of abnormal brain regions, average regional homogeneity of abnormal brain regions, average low-frequency signal amplitude of abnormal brain regions, structural connectivity of abnormal brain regions, average fractional anisotropy of abnormal brain regions, and average diffusivity of abnormal brain regions; according to the multimodal fusion indices, obtain a multimodal fusion index analysis report through automated software, see Figure 12 , which is a schematic diagram of the multimodal fusion index analysis report.

[0043] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present invention.

Claims

1. A magnetic resonance method for evaluating the efficacy of rehabilitation of children with cerebral palsy, characterized in that: include: Obtain structural imaging data and resting-state imaging data of individual pediatric patients to be evaluated; Preprocessing the resting state image data and the structural image data; The preprocessed resting-state image data and structural image data were registered to the children's standard brain template respectively; Respectively obtain multiple clinical indicators of resting-state image data and structural image data registered in the standard brain template for children and normalize them; splice the normalized multiple clinical indicators into multi-index vector data, input them into the pre-trained support vector machine model, and output the clinical indicator differences; obtain the abnormal brain areas of the individual child patients to be evaluated based on the clinical indicator differences; Based on the abnormal brain areas of individual child patients to be evaluated, a multimodal fusion indicator analysis report on the efficacy of cerebral palsy rehabilitation in children is determined.

2. A magnetic resonance method for evaluating the therapeutic effect of cerebral palsy rehabilitation in children as claimed in claim 1, characterized in that: Preprocess the static image data, including: Convert resting-state imaging data from DICOM format to NIFTI format; Perform time layer correction on the NIFTI format resting state image data; and perform head motion correction on the corrected resting state image data; The aligned resting-state image data were smoothed using a Gaussian smoothing kernel; Remove artifacts related to head motion from smoothed resting-state image data; Temporal filtering is used to filter out the physiological signal noise of the resting-state image data after artifact removal to obtain the preprocessed resting-state image data.

3. A magnetic resonance method for evaluating the therapeutic effect of cerebral palsy rehabilitation in children as claimed in claim 1, characterized in that: The structural imaging data includes: diffusion imaging data and T1-weighted imaging data; Preprocess the diffusion imaging data, including: Convert resting-state imaging data from DICOM format to NIFTI format; Perform slice outlier detection on the diffusion imaging data, and remove Gibbs-ringing artifacts and noise of the diffusion imaging data to obtain denoised diffusion imaging data; synthesizing an undistorted non-diffusion weighted image from the denoised diffusion imaging data and combining it with the original non-diffusion weighted image, and then correcting the diffusion imaging data; performing bias field correction on the corrected diffusion imaging data; A self-supervised learning algorithm is used to filter the diffusion imaging data after bias field correction to obtain preprocessed diffusion imaging data; Preprocess the TI-weighted imaging data, including: Resample T1-weighted imaging data to a space of specified size; Using the trained 3DUnet segmentation model, the brain tissue of the individual child patient to be evaluated is extracted from space; The brain tissue is resampled again to another space of specified size.

4. A magnetic resonance method for evaluating the therapeutic effect of cerebral palsy rehabilitation in children as claimed in claim 1, characterized in that: The clinical indicators include functional indicators and structural indicators; the multiple clinical indicators of resting-state image data and structural image data registered in the children's standard brain template are obtained respectively, specifically including: Functional indices were calculated based on the resting-state imaging data registered in the children's standard brain template. The functional indices included the whole-brain functional connectivity matrix, regional homogeneity and its average value, and low-frequency signal amplitude and its average value. Structural indices were calculated based on the registered structural imaging data in the children's standard brain template. The structural indices included the whole-brain structural connectivity matrix, fractional anisotropy and its average value, and mean diffusivity and its average value.

5. A magnetic resonance method for evaluating the therapeutic effect of rehabilitation of children with cerebral palsy as claimed in claim 4, characterized in that: The whole-brain functional connectivity matrix is ​​constructed by dividing the standard brain template of children into multiple regions of interest, calculating the functional connectivity strength between each region of interest, and based on the functional connectivity strength between the regions of interest; The regional homogeneity is obtained by calculating the coordination coefficient between each voxel and its neighboring voxels in the children's standard brain template; The amplitude of the low-frequency signal is obtained by converting the registered resting-state image data into a frequency domain signal through Fourier transformation, and calculating the power spectral density of each voxel in the low-frequency band of the frequency domain signal in the children's standard brain template.

6. A magnetic resonance method for evaluating the therapeutic effect of cerebral palsy rehabilitation in children as claimed in claim 4, characterized in that: The whole-brain structural connection matrix is ​​constructed by reconstructing the white matter fiber bundles of the registered structural image data through fiber tracking technology, calculating the connection strength of each fiber bundle, and based on the connection strength of each fiber bundle.

7. A magnetic resonance method for evaluating the therapeutic effect of cerebral palsy rehabilitation in children as claimed in claim 1, characterized in that: The multimodal fusion indicator analysis report for determining the efficacy of rehabilitation for children with cerebral palsy specifically includes: Extracting multimodal fusion indicators of abnormal brain regions of individual children to be evaluated; the multimodal fusion indicators include abnormal brain region functional connectivity, abnormal brain region average regional homogeneity, abnormal brain region average low-frequency signal amplitude, abnormal brain region structural connectivity, abnormal brain region average anisotropy fraction, and abnormal brain region average diffusivity; According to the multimodal fusion indicators, a multimodal fusion indicator analysis report is obtained through automated software.

Images