70 results about "T1 weighted" patented technology

The invention discloses a nerve regulation system for cognitive impairment, and the system comprises a cognitive behavior evaluation module which is used for recognizing a cognitive defect field which needs to be improved by a target object; the image data acquisition module is used for acquiring and preprocessing task state nuclear magnetic data and T1 weighted structure images of a target object; the brain region activation analysis module is used for determining a neural regulation potential target spot according to the preprocessed task state nuclear magnetic data; the brain network analysis module is used for identifying key nodes related to cognitive defects; the stimulation target screening module is used for determining an individualized stimulation target of the target object according to the nerve regulation potential target and the key node of the target object; the time domain interference electrical stimulation simulation and implementation module is used for performing electrical stimulation on the individualized stimulation target spot; and the dynamic feedback adjustment module is used for dynamically adjusting the stimulation parameters and enhancing the personalized regulation and control effect. By adopting the method and the device, the specific brain region directly related to the cognitive impairment can be determined, so that the targeting and the accuracy of treatment are enhanced.

The invention discloses a temporal lobe epilepsy resection operation curative effect evaluation system based on brain image fusion features. According to the system, pre-operation and post-operation multi-mode brain image data including T1 weighted images, diffusion tensor imaging (DTI) and functional magnetic resonance imaging (fMRI) are collected and preprocessed, and a post-operation recovery model is generated through a recurrent neural network (RNN), a long-short-term memory network (LSTM) and other multi-time-sequence analysis methods. And the personalized curative effect evaluation module is combined with clinical characteristic data of the patient. And the report generation and visualization module generates an easy-to-understand report from the evaluation data through a visualization technology for medical teams to check and make decisions. Through multi-modal data fusion and an advanced algorithm model, the accuracy and reliability of curative effect evaluation are improved, the brain structure and function changes of the patient are comprehensively reflected, the clinical application prospect is wide, and the postoperative life quality of the patient is remarkably improved.

The invention discloses an epilepsy region positioning method and system based on brain power supply imaging and a dynamic brain network, and the method comprises the steps: obtaining T1 weighted magnetic resonance imaging data of a user, and constructing an individual three-dimensional head model through a boundary element method; acquiring scalp electroencephalogram data of a user, and preprocessing the scalp electroencephalogram data; based on an individual three-dimensional head model, performing inverse problem solving on the preprocessed scalp electroencephalogram data by using a standardized low-resolution brain power source imaging algorithm to obtain source current density signals of 68 brain regions; decomposing into six frequency bands, calculating the power spectrum density of each brain region and carrying out normalization processing, and screening effective frequency bands; based on the source current density signals of the 68 brain regions of the effective frequency band, information flow directions and intensities of different brain regions are calculated by adopting a directional transfer function method, a directional transfer function matrix of the effective frequency band is formed, and a directed brain network is constructed; and calculating a graph theory index and / or an epilepsy index of each brain region, carrying out maximum value normalization analysis, and determining an epilepsy region positioning result.

The invention relates to the technical field of medical image auxiliary diagnosis, in particular to a nuclear magnetic resonance image auxiliary diagnosis method and system based on multi-modal fusion. The method comprises the following steps: step 1, synchronously acquiring a three-dimensional T1 weighted structure image, a T2 weighted fluid attenuation inversion recovery image and diffusion weighted imaging data of a subject, carrying out spatial registration by taking the T1 weighted image as a reference, and executing skull stripping and gray scale standardization; 2, individualized brain region segmentation is carried out based on a brain anatomical map, the lesion sensitivity weight of each modal is calculated for each segmented brain region, and the weight is obtained by quantifying the following parameters; step 3, extracting multi-modal image features in each brain region; and 4, inputting the fusion features of the whole brain region into a multi-task classifier. The standardization and alignment of the multi-mode MRI image in the space and gray level are realized, and the problems of space mismatch and feature interference among different modes are effectively solved.

The invention discloses a method for generating a magnetic resonance gadolinium contrast enhancement image based on contrast learning and an SDACL model, and further comprises the step of carrying out patch-based multi-layer contrast learning on the SDACL model and adjusting a designed loss function, the SDACL model comprises a domain X representing a multi-mode MRI channel superposition image, a domain Y representing a T1CE image, generators G and F and a discriminator network; according to the method, T1, T2 and Flair magnetic resonance plain scanning sequence images of a glioma patient are utilized to generate a T1 weighted enhanced sequence image, an attention dilation generator is adopted, a receptive field is expanded by utilizing an expansion residual block, and local features and global dependency of an input image are combined by means of a convolutional attention module (CBAM). According to the method, the feature representation and context learning of the multi-sequence MRI are enhanced, a comprehensive loss function is further designed, the comprehensive loss function comprises patch-level contrast loss and structural similarity loss, noise can be effectively restrained, and the structural consistency of a generated image and a real image is guaranteed.

The invention discloses a method, device and equipment for generating a head model, a storage medium and a program product. The method is characterized by comprising the following steps of: receiving and verifying original three-dimensional T1 weighted image data and an original three-dimensional grid model; performing tissue segmentation based on the verified three-dimensional T1 weighted image data and the original three-dimensional T1 weighted image data to obtain a cerebral grey matter mask, a cerebral white matter mask, a scalp mask and a skull mask; performing topological structure repair on the brain grey matter mask and the brain white matter mask to obtain a brain mask; respectively converting the scalp mask, the skull mask and the brain mask into a scalp mesh model, a skull mesh model and a brain mesh model; calculating and applying a spatial transformation matrix from the verified three-dimensional grid model to the scalp grid model, and performing spatial registration and fusion on the scalp grid model, the skull grid model and the brain grid model to obtain a head model; the method has the advantages that the efficiency, precision and robustness of head model generation are effectively improved, and the method has good cross-platform deployment capability.

The invention relates to the technical field of image processing, in particular to a diabetic cognitive function obstruction detection method based on an MRI (Magnetic Resonance Imaging) image, which comprises the following steps: acquiring a T1 weighted image and a T2 weighted image; obtaining brain pixel points according to the positions of the pixel points in the two weighted images; according to the gray expression of the brain pixel points in the two weighted images, the possibility that the brain pixel points are brain quality pixel points is obtained, and the tissue expression degree of each pixel point in the two weighted images is further obtained; according to the pixel points of the brain pixel points in the two weighted images, combining the tissue expression degree of each pixel point in the two weighted images to obtain the edge features of the brain pixel points; and de-noising the two weighted images according to the edge features of the brain pixel points and the possibility that the brain pixel points are brain quality pixel points. According to the method, the gray level expressions of the pixel points in the two weighted images are analyzed, the unique features of the noise are combined, and smoothing parameters are given to each pixel point, so that the loss of detail features is avoided.

The invention relates to a heterogeneous human head model construction method and system oriented to multi-physics field simulation, based on real human head T1 weighted magnetic resonance image data, key tissues such as white matter, gray matter and cerebrospinal fluid are accurately segmented, and a three-dimensional geometric model used for generating carrying tissue identification information is researched and developed; converting the model data into a file format conforming to the NASTRAN standard by utilizing an independently developed format conversion program, and completely retaining organization classification information; finally, different tissue areas are automatically recognized through physical field simulation software, corresponding frequency dependent material attributes are distributed, and a finite element model capable of being directly used for simulation is formed. According to the method, seamless integration of the high-precision heterogeneous human head model and the simulation platform is achieved, tedious manual repairing and assignment operation is not needed, the method has the advantages of being high in precision, high in efficiency and good in compatibility, and the authenticity and reliability of biomedical simulation such as transcranial magnetic stimulation and radio frequency coil design can be remarkably improved.

A method for magnetic resonance imaging (MRI) may include obtaining a plurality of first magnetic resonance (MR) data sets related to a region of interest (ROI) of a subject. The plurality of first MR data sets may be collected based on two or more different values of a scan parameter. The method may also include determining a plurality of second MR data sets based on the plurality of first MR data sets. Each of the plurality of second MR data sets may correspond to at least two of the plurality of first MR data sets. The method may also include generate, based on the plurality of second MR data sets, a plurality of T1 weighted images of the ROI each of which corresponds to a target time point.

The invention relates to a nuclear magnetic resonance image classification diagnosis method for adult diffuse glioma. The method comprises the following specific steps: step 1, respectively acquiring T2-fluid attenuation inversion recovery T2-FLAIR and contrast enhancement T1 weighted imaging CE-T1WI magnetic resonance images of glioma labeled by a doctor according to histological and genomics information; step 2, carrying out data preprocessing operation on the T2-FLAIR image data and the CE-T1WI image data; step 3, each pair of T2-FLAIR and CE-T1WI images after preprocessing is sent to a decision tree diagnosis model, and an Attention-DenseNet model is used for learning and extracting the characteristics of the T2-FLAIR and CE-T1WI images; 4, obtaining a prediction result based on a plurality of independently trained Attention-DenseNet models, fusing the outputs of the plurality of models by adopting a voting mechanism or a weighted average strategy, and generating a final classification decision; and 5, sending the obtained final classification to a platform, and making final diagnosis based on a classification result.

A diffusion model-based diffusion magnetic resonance image super-resolution reconstruction method comprises the following steps of 1) performing downsampling and normalization processing on high-resolution diffusion magnetic resonance data, and constructing a low-resolution-high-resolution image pair and a corresponding T1 weighted image as training samples; (2) in the forward process of the diffusion model, Gaussian noise is gradually added and high-resolution and low-resolution image residuals are fused, so that a noise adding result approaches low-resolution image distribution; 3) constructing a network architecture based on a cross attention mechanism, and learning a mapping relation from dispersion data to high-resolution data under different noise levels by combining multi-scale feature fusion and structural prior information of a T1 weighted image; and 4) performing a reverse denoising process by using the trained network to realize super-resolution reconstruction of the dispersion magnetic resonance image. According to the method, the super-resolution reconstruction quality of the diffusion magnetic resonance image is remarkably improved by fusing the multi-modal prior information and the progressive up-sampling strategy.

The invention discloses an MRI (Magnetic Resonance Imaging) image registration generation method based on a Cycle consistency framework, which comprises the following steps of: inputting a T1 weighted image into an image generation network to obtain a T2 weighted image, inputting the T2 weighted image generated by the image generation network and an original T2 weighted image into an image registration network to obtain an aligned T2 weighted image, and outputting the aligned T2 weighted image to a Cycle consistency framework; and inputting the aligned T2 weighted image into an image generation network to obtain a T1 weighted image, and inputting the T1 weighted image generated by the image generation network and the original T1 weighted image into an image registration network to obtain an aligned T1 weighted image. According to the invention, the defects in the prior art can be improved, and the generation quality of the MRI image is improved.

The invention discloses a brain abnormal development trajectory identification method and system, and belongs to the technical field of magnetic resonance image analysis, and the method comprises the following steps: obtaining T1 weighted magnetic resonance images of a plurality of healthy individuals, and carrying out the offset correction of the structure index of each brain region in the images, a generalized additive position-scale-shape model of the healthy crowd is constructed; generating a norm development trajectory curve and a normal change range thereof; obtaining a T1 weighted magnetic resonance image of a to-be-evaluated individual, and selecting a target brain region of the to-be-evaluated individual; obtaining various offset corrected structure indexes of the target brain region of the individual to be evaluated; obtaining a mean value and a standard deviation of each structure index at the position of the brain region corresponding to the same-age healthy population of the individual to be evaluated; and generating a recognition report of the abnormal brain development trajectory of the to-be-evaluated individual. The problem that it is difficult to accurately, clearly and visually reflect the abnormal recognition condition of the brain trajectory of the individual to be evaluated and the position of the corresponding abnormal brain region is solved.

A method for magnetic resonance imaging (MRI) may include obtaining a plurality of first magnetic resonance (MR) data sets related to a region of interest (ROI) of a subject. The plurality of first MR data sets may be collected based on two or more different values of a scan parameter. The method may also include determining a plurality of second MR data sets based on the plurality of first MR data sets. Each of the plurality of second MR data sets may correspond to at least two of the plurality of first MR data sets. The method may also include generate, based on the plurality of second MR data sets, a plurality of T1 weighted images of the ROI each of which corresponds to a target time point.

The invention provides a structural magnetic resonance imaging analysis method and system and a storage medium, and the method comprises the steps: obtaining sample data which is a T1 weighted structural image subjected to structural magnetic resonance imaging, determining a map of morphological indexes of a region of interest in an original brain network space according to the sample data, re-sampling the map to a preset standard brain network space template, carrying out Gaussian kernel smoothing processing to obtain standard brain network space template data, mapping the standard brain network space template data to a pre-constructed brain map template, determining feature input of a machine learning model according to the mapped brain map template, and carrying out the feature input of the machine learning model; and training the classification performance of the machine learning model based on the feature input to obtain a classification model. The effectiveness and reliability of training can be improved, so that the classification accuracy of the classification model obtained based on training can be improved.

The invention discloses an automatic hepatic fibrosis staging method and system based on deep learning, and the method comprises the following steps: S1, obtaining T1WI and T2WI original image data of a patient, carrying out the data preprocessing and liver segmentation of the original image data, and obtaining a T1 weighted image data set and a T2 weighted image data set; s2, a DMF-Vheat hepatic fibrosis staging model is constructed, the staging model adopts a double-sequence classification architecture and comprises a T1 branch network, a T2 branch network and a fusion network, and a heat conduction operator layer and a mixed attention module are adopted to extract and process complementary information of T1WI and T2WI sequences; s3, putting the T1 weighted image data set and the T2 weighted image data set into the staging model for training, and optimizing the staging model; and S4, outputting a hepatic fibrosis staging result through the staging model. According to the method, complementary information of T1WI and T2WI is fully utilized through the multi-mode deep learning network, and the accuracy and practicability of hepatic fibrosis staging are remarkably improved.

The invention relates to a structural network-genetic map biological network model for predicting ischemic stroke and a construction method thereof, and the method comprises the steps: extracting and calculating seven multi-scale morphological features and pairwise Pearson correlation coefficients among the features from T1 weighted imaging data and diffusion tensor imaging data; constructing a 308 * 308 morphological similarity network matrix and a brain network module for identifying ischemic stroke neural dysfunction; 1782 sampling points are extracted from the Airy human brain map, and each sampling point comprises expression data of 10185 genes; the method comprises the following steps: mapping space coordinates of AHBA sampling points to a cortex package of a Desikan-Killiany map, carrying out normalization processing to output 308 * 10185 brain region gene-by-gene expression matrixes, and constructing a structural network-gene map biological network model for predicting ischemic stroke by adopting a partial least square regression method and a bootstrap method. Compared with the prior art, the model determines the specific molecular mechanism related to the phenotypic structure change of ischemic stroke injury, and the stroke occurrence probability is predicted according to the specific molecular mechanism.

The invention discloses a neurodegenerative change assessment method and device based on nerve melanin MRI, and relates to the technical field of medical imaging and neuroscience assessment, and the method comprises the steps: obtaining MRI image data of a to-be-assessed individual; the MRI image data comprises multi-sequence MRI image data including three-dimensional T1 weighted imaging, two-dimensional T2 weighted imaging and magnetization transfer imaging; preprocessing the MRI image data to obtain preprocessed MRI image data; extracting a region of interest from the preprocessed MRI image data to obtain the region of interest; performing MRI signal feature analysis on the region of interest to obtain MRI signal features; and inputting the MRI signal features into a neurodegenerative change evaluation model, and outputting a neurodegenerative change evaluation result of the individual to be evaluated. According to the application, the non-invasive and accurate evaluation of the neurodegenerative change can be realized by using the unique signal characteristics of the nerve melanin in MRI imaging.

The invention relates to a method and a device for automatically identifying benign and malignant kidney cystic lesions based on magnetic resonance images. The method comprises the following steps: S1, preprocessing and standardizing a T2 weighted image, a diffusion weighted image, an apparent diffusion coefficient image, a T1 weighted image, a skin medullary phase image, a parenchyma phase image and an excretion phase image; s2, respectively training automatic segmentation models corresponding to different images in a targeted manner, and predicting a focus by using the automatic segmentation models; s3, extracting morphological features, first-order features and textural features of the lesions from all the lesions, wherein the morphological features, the first-order features and the textural features comprise features of capsule walls, partitions and nodules of the lesions; screening the extracted features, and constructing a classification model by using the features with good robustness; and S4, preprocessing and standardizing the image of the current patient, respectively inputting the image into each corresponding automatic segmentation model, and operating the segmentation model and the classification model to realize benign and malignant recognition based on image recognition. According to the invention, integrated and automatic benign and malignant accurate diagnosis of kidney cystic lesions is realized.

The invention discloses a novel nuclear magnetic resonance contrast agent Gd-DOTA-PP for targeting citrullinated histone to achieve neutrophil extracellular trap net imaging and a preparation method and application thereof.The method comprises the steps that citrullinated histone targeting polypeptide PP and ligand molecules DOTA-NHS are combined, and a compound DOTA-PP is obtained; and then gadolinium (Gd) ions are introduced into the DOTA-PP through a coordination reaction, and the novel contrast agent Gd-DOTA-PP is prepared. The contrast agent can be selectively combined with an important marker, namely citrullinated histone, of neutrophil extracellular trapping nets (NETs), and T1 weighted imaging signals of a target area are remarkably enhanced. The preparation method is simple and convenient to operate, the preparation process is highly controllable, and the obtained contrast agent has excellent biocompatibility and targeting characteristic. The accurate targeting ability of the probe enables the influence signal of the probe in NETs enrichment areas such as tumor early metastasis, tumor infiltration lymph nodes and the like to be obviously enhanced, a powerful tool is provided for tumor diagnosis, especially accurate diagnosis of the early metastasis, and the probe has important clinical application value.

The application discloses a brain structure analysis method and system based on gray matter and white matter imageomics hierarchical coupling, and the method comprises the following steps: acquiring and preprocessing original three-dimensional T1 weighted structure magnetic resonance images, generating gray matter volume smoothing images and white matter volume smoothing images; extracting brain region level imageomics original features from the gray matter volume smoothing images and the white matter volume smoothing images respectively, obtaining a gray matter brain region level original feature set and a white matter brain region level original feature set; based on repeated scanning data, performing stability evaluation screening, same brain region cross-gray matter and white matter tissue candidate screening and redundancy removal processing on the gray matter brain region level original features and the white matter brain region level original features, and generating a brain region level reserved feature list; in each brain region, performing feature layering on the brain region level reserved feature list, and generating a first-order statistical feature layer and a texture feature layer; and constructing a gray matter and white matter coupling index based on the first-order statistical feature layer and the texture feature layer, so as to form a brain region level gray matter and white matter hierarchical coupling matrix.

The invention discloses ultra-small magnetic iron oxide nanoparticles taking small molecules as a stabilizer as well as a preparation method and application of the ultra-small magnetic iron oxide nanoparticles, and belongs to the technical field of nano materials. The magnetic iron oxide nanoparticles provided by the invention comprise magnetic iron oxide particles and hydrophilic small molecules connected to the surfaces of the magnetic iron oxide particles, the average molecular weight of the hydrophilic small molecules is less than 1000 Daltons. Hydrophilic small molecules are connected to the surfaces of the magnetic iron oxide particles, good stabilizing and dispersing effects can be achieved, and compared with existing magnetic iron oxide nanoparticles with macromolecules as a stabilizer, the magnetic iron oxide nanoparticles are lower in viscosity, higher in biocompatibility, higher in longitudinal relaxation r1 value and better in stability. The clinical practicability and the development prospect are better, and the preparation of a safe and reliable MRI contrast agent, especially a T1 weighted contrast agent, is facilitated.

Due to various symptoms of Parkinson's disease, accurate diagnosis of Parkinson's disease is still challenging. Although multi-modal magnetic resonance imaging (MRI) can provide complementary information, an existing method has the defects in fusion strategies and modeling modes: on one hand, naive multi-modal fusion is difficult to describe a complex cross-modal relationship; on the other hand, the modeling based on the local patch ignores the inherent anatomical connection of the brain. The invention provides a Parkinson's disease diagnosis model construction method based on a cross-modal diagram attention network. According to the method, robust three-dimensional feature representation is learned under the condition of limited medical image data through self-supervised comparison pre-training; a bidirectional cross-modal attention module is introduced, and interaction between T1 weighted MRI and quantitative magnetic susceptibility mapping features is enhanced; and a brain region graph based on anatomical prior is constructed, and a graph attention network is utilized to model a brain interval spatial dependency relationship. Experiments on a hospital data set show that compared with a representative method, the model constructed by the method has better performance.

Calculating a brain age predicts progression from cognitively normal status to cognitive impairment. Software acquires diffusion magnetic resonance images (dMRI) and T1 weighted magnetic resonance images (T1w MRI) of the brain and calculate a pair of diffusion tensor imaging (DTI) scalar maps. The scalar maps may be a fractional anisotropy (FA) scalar map and / or a mean diffusivity (MD) scalar map. Concatenating the FA scalar map and the MD scalar map form a frame of scalar map data. The method includes a rigid transformation from scalar map data to the T1w images to store a rigid transformed image; applying an affine transformation from the rigid transformed image to a brain template; storing a rigid plus affine transformed image; and applying a non-rigid transformation to warp the rigid plus affine transformed image to the selected brain template to form a non-rigid white matter age input to a brain age calculation model.

The invention relates to the technical field of machine learning, in particular to a knee osteoarthritis assessment method based on machine learning, which comprises the following steps: acquiring knee joint T1 and T2 weighted images of a patient with knee osteoarthritis; manually segmenting the knee joint cartilage in the T1 weighted image and measuring the volume of the knee joint cartilage; cutting the T2 weighted image to obtain a meniscus image, manually segmenting the meniscus image to obtain annotation data, and constructing an image data set; training a meniscus automatic segmentation model, and segmenting the meniscus in the meniscus image; carrying out three-classification on meniscus pixels of the middle five layers of the meniscus image, and calculating a meniscus space specificity signal index; the performance of the meniscus space specific signal index in diagnosis of knee osteoarthritis is systematically evaluated. According to the knee joint meniscus damage diagnosis method, the knee joint T1 and T2 weighted images of a patient with knee osteoarthritis are collected, the meniscus automatic segmentation model and a meniscus space specificity signal index calculation method are applied, the damage condition of the knee joint meniscus is accurately evaluated, and knee osteoarthritis diagnosis is achieved. The process can provide a reliable diagnosis basis for orthopedists, so that clinical diagnosis of knee osteoarthritis is effectively assisted.

The invention discloses a depression prognosis prediction model construction method, system and device, a medium and application, and belongs to the field of depression prognosis prediction.The construction method comprises the following steps that mGLUR5 receptor availability PET images and T1 weighted MRI images of different patients are obtained; preprocessing the image; and constructing a model. The construction method is based on an advanced molecular imaging technology, the availability change of mGluR5 is measured, and the constructed depression prognosis prediction model has great innovation and application values. The built depression prognosis prediction model and the system, equipment and medium thereof have excellent measurement stability and reliability. The invention further discloses application of the mGluR5 as the biomarker in construction of a depression prognosis prediction model, and by accurately evaluating the availability of the mGluR5, a doctor can adjust a treatment scheme and estimate a prognosis effect according to the specific biomarker of each patient.

The invention discloses an infant white matter injury cognitive prognosis evaluation method, system and device, and relates to the technical field of medical treatment, the method comprises the following steps: obtaining T1 weighted imaging and diffusion tensor images of the brain of a to-be-evaluated infant; delineating a focal white matter damage area on the T1 weighted image, and generating a binarized individual focus mask; preprocessing the diffusion tensor image to generate a partial anisotropic graph; based on the individual focus mask and the diffusion tensor image, performing parallel calculation on focus volume and white matter structure loss-of-connection scores involving fiber bundles; inputting the lesion volume involving the fiber bundle and the white matter structure loss of connection score into a pre-trained decision tree model, and outputting an assessment result of the infant cognitive development delay risk according to a preset judgment rule; according to the method, a new quantitative index is provided for PWML dominant lesions and can be derived to conventional MRI images for analysis, DTI data does not need to be collected, and a new solution is provided for early evaluation of prognosis.

The application provides a brain cortex thickness morphological analysis method and system of MR images, and relates to the technical field of medical image data processing.The method comprises the following steps: obtaining a head three-dimensional high-resolution magnetic resonance T1 weighted image sequence of a patient with intractable epilepsy to be evaluated; performing bias field correction and brain tissue stripping on the image sequence to construct an initial brain parenchyma three-dimensional voxel model containing individual lateral ventricle expansion morphological characteristics; based on the initial brain parenchyma three-dimensional voxel model, extracting an initial isosurface of the interface between gray matter and white matter; calculating a local curvature tensor for a deep brain sulcus region in the initial isosurface; performing topological defect repair and mesh smoothing on the initial isosurface according to the local curvature tensor to obtain an individualized cortex surface mesh model eliminating ventricle expansion deformation artifacts.The application realizes accurate registration of image space and surgical physical space.

The invention discloses an image generation method and system based on glioma magnetic resonance imaging, and relates to the technical field of image processing. Comprising the following steps: acquiring a brain magnetic resonance image to be processed of glioma; performing multi-time step noise adding operation on the brain magnetic resonance image to be processed to obtain an intermediate variable image; performing feature compression on the intermediate variable image to obtain image features; performing encoding operation on the image features to obtain potential representation; performing image reconstruction on the potential representation, and performing cross-modal fusion to obtain a transformed potential representation; feature extraction is carried out from three dimensions of depth, height and width, and feature extraction results are fused to obtain a T1 weighted contrast enhanced image; according to the method, the T1 weighted contrast enhanced image is directly synthesized from the brain magnetic resonance image, and the structural fidelity and stability of the synthesized image are improved.

The application provides an undersampling magnetic resonance image reconstruction method based on a reference image and data correction, comprising the following steps: acquiring a full sampling T1 weighted image as a reference image I ref ; acquiring a full sampling T2 weighted image I T2 , and converting it into undersampling k-space data y; initializing an image sequence I s and initializing a mutual information value sequence MI s ; using a convolutional neural network based on a residual module, establishing an undersampling magnetic resonance image reconstruction model based on the reference image I ref and the undersampling k-space data y; training the undersampling magnetic resonance image reconstruction model; recording the reconstructed image in the reconstruction process and the mutual information value thereof with the reference image; selecting the best reconstructed image based on the mutual information value; and performing iterative k-space data correction on the best reconstructed image to obtain a final reconstructed image. The application can improve the accuracy of the reconstructed image.