A method, system, terminal, and storage medium for constructing individual morphological connectivity graphs

Abstract

This invention relates to the field of medical image processing technology, and discloses a method, system, terminal, and storage medium for constructing individual morphological connectivity maps. The method includes: preprocessing a three-dimensional T1-weighted structural magnetic resonance image of an individual to obtain a gray matter volume map in standard space; extracting multiple radiomics features from the gray matter volume map to generate multiple initial feature maps; standardizing and smoothing the multiple initial feature maps to obtain multiple target feature maps, and constructing a multidimensional feature vector for each voxel; selecting seed points and extracting their feature vectors; calculating the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the seed point feature vector on a voxel-by-voxel basis; and generating a morphological connectivity map based on the correlation coefficient. This invention achieves high spatial resolution morphological connectivity maps at the individual level without the need for predefined brain regions, overcoming the limitations of traditional group-level and brain region-dependent methods.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method, system, terminal, and computer-readable storage medium for constructing individual morphological connectivity maps. Background Technology

[0002] Morphological brain networks are important tools for studying the covariant relationships of brain structures. They characterize the organizational patterns of brain structures by calculating the statistical dependencies of morphological features (such as gray matter volume and cortical thickness) across different brain regions. Traditional methods primarily construct morphological covariance networks at the population level, failing to reflect individual differences. While methods for constructing individual-level morphological brain networks have emerged in recent years, these methods still have the following limitations: 1. Reliance on predefined brain regions: Existing methods (including low-order and high-order methods) all require pre-dividing brain regions based on brain atlases. This makes it impossible to directly and flexibly study the correlation between specific regions of interest to the user and other parts of the whole brain. At the same time, different brain region division schemes can significantly affect the final network results, lacking a unified standard.

[0003] 2. Loss of voxel-level information: Brain region-based methods require aggregating information from a large number of voxels within the brain region into a representative value (such as mean, probability distribution, etc.). This aggregation process loses detailed information within the brain region, making the constructed network unable to reflect fine-grained features at the voxel level, potentially masking important biological information.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a method, system, terminal, and computer-readable storage medium for constructing individual morphological connectivity maps, aiming to solve the problems of relying on predefined brain regions and losing voxel-level information when constructing individual morphological connectivity maps in the prior art.

[0006] To achieve the above objectives, the present invention provides a method for constructing an individual morphological connectivity graph, the method comprising the following steps: Preprocessing of the individual's three-dimensional T1-weighted structured magnetic resonance image yields a gray matter volume map in standard space; Multiple radiomics features are extracted from the gray matter volume map, and a feature map is generated for each radiomics feature to obtain multiple initial feature maps. The multiple initial feature maps are standardized and smoothed to obtain multiple target feature maps, and a multidimensional feature vector for each voxel is constructed. Select brain regions of interest as seed points and extract the seed point feature vectors of the seed points; The correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map.

[0007] Furthermore, the preprocessing of the individual's three-dimensional T1-weighted structural magnetic resonance image to obtain a gray matter volume map in standard space includes: Gray matter images were segmented from three-dimensional T1-weighted structural magnetic resonance images of individuals using an adaptive maximum a posteriori technique. The gray matter image is normalized to the standard Montreal Neuroscience Institute space to obtain the target gray matter image; The target gray matter map is modulated using the Jacobian determinant obtained during the standardization process to obtain a gray matter volume map.

[0008] Furthermore, the step of extracting multiple radiomics features from the gray matter volume map, generating a feature map for each radiomics feature, and obtaining multiple initial feature maps includes: Generate a binarized gray matter mask based on a priori brain tissue probability map; Using a radiomics feature extraction tool, the value of each radiomics feature is calculated voxel-by-voxel from the gray matter volume map within the area defined by the gray matter mask; Based on the value of each radiomics feature, an image covering the whole brain gray matter region is generated for each radiomics feature, resulting in multiple initial feature maps.

[0009] Furthermore, the standardization and smoothing of the multiple initial feature maps to obtain multiple target feature maps includes: For each of the multiple initial feature maps, a standardization process is performed. Using a spatial smoothing filter with a preset kernel function, each of the standardized initial feature maps is convolved to obtain multiple target feature maps.

[0010] Furthermore, the construction of the multidimensional feature vector for each voxel includes: Spatial registration is performed on all the target feature maps; For each voxel within the gray mask, extract the value of the voxel at the corresponding position in the target feature map after all spatial registrations; The values ​​are combined in order to form a multidimensional vector, resulting in a multidimensional feature vector.

[0011] Furthermore, the selection of brain regions of interest as seed points includes: Determine the center coordinates of the region of interest; Using the center coordinates as the center, a spherical region with a preset radius is generated as a seed point.

[0012] Further, the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map, including: Based on the seed point feature vector and the multidimensional feature vector of each voxel in the gray mask, the Pearson correlation coefficient is calculated on a voxel-by-voxel basis. Based on the Pearson correlation coefficients of all voxels, a raw whole-brain correlation map is generated; The original whole-brain correlation map was subjected to Fisher Z-transform to obtain a Z-value map, which was defined as a morphological connectivity map.

[0013] Furthermore, to achieve the above objectives, the present invention also provides an individual morphological connectivity graph construction system, which is used to implement the individual morphological connectivity graph construction method described above, wherein the individual morphological connectivity graph construction system includes: The volumetric image generation module is used to preprocess the three-dimensional T1-weighted structured magnetic resonance images of an individual to obtain a gray matter volumetric image in standard space. The feature vector construction module is used to extract multiple radiomics features from the gray matter volume map, generate a feature map for each radiomics feature to obtain multiple initial feature maps, perform standardization and smoothing on the multiple initial feature maps to obtain multiple target feature maps, and construct a multidimensional feature vector for each voxel. The feature vector extraction module is used to select brain regions of interest as seed points and extract the seed point feature vectors of the seed points; The connectivity graph generation module is used to calculate the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis, forming a whole-brain correlation graph, and transforming the whole-brain correlation graph to obtain a morphological connectivity graph.

[0014] In addition, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and an individual morphological connectivity graph construction program stored in the memory and executable on the processor, wherein when the individual morphological connectivity graph construction program is executed by the processor, it implements the steps of the individual morphological connectivity graph construction method as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an individual morphological connectivity graph construction program, which, when executed by a processor, implements the steps of the individual morphological connectivity graph construction method as described above.

[0016] The beneficial effects of this invention are as follows: The present invention preprocesses three-dimensional T1-weighted structural magnetic resonance images of individuals to obtain gray matter volumetric maps in a standard space. This unifies brain images from different individuals into a standard space, eliminating interference caused by differences in brain size and shape between individuals and ensuring spatial comparability of data from different individuals. Simultaneously, by accurately segmenting and modulating the gray matter volumetric maps, the focus is placed on gray matter tissues closely related to cognitive function, laying an accurate and reliable foundation for subsequent morphological feature extraction.

[0017] This invention extracts multiple radiomics features from the gray matter volume image, generates a feature map for each radiomics feature, obtains multiple initial feature maps, and performs standardization and smoothing on these initial feature maps to obtain multiple target feature maps. A multidimensional feature vector is then constructed for each voxel. This successfully constructs a rich, multidimensional quantitative description for each voxel from a single-scan static structural image. It transforms single grayscale (volume) information into a multi-dimensional feature set containing deep information such as texture and shape. This ingeniously solves the core challenge of structural magnetic resonance imaging lacking time-series data, providing a data foundation for analytical methods simulating functional connectivity.

[0018] This invention selects brain regions of interest as seed points and extracts their feature vectors, breaking away from the reliance on predefined brain atlases in traditional methods. This allows researchers to freely choose any brain region of interest as the starting point for their research, based on specific scientific questions or clinical hypotheses. This flexibility enables the method to directly explore the connectivity patterns between specific brain regions and the whole brain, making it more intuitive and explanatory.

[0019] This invention calculates the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis, forming a whole-brain correlation map. This whole-brain correlation map is then transformed to obtain a morphological connectivity map, achieving visualization of whole-brain morphological connectivity patterns at the individual level while preserving voxel-level resolution. By calculating correlation on a voxel-by-voxel basis, this method avoids the information loss caused by aggregating and averaging voxel information in brain region-based methods, revealing more refined connectivity patterns. The final generated morphological connectivity map is similar to the functional connectivity map in functional magnetic resonance imaging (fMRI), visually displaying the distribution of connection strength between the seed point and the whole brain, and providing possibilities for individualized difference studies and the discovery of clinical biomarkers. Attached Figure Description

[0020] Figure 1 This is a flowchart of a preferred embodiment of the individual morphological connection diagram construction method of the present invention; Figure 2 This is a structural diagram of a preferred embodiment of the individual morphological connection diagram construction system of the present invention; Figure 3This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0021] This application provides a method, system, terminal, and storage medium for constructing an individual morphological connectivity map. To make the objectives, technical solutions, and effects of this application clearer and more explicit, the following detailed description is provided with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining this application and are not intended to limit this application.

[0022] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0023] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0024] In the field of functional magnetic resonance imaging (fMRI), seed-point-based functional connectivity is a mature and popular technique. By calculating the correlation between the time series data of a seed-point region and the time series data of every voxel in the whole brain, it can directly and intuitively reveal the functional connectivity patterns between specific brain regions and the whole brain, while preserving voxel-level information. However, directly transferring this approach to sMRI (structural magnetic resonance imaging) faces a core challenge: sMRI typically only provides static structural data from a single scan, lacking the time-varying time series data found in fMRI.

[0025] Radiomics is an emerging technology capable of extracting a large number of quantitative features from medical images (such as sMRI) in high throughput. These features can reflect information such as texture, shape, and intensity distribution that are not directly recognizable by the human eye. Previous studies have shown that radiomics features can be used to construct stable morphological networks at the brain region level. This offers a potential solution to the aforementioned challenges.

[0026] The preferred embodiment of the present invention describes a method for constructing individual morphological connectivity maps, such as... Figure 1 As shown, the method for constructing the individual morphological connectivity map includes the following steps: Step S10: Preprocess the three-dimensional T1-weighted structured magnetic resonance image of the individual to obtain a gray matter volume map in standard space.

[0027] The purpose of this step is to standardize the original individual brain structure images, eliminate spatial differences between individuals, and accurately segment gray matter tissue, laying the foundation for subsequent feature extraction.

[0028] Specifically, this step can be accomplished using a mature image processing toolkit, such as the CAT12 toolbox of the SPM12 software in the MATLAB environment. The preprocessing of the individual's three-dimensional T1-weighted structural magnetic resonance image to obtain a gray matter volume map in standard space includes the following sub-steps: S11. Gray matter images are segmented from the three-dimensional T1-weighted structural magnetic resonance images of an individual using the adaptive maximum a posteriori technique.

[0029] In this embodiment, the adaptive maximum a posteriori technique can accurately classify tissues in the T1-weighted image based on the tissue prior probability map, resulting in a clean grayscale image.

[0030] S12. Normalize the gray matter image to the standard Montreal Neuroscience Institute space to obtain the target gray matter image.

[0031] In this embodiment, advanced image registration algorithms (such as geodesic projection) are used to nonlinearly transform the gray matter image of an individual space to the standard Montreal Neurological Institute (MNI) space, thereby achieving spatial alignment of brain images of different individuals.

[0032] S13. Modulate the target gray matter map using the Jacobian determinant obtained during the standardization process to obtain a gray matter volume map.

[0033] In this embodiment, by applying the Jacobian determinant derived from the spatial normalization transformation to modulate the target gray matter map, the gray matter volume changes caused by volume stretching or compression during the spatial normalization process can be compensated, thereby obtaining a gray matter volume map that more accurately reflects the actual gray matter volume of an individual.

[0034] Step S20: Extract multiple radiomics features from the gray matter volume map, generate a feature map for each radiomics feature to obtain multiple initial feature maps, perform standardization and smoothing on the multiple initial feature maps to obtain multiple target feature maps, and construct a multidimensional feature vector for each voxel.

[0035] The purpose of this step is to extract deep morphological information from structural images that goes beyond the range of human visual recognition, construct a multidimensional feature description of each voxel, and simulate multidimensional data similar to "time series" in functional magnetic resonance imaging, thus creating conditions for subsequent calculation of the correlation between voxels.

[0036] Further, the step of extracting multiple radiomics features from the gray matter volume map, generating a feature map for each radiomics feature, and obtaining multiple initial feature maps includes: S211. Generate a binarized gray matter mask based on a priori brain tissue probability map.

[0037] In this embodiment, a gray matter probability map can be extracted from the prior probability map provided by toolkits such as SPM12, and a threshold (e.g., 0.4) can be set. Regions with probabilities higher than this threshold are defined as gray matter regions, and a binarized gray matter mask is generated to limit the region of interest for feature extraction.

[0038] S212. Using a radiomics feature extraction tool, calculate the value of each radiomics feature from the voxel-by-voxel in the gray matter volume map within the area defined by the gray matter mask.

[0039] In this embodiment, high-throughput radiomics feature extraction can be performed using open-source Python packages (such as pyradiomics). The extracted features can encompass histogram-based features (such as first-order statistical features) and texture-based features. Texture features further include categories such as gray-level co-occurrence matrix features, gray-level run-length matrix features, gray-level region size matrix features, neighborhood gray-level difference matrix features, and gray-level dependency matrix features. Through voxel-by-voxel calculation, numerical values ​​covering the entire brain's gray matter region are generated for each selected feature.

[0040] S213. Based on the value of each radiomics feature, generate an image covering the whole brain gray matter region for each radiomics feature to obtain multiple initial feature maps.

[0041] In this embodiment, the values ​​of each feature across all gray matter voxels are recombined into a three-dimensional image with the same spatial dimensions as the original gray matter volumetric map, thus obtaining the initial feature map for that feature. Extracting N features yields N initial feature maps.

[0042] Furthermore, the standardization and smoothing of the multiple initial feature maps to obtain multiple target feature maps includes: S221. Standardize each of the multiple initial feature maps.

[0043] In this embodiment, the Z-score standardization method is typically used to standardize all voxel values ​​within each initial feature map so that the mean is 0 and the standard deviation is 1, in order to eliminate the differences in dimensions and numerical ranges between different features.

[0044] S222. Using a spatial smoothing filter with a preset kernel function, perform convolution processing on each of the standardized initial feature maps to obtain multiple target feature maps.

[0045] In this embodiment, a spatial smoothing filter (such as a Gaussian smoothing filter) is used to smooth the normalized feature map. The full width at half maximum (FWHM) of the smoothing kernel can be set as needed (e.g., 8 mm). Smoothing helps improve the signal-to-noise ratio and conforms to the general assumptions of magnetic resonance imaging data processing.

[0046] Furthermore, the construction of the multidimensional feature vector for each voxel includes: S231. Spatial registration of all the target feature maps.

[0047] Since all feature maps originate from the same mask (gray matter mask) in the same space (MNI space), they are essentially spatially registered. This step aims to confirm the spatial consistency of all feature maps.

[0048] S232. For each voxel in the gray mask, extract the value of the target feature map corresponding to the voxel after all spatial registrations.

[0049] In this embodiment, for any voxel within the gray mask, the feature value at the voxel's location is extracted from the K target feature maps obtained after processing.

[0050] S233. Combine the values ​​in order to form a multidimensional vector to obtain a multidimensional feature vector.

[0051] In this embodiment, K feature values ​​extracted from a voxel are combined into a K-dimensional vector in a predetermined order (such as the order of feature extraction), which is the multidimensional feature vector of that voxel. Each voxel within the whole-brain gray matter mask corresponds to such a K-dimensional vector, thereby constructing a voxel-level multidimensional feature dataset.

[0052] Step S30: Select the brain region of interest as the seed point and extract the seed point feature vector of the seed point.

[0053] The purpose of this step is to identify the starting brain regions (seed points) that need to be examined in relation to the whole-brain connectivity patterns, based on the research objectives.

[0054] Furthermore, the selection of the brain region of interest as the seed point includes: Determine the center coordinates of the region of interest; Using the center coordinates as the center, a spherical region with a preset radius is generated as a seed point.

[0055] In this embodiment, the center coordinates of the region of interest can be determined based on existing neuroanatomical knowledge or functional brain atlas literature. For example, if studying the posterior cingulate cortex nodes of the default mode network, the standard coordinates of this region in MNI space reported in the literature can be used. Then, a sphere with a specified radius (e.g., 5 mm) is generated with this coordinate point as the center, and this spherical region is defined as the seed point.

[0056] In step S30, the seed point feature vector of the seed point is extracted, specifically as follows: Calculate the average value of all voxels within the seed point region (i.e., within the sphere) on each target feature map. Specifically, for the k-th feature map, calculate the average value of all voxels within the seed point on that feature map. Combine the K average values ​​corresponding to the K feature maps in order to obtain the seed point feature vector (a K-dimensional vector) representing the features of the entire seed point region.

[0057] Step S40: Calculate the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis to form a whole-brain correlation map. Transform the whole-brain correlation map to obtain a morphological connectivity map.

[0058] The purpose of this step is to quantify the similarity of the seed point to each voxel in the morphological feature pattern, thereby constructing a whole-brain morphological connectivity pattern map with the seed point as the core at the individual level.

[0059] Further, the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map, including: S41. Calculate the Pearson correlation coefficient on a voxel-by-voxel basis based on the seed point feature vector and the multidimensional feature vector of each voxel in the gray mask.

[0060] In this embodiment, for each voxel within the gray matter mask, the Pearson correlation coefficient between its K-dimensional eigenvector and the seed point's K-dimensional eigenvector is calculated. This coefficient reflects the degree of linear correlation between the voxel and the seed point in the multidimensional radiomics feature pattern.

[0061] S42. Generate the original whole-brain correlation map based on the Pearson correlation coefficients of all voxels.

[0062] In this embodiment, the Pearson correlation coefficient value calculated for each voxel location is filled back into the corresponding voxel position in the three-dimensional brain space matrix, thereby generating an original correlation map covering the entire brain.

[0063] S43. Perform Fisher Z-transform on the original whole-brain correlation map to obtain a Z-value map, and define the Z-value map as a morphological connectivity map.

[0064] In this embodiment, to improve the distribution of correlation coefficients to be closer to a normal distribution and facilitate subsequent statistical analysis, a Fisher's Z-transform is performed on each correlation coefficient value in the original whole-brain correlation map. The transformed Z-value map is the final individual morphological connectivity map, which visually shows the connection strength between each voxel and the selected seed point in morphological feature patterns across the whole brain.

[0065] Step S50 (Optional Population Analysis): Perform population-level statistical analysis on the morphological connectivity maps of a group of subjects, and after multiple comparison corrections, obtain the brain regions that have significant morphological connections with the seed points.

[0066] The purpose of this step is to perform group-level analysis based on individual connectivity maps to identify brain regions that are stable in the population and have significant morphological connections with seed points.

[0067] In this embodiment, if the study includes a group of subjects (e.g., a patient group and a control group), further group analysis is performed on the morphological connectivity maps (Z-maps) of each subject obtained in step S40. For example, to identify brain regions with significant positive connections to the seed point, a one-sample t-test can be performed on the Z-maps of the entire group of subjects (testing whether the mean is significantly greater than 0). To control the risk of false positives from multiple comparisons, an appropriate multiple comparison correction method should be used (e.g., cluster-level correction based on Gaussian random field theory or voxel-level correction based on the false discovery rate). After correction, the surviving significant brain regions are considered to have significant morphological connections to the seed point in this group. This step can be performed in statistical software such as SPM12 or FSL.

[0068] Example Suppose a group of subjects (e.g., healthy volunteers or patients with a certain disease) undergo MRI scans to obtain raw 3DT1-weighted images. First, step S10 is performed on the data of each subject, using the CAT12 toolbox for preprocessing to obtain a gray matter volumetric map of each subject in the MNI space. Next, step S20 is performed, using pyRadiomics to extract 93 radiomics features from each gray matter volumetric map, generating 93 initial feature maps. After standardization and 8mm FWHM Gaussian smoothing, 93 target feature maps are obtained, and a 93-dimensional feature vector is constructed for each voxel.

[0069] Suppose the research objective is to explore the morphological connectivity between the posterior cingulate cortex, one of the core nodes of the default mode network, and the whole brain. Then, step S30 is executed: the core coordinates of the posterior cingulate cortex in the MNI space are determined according to the literature (e.g., coordinates [0, -53, 26]). Using these coordinates as the center, a spherical seed point with a radius of 5 mm is generated, and the average value of 93 features within the seed point is calculated to obtain the seed point feature vector (a 1x93 vector).

[0070] Then, step S40 is executed. For each voxel in the gray matter mask, the Pearson correlation coefficient between its 93-dimensional feature vector and the 93-dimensional feature vector of the seed point is calculated to generate the original whole-brain correlation map. After Fisher Z-transform, the morphological connectivity map of each subject with respect to the posterior cingulate cortex is finally obtained.

[0071] Finally, if group analysis is required (step S50), a one-sample t-test is performed on the morphological connectivity maps of all subjects in the group, and multiple comparison corrections are performed using Gaussian random field theory to obtain the distribution of brain regions with significant morphological connectivity to the posterior cingulate cortex at the population level.

[0072] Through the above steps, this invention successfully transfers the concept of seed point-based functional connectivity analysis to structural magnetic resonance imaging data, and uses radiomics features to construct a voxel-level "pseudo-time series". This enables the creation of morphological connectivity maps at the individual level that retain voxel-level information and target specific brain regions of interest, providing a new and effective tool for studying individual differences in brain structural connectivity and its changes in neuropsychiatric diseases.

[0073] The beneficial effects of this invention are as follows: The present invention preprocesses three-dimensional T1-weighted structural magnetic resonance images of individuals to obtain gray matter volumetric maps in a standard space. This unifies brain images from different individuals into a standard space, eliminating interference caused by differences in brain size and shape between individuals and ensuring spatial comparability of data from different individuals. Simultaneously, by accurately segmenting and modulating the gray matter volumetric maps, the focus is placed on gray matter tissues closely related to cognitive function, laying an accurate and reliable foundation for subsequent morphological feature extraction.

[0074] This invention extracts multiple radiomics features from the gray matter volume image, generates a feature map for each radiomics feature, obtains multiple initial feature maps, and performs standardization and smoothing on these initial feature maps to obtain multiple target feature maps. A multidimensional feature vector is then constructed for each voxel. This successfully constructs a rich, multidimensional quantitative description for each voxel from a single-scan static structural image. It transforms single grayscale (volume) information into a multi-dimensional feature set containing deep information such as texture and shape. This ingeniously solves the core challenge of structural magnetic resonance imaging lacking time-series data, providing a data foundation for analytical methods simulating functional connectivity.

[0075] This invention selects brain regions of interest as seed points and extracts their feature vectors, breaking away from the reliance on predefined brain atlases in traditional methods. This allows researchers to freely choose any brain region of interest as the starting point for their research, based on specific scientific questions or clinical hypotheses. This flexibility enables the method to directly explore the connectivity patterns between specific brain regions and the whole brain, making it more intuitive and explanatory.

[0076] This invention calculates the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis, forming a whole-brain correlation map. This whole-brain correlation map is then transformed to obtain a morphological connectivity map, achieving visualization of whole-brain morphological connectivity patterns at the individual level while preserving voxel-level resolution. By calculating correlation on a voxel-by-voxel basis, this method avoids the information loss caused by aggregating and averaging voxel information in brain region-based methods, revealing more refined connectivity patterns. The final generated morphological connectivity map is similar to the functional connectivity map in functional magnetic resonance imaging (fMRI), visually displaying the distribution of connection strength between the seed point and the whole brain, and providing possibilities for individualized difference studies and the discovery of clinical biomarkers.

[0077] In summary, this invention innovatively transfers the mature paradigm of functional connectivity analysis to the field of structural imaging, solves the key data bottleneck through radiomics technology, and thus realizes the creation of high spatial resolution morphological connectivity maps at the individual level without the need for predefined brain regions, breaking through the limitations of traditional group-level and brain region-dependent methods.

[0078] Furthermore, such as Figure 2 As shown, the present invention also provides an individual morphological connectivity graph construction system, which is used to implement the individual morphological connectivity graph construction method described above, wherein the individual morphological connectivity graph construction system includes: The volume map generation module 51 is used to preprocess the three-dimensional T1-weighted structured magnetic resonance image of an individual to obtain a gray matter volume map in standard space. The feature vector construction module 52 is used to extract multiple radiomics features from the gray matter volume map, generate a feature map for each radiomics feature, obtain multiple initial feature maps, perform standardization and smoothing on the multiple initial feature maps to obtain multiple target feature maps, and construct a multidimensional feature vector for each voxel. The feature vector extraction module 53 is used to select the brain region of interest as a seed point and extract the seed point feature vector of the seed point; The connectivity graph generation module 54 is used to calculate the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis, forming a whole-brain correlation graph, and transforming the whole-brain correlation graph to obtain a morphological connectivity graph.

[0079] Furthermore, such as Figure 3 As shown, based on the above-mentioned method and system for constructing individual morphological connection diagrams, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 3 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0080] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage units. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores an individual morphological connectivity graph construction program 40, which can be executed by the processor 10 to implement the individual morphological connectivity graph construction method of this application.

[0081] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the individual morphological connectivity graph construction method.

[0082] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0083] In one embodiment, when the individual morphological connectivity graph construction program 40 is executed by the processor 10, it performs the following steps: Preprocessing of the individual's three-dimensional T1-weighted structured magnetic resonance image yields a gray matter volume map in standard space; Multiple radiomics features are extracted from the gray matter volume map, and a feature map is generated for each radiomics feature to obtain multiple initial feature maps. The multiple initial feature maps are standardized and smoothed to obtain multiple target feature maps, and a multidimensional feature vector for each voxel is constructed. Select brain regions of interest as seed points and extract the seed point feature vectors of the seed points; The correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map.

[0084] The preprocessing of the individual's three-dimensional T1-weighted structural magnetic resonance image to obtain a gray matter volume map in standard space includes: Gray matter images were segmented from three-dimensional T1-weighted structural magnetic resonance images of individuals using an adaptive maximum a posteriori technique. The gray matter image is normalized to the standard Montreal Neuroscience Institute space to obtain the target gray matter image; The target gray matter map is modulated using the Jacobian determinant obtained during the standardization process to obtain a gray matter volume map.

[0085] The process involves extracting multiple radiomics features from the gray matter volume map, generating a feature map for each radiomics feature, and obtaining multiple initial feature maps, including: Generate a binarized gray matter mask based on a priori brain tissue probability map; Using a radiomics feature extraction tool, the value of each selected feature is calculated voxel-by-voxel from the gray matter volume map within the region defined by the gray matter mask; Based on the value of each selected feature, an image covering the entire gray matter region of the brain is generated for each feature, resulting in multiple initial feature maps.

[0086] The standardization and smoothing process of the multiple initial feature maps to obtain multiple target feature maps includes: For each of the multiple initial feature maps, a standardization process is performed. Using a spatial smoothing filter with a preset kernel function, each of the standardized initial feature maps is convolved to obtain multiple target feature maps.

[0087] The construction of the multidimensional feature vector for each voxel includes: Spatial registration is performed on all the target feature maps; For each voxel within the gray mask, extract the value of the voxel at the corresponding position in the target feature map after all spatial registrations; The values ​​are combined in order to form a multidimensional vector, resulting in a multidimensional feature vector.

[0088] The selection of brain regions of interest as seed points includes: Determine the center coordinates of the region of interest; Using the center coordinates as the center, a spherical region with a preset radius is generated as a seed point.

[0089] The correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map, including: Based on the seed point feature vector and the multidimensional feature vector of each voxel in the gray mask, the Pearson correlation coefficient is calculated on a voxel-by-voxel basis. Based on the Pearson correlation coefficients of all voxels, a raw whole-brain correlation map is generated; The original whole-brain correlation map was subjected to Fisher Z-transform to obtain a Z-value map, which was defined as a morphological connectivity map.

[0090] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an individual morphological connectivity graph construction program, which, when executed by a processor, implements the steps of the individual morphological connectivity graph construction method as described above.

[0091] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0092] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0093] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for constructing an individual morphological connectivity graph, characterized in that, The method for constructing the individual morphological connectivity graph includes the following steps: Preprocessing of the individual's three-dimensional T1-weighted structured magnetic resonance image yields a gray matter volume map in standard space; Multiple radiomics features are extracted from the gray matter volume map, and a feature map is generated for each radiomics feature to obtain multiple initial feature maps. The multiple initial feature maps are standardized and smoothed to obtain multiple target feature maps, and a multidimensional feature vector for each voxel is constructed. Select brain regions of interest as seed points and extract the seed point feature vectors of the seed points; The correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map.

2. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The preprocessing of the individual's three-dimensional T1-weighted structural magnetic resonance image to obtain a gray matter volume map in standard space includes: Gray matter images were segmented from three-dimensional T1-weighted structural magnetic resonance images of individuals using an adaptive maximum a posteriori technique. The gray matter image is normalized to the standard Montreal Neuroscience Institute space to obtain the target gray matter image; The target gray matter map is modulated using the Jacobian determinant obtained during the standardization process to obtain a gray matter volume map.

3. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The process involves extracting multiple radiomics features from the gray matter volume map, generating a feature map for each radiomics feature, and obtaining multiple initial feature maps, including: Generate a binarized gray matter mask based on a priori brain tissue probability map; Using a radiomics feature extraction tool, the value of each radiomics feature is calculated voxel-by-voxel from the gray matter volume map within the area defined by the gray matter mask; Based on the value of each radiomics feature, an image covering the whole brain gray matter region is generated for each radiomics feature, resulting in multiple initial feature maps.

4. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The standardization and smoothing process of the multiple initial feature maps to obtain multiple target feature maps includes: For each of the multiple initial feature maps, a standardization process is performed. Using a spatial smoothing filter with a preset kernel function, each of the standardized initial feature maps is convolved to obtain multiple target feature maps.

5. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The construction of the multidimensional feature vector for each voxel includes: Spatial registration is performed on all the target feature maps; For each voxel within the gray mask, extract the value of the voxel at the corresponding position in the target feature map after all spatial registrations; The values ​​are combined in order to form a multidimensional vector, resulting in a multidimensional feature vector.

6. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The selection of brain regions of interest as seed points includes: Determine the center coordinates of the region of interest; Using the center coordinates as the center, a spherical region with a preset radius is generated as a seed point.

7. The method for constructing an individual morphological connectivity map according to claim 1, characterized in that, The correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point is calculated on a voxel-by-voxel basis to form a whole-brain correlation map. The whole-brain correlation map is then transformed to obtain a morphological connectivity map, including: Based on the seed point feature vector and the multidimensional feature vector of each voxel in the gray mask, the Pearson correlation coefficient is calculated on a voxel-by-voxel basis. Based on the Pearson correlation coefficients of all voxels, a raw whole-brain correlation map is generated; The original whole-brain correlation map was subjected to Fisher Z-transform to obtain a Z-value map, which was defined as a morphological connectivity map.

8. A system for constructing individual morphological connection diagrams, characterized in that, The individual morphological connectivity graph construction system is used to implement the individual morphological connectivity graph construction method as described in any one of claims 1-7, wherein the individual morphological connectivity graph construction system comprises: The volumetric image generation module is used to preprocess the three-dimensional T1-weighted structured magnetic resonance images of an individual to obtain a gray matter volumetric image in standard space. The feature vector construction module is used to extract multiple radiomics features from the gray matter volume map, generate a feature map for each radiomics feature to obtain multiple initial feature maps, perform standardization and smoothing on the multiple initial feature maps to obtain multiple target feature maps, and construct a multidimensional feature vector for each voxel. The feature vector extraction module is used to select brain regions of interest as seed points and extract the seed point feature vectors of the seed points; The connectivity graph generation module is used to calculate the correlation coefficient between the multidimensional feature vector of each voxel in the whole brain and the feature vector of the seed point on a voxel-by-voxel basis, forming a whole-brain correlation graph, and transforming the whole-brain correlation graph to obtain a morphological connectivity graph.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and an individual morphological connectivity graph construction program stored in the memory and executable on the processor, wherein when the individual morphological connectivity graph construction program is executed by the processor, it implements the steps of the individual morphological connectivity graph construction method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an individual morphological connectivity graph construction program, which, when executed by a processor, implements the steps of the individual morphological connectivity graph construction method as described in any one of claims 1-7.

Images