WO2021109727A1 - 脑功能图谱的绘制方法和系统 - Google Patents

脑功能图谱的绘制方法和系统 Download PDF

Info

Publication number
WO2021109727A1
WO2021109727A1 PCT/CN2020/121817 CN2020121817W WO2021109727A1 WO 2021109727 A1 WO2021109727 A1 WO 2021109727A1 CN 2020121817 W CN2020121817 W CN 2020121817W WO 2021109727 A1 WO2021109727 A1 WO 2021109727A1
Authority
WO
WIPO (PCT)
Prior art keywords
functional
voxel
atlas
brain function
brain
Prior art date
Application number
PCT/CN2020/121817
Other languages
English (en)
French (fr)
Inventor
魏可成
李海洋
胡清宇
Original Assignee
北京优脑银河科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京优脑银河科技有限公司 filed Critical 北京优脑银河科技有限公司
Priority to EP20895505.4A priority Critical patent/EP4071765A4/en
Priority to JP2022533142A priority patent/JP7356590B2/ja
Priority to US17/781,595 priority patent/US20220414979A1/en
Publication of WO2021109727A1 publication Critical patent/WO2021109727A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R33/00—Arrangements or instruments for measuring magnetic variables
    • G01R33/20—Arrangements or instruments for measuring magnetic variables involving magnetic resonance
    • G01R33/44—Arrangements or instruments for measuring magnetic variables involving magnetic resonance using nuclear magnetic resonance [NMR]
    • G01R33/48—NMR imaging systems
    • G01R33/4806—Functional imaging of brain activation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/20—Drawing from basic elements, e.g. lines or circles
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H15/00—ICT specially adapted for medical reports, e.g. generation or transmission thereof
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
    • G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2210/00—Indexing scheme for image generation or computer graphics
    • G06T2210/41—Medical
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the invention relates to the technical field of medical image processing, in particular to a method and system for drawing a brain function atlas.
  • German neuroanatomist Korbinian Brodmann first drew a human brain map (Brodmann brain map), pointing out that different areas of the brain are responsible for different functions. Since then, brain atlas has always been an important direction of brain science research. How the human brain is partitioned, the boundaries of regions and the connections between these regions are of great significance to the basic and clinical research of brain science.
  • the brain atlas can only be drawn in a "group” way, that is, studying the brains of a group of people and performing statistical analysis to draw an "average” map.
  • the "group” brain atlas can discover many commonalities and laws of humans, which is of great significance in scientific research.
  • the “group” atlas represents the American brain atlas, which was published on Nature by Glasser et al in 2016 (Glasser et al, Nature, A multi-modal resolution of human cerebral cortex, 2016).
  • Patent US 9,662,039 B2 (Liu et al.) describes a method of drawing an "individual" brain map. This method can successfully map 18 functional networks, which is a great improvement. However, when using this method to draw a higher-precision brain function map (for example, 112 networks), there will be stability problems and weak anti-noise shortcomings, so the results are unreliable and noisy. As far as we know, there is no good solution for drawing "individual" brain maps.
  • the main technical problem solved by the present invention is to provide a method for drawing a brain function map, which overcomes the problems of low stability, weak noise resistance, and unreliable results in drawing high-precision individual brain function maps in the prior art, so as to achieve accuracy. Reliably draw an individual's whole brain functional network, with high stability, high reliability and low noise at different precisions.
  • the technical problem to be solved by the present invention is also to provide a drawing system of brain function atlas.
  • the first aspect of the present invention provides a method for drawing a brain function atlas, which includes: initializing an individual’s brain function atlas using a brain function atlas template to obtain an initial individual brain function atlas.
  • the initial individual brain function atlas divides the individual brain There are multiple functional areas; the initial individual brain function atlas is divided into several large areas, and each large area includes several functional areas; into iteration, each iteration process includes: calculating each voxel in each large area in turn The degree of connection with each functional area in the large area, adjust each voxel to the functional area with the highest degree of connection with the voxel, until the adjustment of all voxels is completed; when the termination conditions are met, the iteration is terminated, Get the final individual brain function atlas. And, based on the final individual brain function atlas, corresponding reports and/or images can be generated.
  • each iteration process specifically includes: calculating the reference time series signal of each functional area according to the time series signal of all voxels in each functional area; determining an unadjusted volume Voxel is the current voxel; calculate the correlation value of the time series signal of the current voxel and the reference time series signal of each functional area in the current large area, and this correlation value is used as the connection between the current voxel and each functional area in the current large area Degree, the current large area is the large area to which the current voxel belongs; adjust the current voxel to the functional area with the highest connection to the current voxel; determine whether all voxels have been adjusted once, if not, return The step of determining an unadjusted voxel as the current voxel; if all the voxels have been adjusted once, the iteration process is ended.
  • the second possible implementation before entering the iteration, it also includes: initializing the credibility of each voxel; adjusting the current voxel to the highest connection with the current voxel After the functional area, it also includes: update the credibility of the current voxel; calculate the reference time series signal of each functional area, including: for each functional area, calculate that all the credibility of the functional area is not lower than the preset The average or median of the time series signal of the threshold voxel is used as the reference time series signal of the functional area.
  • the credibility of the current voxel is updated, including: the time series signal of the current voxel obtained from the calculation and the functional areas in the current large area From the correlation value of the reference time series signal, select the largest correlation value and the second largest correlation value; calculate and update the credibility of the current voxel, the credibility of the current voxel is equal to the largest correlation value and the second largest correlation value Ratio.
  • the iteration when the termination condition is met, the iteration is terminated, including: the number of iterations When the preset number of times is reached or the convergence criterion is reached, the iteration is terminated.
  • a population brain function atlas is preselected or generated as a brain function atlas template .
  • a group brain function atlas is preselected or generated as a brain function atlas template .
  • the steps include: divide the brain into several large regions; calculate a group-level functional connection diagram in each large region, the functional connection diagram is the voxels in the large region and the N regions of interest ROI in the large region
  • the atlas serves as the required brain function atlas template.
  • the initial individual brain function atlas is divided into several large areas, Including: By dividing the left and right cortex of the brain into five regions, the frontal lobe, the parietal lobe, the occipital lobe, the temporal lobe, and the pan-central sulcus region, the initial individual brain function map is divided into ten regions.
  • the second aspect of the present invention provides a method for drawing a population brain function map, including:
  • a group function connection diagram is calculated in each large area, and the group function connection diagram is a function connection matrix of voxels in the large area and N regions of interest in the large area, where N is a natural number; preferably, N is a natural number not less than 100;
  • the calculation of the group function connection diagram includes:
  • the group function connection diagram is the function connection matrix of voxels in the large area and N regions of interest in the large area, where N is a natural number;
  • the third aspect of the present invention provides a brain function atlas drawing system, including: an initialization module for initializing the individual brain function atlas using a brain function atlas template to obtain an initial individual brain function atlas.
  • the initial individual brain function The atlas divides the brain into multiple functional areas; the preprocessing module is used to divide the initial individual brain function atlas into several large areas, and each large area includes several functional areas; the iterative processing module is used to enter iterations, each An iterative process includes: sequentially calculating the connectivity of each voxel in each large area to each functional area in the large area, and adjusting each voxel to the functional area with the highest connection to the voxel until completion Adjustment of all voxels; when the termination conditions are met, the iteration is terminated, and the final individual brain function map is obtained.
  • the fourth aspect of the present invention provides a data processing system, including a processor and a memory; the memory is used to store computer execution instructions, and when the data processing system is running, the processor executes the computer stored in the memory The instruction is executed to cause the data processing system to execute the brain function atlas drawing method as described in the first aspect or any one of its possible implementation manners.
  • a fifth aspect of the present invention provides a medical image processing system, including: a data acquisition system, a data output interaction system, and the data processing system as described in the third aspect;
  • the data collection system is used to collect individual magnetic resonance scan data, and upload the collected data to the data processing system;
  • the data processing system is used to execute the brain function atlas drawing method as described in the first aspect to obtain the final individual brain function atlas, and based on the final individual brain function atlas, generate corresponding reports and/or image;
  • the data output interactive system is used to obtain the report and/or image from the data processing system, and display the report and/or image;
  • the data processing system includes a local server and/or a cloud computing platform; when the data processing system includes both the local server and the cloud computing platform, the local server and the cloud computing platform are based on load balancing Strategy and/or sharing strategy to jointly execute the brain function map drawing method described in the first aspect or any one of its possible implementation manners.
  • a sixth aspect of the present invention provides a computer-readable storage medium storing one or more programs, the one or more programs including computer-executable instructions, and when the computer-executable instructions are executed by a data processing system including a processor, The data processing system executes the brain function atlas drawing method as described in the first aspect or any one of its possible implementation manners.
  • the present invention performs iterative calculations on the initial individual brain function atlas, and continuously adjusts the functional area to which each voxel belongs through the iterative process, and finally can draw a high-precision individual brain function atlas with a variety of different precisions or resolutions. , Can achieve high stability, high reliability and low noise.
  • the whole brain can be divided into 56 functional areas, 112 functional areas, 213 functional areas and many other different precisions, and the results are stable and reliable, and successful clinical verification.
  • FIG. 1 is a schematic flowchart of a method for drawing a brain function atlas provided by an embodiment of the present invention.
  • Fig. 2 is a schematic flowchart of each iteration process in an embodiment of the present invention.
  • Fig. 3 is a schematic diagram of a process of generating a population brain function atlas template in an embodiment of the present invention.
  • Fig. 4 is a schematic flowchart of an embodiment of a specific application scenario of the present invention.
  • Fig. 5 is a schematic structural diagram of a system for drawing a brain function atlas provided by an embodiment of the present invention.
  • FIG. 6 is a schematic structural diagram of a data processing system provided by an embodiment of the present invention.
  • Figures 7-1 to Figure 7-3 are network architecture diagrams of medical image processing systems provided by embodiments of the present invention.
  • Fig. 8 is a clinical verification result of the brain function area drawn by the present invention and the brain function area measured by ECS;
  • FIG. 9 is a clinical verification result of the brain function area drawn by the present invention and the brain function area measured by long-time sampling task state NMR;
  • Figure 10 is an atlas of brain function drawn by an embodiment of the present invention.
  • Fig. 11 is a brain function map drawn by an embodiment of the present invention and a brain function map drawn by patent US 9,662,039 B2.
  • Brain functional atlas full English name: brain functional atlas
  • brain atlas or “brain map” or “functional atlas” or “brain functional network atlas”
  • brain functional network atlas refers to the division of brain functions in the cerebral cortex, marking the responsibility of the brain Maps of different areas with different functions; the marked areas are also called “functional areas” or “functional networks”.
  • Individualized brain functional atlas (full English name: individualized brain functional atlas), referred to as individual brain atlas, is an atlas that characterizes individual brain functions.
  • the group brain functional atlas (full English name: group level brain functional atlas), referred to as the group brain atlas, is a statistical atlas of brain function that characterizes a group of many people.
  • Brain functional networks (English full name: brain functional networks), a representation of the functional connections of the brain.
  • Functional magnetic resonance imaging (English full name: functional magnetic resonance imaging, English abbreviation: fMRI), abbreviated as functional magnetic resonance or magnetic resonance, functional magnetic resonance imaging shows the activation area of the brain when it is stimulated by external sources.
  • Task-state functional magnetic resonance (English full name: task functional magnetic resonance imaging, English abbreviation: task fMRI) is to let the subject do a specified task (for example, moving the tongue) during a magnetic resonance imaging scan, and hope to use magnetic resonance To find the brain function area or network responsible for the task.
  • Voxel also known as voxel, is an abbreviation for volume pixel. Voxels are conceptually similar to the smallest unit of two-dimensional space-pixels, which are used in the image data of two-dimensional computer images. A voxel is the smallest unit of digital data in the three-dimensional space segmentation, and is used in three-dimensional imaging, scientific data, and medical imaging.
  • BOLD (full English name: Blood Oxygen Level Dependency) signal is the blood oxygen level signal collected by functional magnetic resonance imaging.
  • Time series signals of voxels In functional magnetic resonance imaging, BOLD signals are generally collected for a period of time (for example, two minutes) for each voxel. The BOLD signal of this time period is the time series signal of the voxel;
  • a region of interest (English full name: region of interest, ROI for short) is a region that needs to be processed in the form of boxes, circles, ellipses, irregular polygons, etc., drawn from the processed image in machine vision and image processing.
  • the Pearson correlation coefficient is a coefficient used to measure whether two data sets are on a line, that is, to measure the linear relationship between the distance variables.
  • the calculation formula is:
  • the formula is defined as: the pearson correlation coefficient ( ⁇ x, y ) of two continuous variables (X, Y) is equal to the covariance cov (X, Y) between them divided by the product of their respective standard deviations ( ⁇ X , ⁇ Y ). The value of the coefficient is always between -1.0 and 1.0. Variables close to 0 are considered non-correlated, and close to 1 or -1 are called strong correlations.
  • an embodiment of the present invention provides a method for drawing a brain function atlas, which may include the steps:
  • MRI scan data such as functional magnetic resonance scan data, specifically such as BOLD signal, which is related to the functional areas of the brain; for example, structural magnetic resonance scan data, such as magnetic Resonance T1 signal, which is related to brain structure
  • pre-select or generate a population brain function map as a brain function map template e.g., MRI scan data (such as functional magnetic resonance scan data, specifically such as BOLD signal, which is related to the functional areas of the brain; for example, structural magnetic resonance scan data, such as magnetic Resonance T1 signal, which is related to brain structure)
  • Functional magnetic resonance is an emerging neuroimaging method. Its principle is to use magnetic resonance imaging to measure changes in blood oxygen levels caused by neuronal activity, and to collect BOLD signals. It is currently mainly used in the study of human and animal brain or Other functional activities of the nervous system. Resting functional magnetic resonance scanning refers to the individual being photographed lying in the MRI scanner in a quiet state, the whole body is relaxed, and the scanning is performed without any tasks or systematic thinking. In particular, the present invention can use resting state functional magnetic resonance scan data as raw data to draw brain function maps.
  • the fMRI scan data is four-dimensional imaging data based on voxels, including time series signals (BOLD signals) of all voxels.
  • the brain function atlas template divides the brain into multiple functional areas, such as 56 or 112 or other numbers.
  • the brain function atlas template can divide the brain into several large areas based on structural information.
  • the brain is divided into two large areas, the left and right cortex.
  • the brain is divided into high-level cortex and low-level cortex.
  • the brain is divided into four large areas according to the left and right cortex according to the high-level cortex and the low-level cortex; more preferably, the left and right cortex of the brain are divided into five large areas according to structural information, namely The frontal lobe, parietal lobe, occipital lobe, temporal lobe, and pan-central sulcus area divide the brain into ten regions. All functional areas belong to these large areas, and each large area includes several functional areas.
  • Initialization step initialize the individual brain function atlas using the brain function atlas template to obtain an initial individual brain function atlas.
  • the initial individual brain function atlas divides the brain into multiple functional areas.
  • the functional partitions of the brain function atlas template can be projected onto the reconstructed cerebral cortex of the individual to obtain the initialized individual brain function atlas.
  • This kind of "projection” is to complete one-to-many, many-to-many, and many-to-one registration through the point-level mapping relationship between different images of the same individual, using mathematical methods such as interpolation. Different registration methods (rigid, radial, nonlinear, etc.) use different formulas and constraints.
  • the BBR algorithm provided by Douglas Greve et al. is a commonly used "projection” method (Douglas Greve and Bruce Fishl, Accurate and Robust brain image alignment using boundary based registration, NeuroImage 2009).
  • Preprocessing step divide the initial individual brain function atlas into several large areas, and each large area includes several functional areas. Among them, several refers to at least two.
  • the initial individual brain function map can be divided into ten regions.
  • the left and right brains can be divided into 4 regions according to the high-level and low-level cortex of the brain.
  • the credibility of each voxel can also be initialized, and the credibility represents the credibility that the voxel functionally belongs to the functional area where it is currently located.
  • the initial value of the credibility of each voxel can be set to 1 through initialization. In subsequent iterative calculations, the credibility will be continuously updated.
  • each iteration process includes: by sequentially calculating the connection degree between each voxel in each large area and each functional area in the large area, adjusting each voxel to the volume The functional area with the most connected voxel until the adjustment of all voxels is completed; when the termination conditions are met, the iteration is terminated, and the final individual brain function map is obtained.
  • corresponding reports and/or images showing brain function divisions can be generated and output.
  • This iterative calculation step is a process of repeated iterative calculations. In each iteration, all voxels are adjusted. After this iteration process is over, it is judged whether the end condition is met. If not, the output result of the previous iteration process is taken as Input, enter the next iteration, and adjust all voxels again; until the end condition is met, the final individual brain function map is output.
  • each voxel has been adjusted many times and adjusted to the most possible functional area, with high accuracy.
  • a better balance can be reached between accuracy and calculation time, and a very high-precision final individual brain function map can be obtained.
  • whether the convergence criterion is met or whether the preset number of iterations is reached can be used as the iteration termination condition.
  • the number of iterations reaches the preset number or reaches the convergence standard, the iteration is terminated.
  • a convergence criterion is: the change in the two iterations before and after is very small: for example, the change in the two iterations before and after is less than 1%, preferably less than 0.1% or 0.5%, etc.; optionally, another convergence criterion is: all The credibility reaches a certain threshold.
  • the threshold can be, for example, 2-10, preferably 3, 4, or 5; optionally, the preset number of times can be, for example, 5 to 1000, and preferably, the preset test is 100 Times, 150 times, or 200 times.
  • the degree of connection between each voxel in each large area and each functional area in the large area may be calculated by using the correlation value of the time series signal of the voxel and the reference time series signal of the functional area.
  • the correlation value represents the correlation, correlation coefficient, or similarity between the time series signal of the voxel and the reference time series signal of the functional area. The higher the similarity between the two, the higher the correlation value, which means the closer the two are, and the degree of connection The higher (or bigger, stronger).
  • multiple correlation calculation methods can be used to calculate the correlation value, which is not limited herein.
  • the Pearson correlation value of the time series signal of the voxel and the reference time series signal of the functional area can be calculated to indicate the degree of connectivity between the voxel and the functional area.
  • each iteration process may specifically include:
  • the current region is the region to which the current voxel belongs.
  • the accuracy of functional area division is improved, and the individual brain function atlas can be updated.
  • the current voxel can be adjusted or re-allocated to any functional area with the maximum correlation value.
  • the largest correlation value and the second largest correlation value can be selected from the calculated correlation values of the current voxel time series signal and the reference time series signal of each functional area in the current large area; calculation and update
  • the credibility of the current voxel is set, the credibility of the current voxel is equal to the ratio of the largest correlation value to the second largest correlation value.
  • the end of this iteration process means that a round of adjustments and updates have been completed to the individual brain function atlas; next, if the iteration termination conditions have not been met, then based on the updated individual brain function atlas obtained in this iteration process, enter the next iteration process.
  • a population brain function atlas can be generated as a brain function atlas template by the following methods, the steps include:
  • A1 Divide the brain into several large areas based on structural information.
  • the left and right cortex of the brain can be divided into five large areas, namely the frontal lobe, parietal lobe, occipital lobe, temporal lobe and pan-central sulcus area. That is, the brain is divided into ten regions.
  • ROI For each voxel in each large area of each individual, calculate the functional connectivity of the voxel to each ROI in the large area to obtain a multi-dimensional vector, such as a 1000-dimensional vector or a 4098-dimensional vector, preferably, ROI can be obtained by uniform sampling;
  • clustering algorithms are used to classify each voxel in each large area, and finally each large area is divided into dozens to dozens of "classes”.
  • clustering algorithms include but are not limited to K-means and related algorithms, spectral clustering algorithm (spectral clustering), Gaussian mixed model (Gaussian Mixed Model).
  • the index of the clustering algorithm uses the "distance" between voxels in a 1000-dimensional space , Divide the "distance” voxels into the same fine-grained functional partition to achieve the "local optimum".
  • A5. Use the merging algorithm to merge the fine-grained functional partitions of multiple large areas, thereby creating a template for the group-level functional partition of the whole brain.
  • the merging method includes, but is not limited to, direct splicing and merging, and edge trimming and edge voxel redistribution are performed after splicing.
  • the obtained population brain function atlas template can have different resolutions, ranging from dozens of functional areas to hundreds of functional areas.
  • FIG. 4 is a flowchart of a method for drawing a brain function atlas in an embodiment of a specific application scenario. The steps include:
  • B5. Divide the individual brain atlas into multiple local areas, that is, multiple large areas, and each large area covers multiple functional areas.
  • step B7 Determine whether the iteration termination condition is met, if not, go to step B8; if yes, go to step B16.
  • step B10 Determine whether all voxels have completed a calculation, if yes, return to step B7; if not, determine that the next voxel that has not been calculated and adjusted is the current voxel, and then go to step B11.
  • a method for drawing a brain function atlas performs iterative calculations on the initial individual brain function atlas, and continuously adjusts each voxel to which each voxel belongs through an iterative process.
  • the functional area can finally draw a high-precision individual brain function map, which can achieve high stability, high reliability and low noise at a variety of different precisions or resolutions.
  • the present invention first divides large areas, and then adjusts or redistributes voxels in each large area with the large area as a unit; this method of individualization of large areas is compared with the prior art The whole brain individualized way of directly distributing voxels in the whole brain area.
  • the voxel adjustment in each large area naturally introduces such as "the voxels of the same fine-grained functional partition should belong to the same large area.
  • the whole brain can be divided into 56 functional areas, 112 functional areas, 213 functional areas and many other different precisions, and the results are stable and reliable, and successful clinical verification.
  • Verification method one please refer to Figure 8.
  • the left column is the cerebral cortex of one person, and the right column is the cerebral cortex of another person: the first row is the use of intraoperative electrical cortical stimulation (ECS) technology Positioning the functional area (that is, the common clinical method of positioning the functional area), the second line is the positioning of the functional area by the method of the present invention. It can be seen that the position of the functional area located by the method of the present invention is the same as the position located by the ECS technology, that is It is verified that the positioning of the present invention is accurate.
  • ECS intraoperative electrical cortical stimulation
  • Verification method 2 Please refer to Figure 9 to verify the motion network. Different color parts are part of the motion network. For example, one part controls the hand movement and the other part controls the body movement (face, head, tongue).
  • the right column is a motion network diagram obtained through a large number of experiments using traditional methods (task state functional nuclear magnetics).
  • the specific positioning method is to allow healthy people to perform tasks under nuclear magnetic functional scanning, such as doing tongue movements or pinching fingers at the same time Scan, exercise for 1.5 hours, and locate the brain function area related to the movement after scanning for 10 consecutive days. Due to a large number of experimental samples, this method can accurately find the corresponding functional area.
  • the left column is the method of the present invention. It can be seen that the positioning area on the left is basically the same as the positioning area on the right.
  • the yellow border in the right picture is the result of the comparison between the present invention and the right picture, showing that the two methods are positioned.
  • the degree of overlap of the sports function network is very high.
  • the present invention does not require any tasks, and the sampling time is short (20 minutes on a 3T MRI machine). Compared with the task state functional NMR, the advantage is very obvious.
  • the short-term confirmation of individual brain functional areas is more conducive to clinical application. ;
  • FIG. 10 is an atlas of brain function obtained by using a specific drawing method of the present invention.
  • the specific drawing steps are as follows: a structural and resting functional magnetic resonance scan of a young normal subject is performed and signal preprocessing is performed.
  • a brain atlas template of 213 functional areas at the population level is selected for initialization, that is, the population-level partitions are projected onto the individual cerebral cortex through non-linear registration.
  • the individual cerebral cortex is divided into five major areas on the left and right according to the structure: frontal lobe, parietal lobe, occipital lobe, temporal lobe, and pan-central sulcus area, and then individualized calculation iterations are performed on each large area on the individual cortex.
  • the time series signals of all voxels in each functional division of the initialization are averaged.
  • the average signal obtained is used as the reference signal of the partition, with a total of 17 reference signals.
  • the time series signal of each voxel in the left frontal lobe and 17 reference signals are calculated by Pearson coefficient as the correlation value, and the 17 correlation values are sorted in descending order. Reallocate the voxel to the functional partition with the largest correlation value and calculate the credibility.
  • the credibility is the ratio of the largest correlation value to the second largest correlation value, and the value range of the credibility is [1,+ ⁇ ).
  • the voxel can be assigned to the functional area with high confidence.
  • FIG. 11 is an effect comparison diagram of the brain function atlas drawn by the method of the present invention and the brain function atlas drawn by the prior art.
  • the method of the present invention is used to draw functional areas, and the brain is divided into functional areas by means of individualization of large areas.
  • the functional area is drawn using the technical solution of patent US 9,662,039B2, which uses the method of simultaneous individualization of the whole brain to divide the functional areas of the brain.
  • Noise is a small area in the picture that violates the principles of neuroscience.
  • Neuroscience principles include: at the same resolution, there should be no other functional areas inside a functional area of the nervous system. If the inside of a red area appears green, it indicates that there is a problem with the logical result; The boundary usually has a certain degree of smoothness. If the boundary is staggered, or there are multiple discontinuous areas in a functional area, it is usually caused by noise interference in the processing method.
  • an embodiment of the present invention provides a brain function atlas drawing system, which may include:
  • the initialization module 51 is used to initialize the individual brain function atlas using the brain function atlas template to obtain an initial individual brain function atlas, which divides the brain into multiple functional areas;
  • the preprocessing module 52 is used to divide the initial individual brain function atlas into several large areas, and each large area includes several functional areas;
  • the iterative processing module 53 is used to enter iterations.
  • Each iteration process includes: successively calculating the connection degree of each voxel in each large area with each functional area in the large area, and adjusting each voxel to the volume The functional area with the most connected voxel until the adjustment of all voxels is completed; when the termination conditions are met, the iteration is terminated, and the final individual brain function map is obtained.
  • brain function atlas drawing system of the embodiment of the present invention may adopt different implementation forms.
  • the server obtains the fMRI scan data from the magnetic resonance equipment, performs calculation processing, and obtains the final individual brain function atlas, and can further generate and output corresponding reports and/or images showing the functional areas of the brain.
  • the server can obtain fMRI scan data from the magnetic resonance device and upload the data to the cloud platform. Then, the cloud platform performs calculation and processing to obtain the final individual brain function atlas, and can further generate and output corresponding reports and/or images showing brain function divisions. The cloud platform can also send the final individual brain function atlas, reports and/or images back to the server.
  • the server can obtain fMRI scan data from the magnetic resonance equipment, and cooperate with the cloud platform for calculation and processing to obtain the final individual brain function atlas, and can further generate and output corresponding reports and/or brain functional divisions image.
  • a brain function atlas drawing system is provided.
  • the system is used to iteratively calculate the initial individual brain function atlas, and continuously adjust each voxel through the iterative process.
  • the functional area it belongs to can finally draw a high-precision individual brain function map, which can achieve high stability, high reliability and low noise at a variety of different precisions or resolutions.
  • the whole brain can be divided into 56 functional areas, 112 functional areas, 213 functional areas and many other different precisions, and the results are stable and reliable, and successful clinical verification.
  • an embodiment of the present invention also provides a data processing system 60, which may include:
  • Processor 61 memory 62, communication interface 63, bus 64,
  • the processor 61, the memory 62, and the communication interface 63 communicate with each other through the bus 64; the communication interface 63 is used to receive and send data; the memory 62 is used to store computer execution instructions, when the data is processed When the system is running, the processor 61 executes the computer-executable instructions stored in the memory 62, so that the data processing system executes the brain function atlas drawing method described in the above method embodiment.
  • the data processing system 60 can be composed of local computer equipment such as a local server or a cloud computing platform, or can be composed of both.
  • An embodiment of the present invention also provides a medical image processing system, including: a data acquisition system 71, a data output interaction system 72, and the data processing system as described above;
  • the data processing system includes a local server 73 and/or a cloud computing platform 74;
  • the data acquisition system 71 includes a magnetic resonance device, etc., which is used to collect individual functional magnetic resonance scan data and upload the collected data to the data processing system;
  • the data processing system is used to execute the brain function atlas drawing method as described in the embodiment of FIG. 1 above, to obtain the final individual brain function atlas, and based on the final individual brain function atlas, to generate corresponding data processing results, for example Reports and/or images;
  • the data output interaction system 72 includes a display device, an input/output device, etc., for obtaining data processing results in the form of reports and/or images from the data processing system, and displaying the data processing results;
  • the data processing system may only include a local server, as shown in Figure 7-1; it can also include only a cloud computing platform, as shown in Figure 7-2; or, it can also include both a local server and a cloud computing platform. As shown in Figure 7-3.
  • the local server and the cloud computing platform jointly execute the brain function map provided by the present invention based on a load balancing strategy and/or a sharing strategy. Drawing method.
  • the interaction relationship between the local server and the cloud computing platform may include, for example:
  • a load balancing strategy between the cloud computing platform and the local server can be, for example: when the load of the local server exceeds a certain set value, for example, 70%, new work enters the cloud computing The platform performs processing.
  • the embodiment of the present invention also provides a computer storage medium, a computer-readable storage medium storing one or more programs, the one or more programs including computer-executable instructions, and the computer-executable instructions include data of a processor.
  • the processing system is executed, the data processing system is caused to execute the brain function atlas drawing method as described in the above method embodiment.
  • the disclosed system, device, and method can be implemented in other ways.
  • the device embodiments described above are merely illustrative, for example, the division of units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or integrated. To another system, or some features can be ignored, or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.
  • the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium.
  • the technical solution of the present invention essentially or the part that contributes to the existing technology or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium.
  • a computer device which can be a personal computer, a server, or a network device, etc.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other media that can store program codes. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Epidemiology (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • General Physics & Mathematics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Theoretical Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Computer Graphics (AREA)
  • Geometry (AREA)
  • Software Systems (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Neurosurgery (AREA)
  • Business, Economics & Management (AREA)
  • General Business, Economics & Management (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)
  • Image Analysis (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

一种脑功能图谱的绘制方法和系统。方法包括：利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱；将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。本方法可绘制高精度的个体脑功能图谱，在多种不同的精度或者说分辨率下，都能实现高稳定性、高可靠性和低噪声。

Description

脑功能图谱的绘制方法和系统 技术领域
本发明涉及医学图像处理技术领域，具体涉及一种脑功能图谱的绘制方法和系统。
背景技术
19世纪初，德国神经解剖学家Korbinian Brodmann首次绘制人类大脑图谱(Brodmann脑图谱)，指出了大脑的不同区域是负责不同功能的。从此，脑图谱一直是脑科学研究的重要方向。人类大脑如何分区，区域的界限及这些区域之间的联系，对脑科学基础及临床研究具有重要的意义。
长期以来，因为对人脑的不了解和技术的局限，脑图谱只能通过“群体”的方式绘制，也就是研究一群人的大脑，进行统计分析来画出“平均”的图谱。“群体”脑图谱可发现很多人类共性和规律，在科研上很有意义。目前代表美国脑图谱的就是“群体”图谱，这个脑图谱是由Glasser等人于2016年发在Nature上(Glasser et al，Nature，A multi-modal parcellation of human cerebral cortex，2016)。
但在临床上，医生关注的是病人个体，需要的是“个体”脑图谱。事实上，人类的大脑千差万别，没有两个人的大脑是完全相同的(Mueller et al,Neuron,Individual Variability in Functional Connectivity Architecture of the Human Brain,2013)。“群体”脑图谱抹杀了个体的独特性，在临床上意义有限，甚致有误导性。所以，“个体”脑图谱的绘制是一个非常需要但一直没有解决的问题。
专利US 9,662,039 B2(Liu et al.)描述了一个绘制“个体”脑图谱的方法。该方法可成功绘制出18个功能网络，是一大进步。但是，当用该方法来绘制更高精度的脑功能图谱时(比方说112个网络)，会出现稳定性问题和抗噪声性弱的缺点，因此结果不可靠，噪声大。据我们所知，目前“个体”脑图谱的绘制还没有好的解决方法。
发明内容
本发明主要解决的技术问题在于提供一种脑功能图谱的绘制方法，克服现有技术绘制高精度个体脑功能图谱中存在的稳定性低、抗噪声性弱、结果不可靠等问题，以实现准确可靠的绘制出个体的全脑功能网络，在不同的精度下都具有高稳定性、高可靠性和低噪声。本发明要解决的技术问题还在于提供一种脑功能图谱的绘制系统。
为解决上述技术问题，本发明采用的技术方案如下：
本发明第一方面提供一种脑功能图谱的绘制方法，包括：利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将个体大脑分为多个功能区；将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。并且，可基于最终的个体脑功能图谱，生成相应的报告和/或图像。
在第一种可能的实现方式中，每一次迭代过程具体包括：根据每个功能区中的所有体素的时间序列信号，计算每个功能区的参考时间序列信号；确定一个未经调整的体素为当前体素；计算当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值，该相关值作为当前体素与当前大区中的各个功能区的连接度，所述当前大区是当前体素所属的大区；将当前体素调整到与当前体素的连接度最高的功能区；判断是否所有体素都已经完成一次调整，若否，则返回所述确定一个未经调整的体素为当前体素的步骤；若所有体素都已经完成一次调整，则结束该次迭代过程。
结合第一种可能的实现方式，在第二种可能的实现方式中，进入迭代之前还包括：对每个体素的可信度进行初始化；将当前体素调整到与当前体素的连接度最高的功能区之后，还包括：更新当前体素的可信度；计算每个功能区的参考时间序列信号，包括：对于每个功能区，计算该功能区的所有可信度不低于预设阈值的体素的时间序列信号的平均值或中位数，作为该功能区的参考时间序列信号。
结合第二种可能的实现方式，在第三种可能的实现方式中，更新当前体素的可信度，包括：从计算得到的当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值中，选出最大相关值和第二大相关值；计算和更新当前体素的可信度，当前体素的可信度等于最大相关值与第二大相关值的比值。
结合第一方面或第一方面的第一种至第三种可能的实现方式中的任一种，在第四种可能的实现方式中，在满足终结条件时，终止迭代，包括：在迭代次数达到预设次数或者达到收敛标准时，终止迭代。
结合第一方面或第一方面的第一种至第四种可能的实现方式中的任一种，在第五种可能的实现方式中，预先选择或生成一个群体脑功能图谱作为脑功能图谱模板。
结合第一方面或第一方面的第一种至第五种可能的实现方式中的任一种，在第六种可能的实现方式中，预先选择或生成一个群体脑功能图谱作为脑功能图谱模板，步骤包括：将大脑分成若干个大区；在每个大区内计算出群体水平的功能连接图，该功能连接图为大区内的体素和该大区内的N个感兴趣区域ROI的功能连接矩阵，其中N为自然数；优选地，N为不小于100的自然数；基于所述功能连接图作为特征，利用聚类算法，将每个大区分成多个细颗粒度功能分区；综合评价聚类算法的指标和最大化功能同质性的指标，确定多个局部最优的分区数；利用合并算法将大区内的细颗粒度功能分区进行合并，创建出全脑的群体脑功能图谱作为所需要的脑功能图谱模板。
结合第一方面或第一方面的第一种至第五种可能的实现方式中的任一种，在第七种可能的实现方式中，将初始的个体脑功能图谱分为若干个大区，包括：通过将大脑的左、右皮层各分为额叶、顶叶、枕叶、颞叶和泛中央沟区域五个大区，将初始的个体脑功能图谱分为十个大区。
本发明第二方面提供一种群体脑功能图谱的绘制方法，包括：
获取一群人的脑部磁共振扫描数据；
将大脑分为若干个大区；
在每个大区内计算出群体功能连接图，该群体功能连接图为大区内的体素和该大区内的N个感兴趣区域的功能连接矩阵，其中，N为自然数；优选地，N为不小于100的自然数；
基于所述群体功能连接图作为特征，利用聚类算法，将每个大区分成多个细颗粒度功能分区；
综合评价聚类算法的指标和最大化功能同质性的指标，确定多个局部最优的分区数；
利用合并算法将大区内的细颗粒度功能分区进行合并，创建出全脑的群体脑功能图谱。
在第二方面可能的实现方式中，所述计算出群体功能连接图，包括：
在每个大区内计算出个体功能连接图，该群体功能连接图为大区内的体素和该大区内的N个感兴趣区域的功能连接矩阵，其中，N为自然数；
计算每个大区中的所有个体功能连接图的平均值，得到群体功能连接图
本发明第三方面提供一种脑功能图谱的绘制系统，包括：初始化模块，用于利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将大脑分为多个功能区；预处理模块，用于将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；迭代处理模块，用于进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。
本发明第四方面提供一种数据处理系统，包括处理器和存储器；所述存储器用于存储计算机执行指令，当所述数据处理系统运行时，所述处理器执行所述存储器存储的所述计算机执行指令，使所述数据处理系统执行如第一方面或者其任一种可能的实现方式中所述的脑功能图谱的绘制方法。
本发明第五方面提供一种医学图像处理系统，包括：数据采集系统，数据输出交互系统，以及，如第三方面所述的数据处理系统；
所述数据采集系统，用于采集个体的磁共振扫描数据，将采集的数据上传到所述数据处理系统；
所述数据处理系统，用于执行如第一方面中任一所述的脑功能图谱的绘制方法，得到最终的个体脑功能图谱，并基于最终的个体脑功能图谱，生成相应的报告和/或图像；
所述数据输出交互系统，用于从所述数据处理系统获取所述报告和/或图像，并显示所述报告和/或图像；
其中，所述数据处理系统包括本地服务器和/或云计算平台；当所述数据处理系统同时包括所述本地服务器和所述云计算平台时，所述本地服务器和所述云计算平台基于负载均衡策略和/或共享策略，共同执行第一方面或者其任一种可能的实现方式中任一所述的脑功能图谱的绘制方法。
本发明第六方面提供一种存储一个或多个程序的计算机可读存储介质，所述一个或多个程序包括计算机执行指令，所述计算机执行指令被包括处理器的数据处理系统执行时，使所述数据处理系统执行如第一方面或者其任一种可能的实现方式中所述的脑功能图谱的绘制方法。
由上可见，在本发明的一些可行的实施方式中，通过采用上述技术方案，取得了以下有益效果：
本发明对初始的个体脑功能图谱进行迭代计算，通过迭代过程持续调整其中每个体素所属的功能区，最终可以绘制出高精度的个体脑功能图谱，在多种不同的精度或者说分辨率下，都能实现高稳定性、高可靠性和低噪声。本发明实践中，可把全脑分成56个功能区，112个功能区，213个功能区等多种不同精度，且结果稳定可靠，并成功临床验证。
附图说明
为了更清楚地说明本发明实施例技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍。
图1是本发明一实施例提供的一种脑功能图谱的绘制方法的流程示意图。
图2是本发明一实施例中每一次的迭代过程的流程示意图。
图3是本发明一实施例中生成一个群体脑功能图谱模板的流程示意图。
图4是本发明一具体应用场景实施例的流程示意图。
图5是本发明一实施例提供的一种脑功能图谱的绘制系统的结构示意图。
图6是本发明一实施例提供的一种数据处理系统的结构示意图；
图7-1至图7-3是本发明实施例提供的医学图像处理系统的网络架构图；
图8是本发明绘制的脑功能区与ECS测出的脑功能区的临床验证结果；
图9是本发明绘制的脑功能区与长时间采样的任务态核磁测出的脑功能区的临床验证结果；
图10是本发明一实施例绘制的脑功能图谱；
图11是本发明一实施例绘制的脑功能图谱与专利US 9,662,039 B2绘制的脑功能图谱。
具体实施方式
为了使本技术领域的人员更好地理解本发明方案，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分的实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本发明保护的范围。
本发明的说明书和权利要求书及上述附图中的术语“包括”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或单元的过程、方法、系统、产品或设备没有限定于已列出的步骤或单元，而是可选地还包括没有列出的步骤或单元，或可选地还包括对于这些过程、方法、产品或设备固有的其它步骤或单元。
为便于理解，首先介绍本发明涉及的若干名词：
脑功能图谱(英文全称：brain functional atlas)，简称“脑图谱”或“脑图”或“功能图谱”或“脑功能网络图谱”；是指对大脑皮层进行脑功能分区，标记出大脑的负责不同功能的不同区域的图谱；被标记的区域也称为“功能区”或“功能网络”。
个体脑功能图谱(英文全称：individualized brain functional atlas)，简称个体脑图谱，是表征个体的脑功能图谱。
群体脑功能图谱(英文全称：group level brain functional atlas)，简称群体脑图谱，是表征包括很多人的群体的统计性的脑功能图谱。
脑功能网络(英文全称：brain functional networks)，对大脑功能连接的表示。
功能磁共振成像(英文全称：functional magnetic resonance imaging，英文简称：fMRI)，简称功能磁共振或磁共振，功能磁共振成像显示脑部受外界刺激时的活化区域。
任务态功能磁共振(英文全称：task functional magnetic resonance imaging，英文简称：task fMRI)是在磁共振成像扫描时，让受试者做指定的任务(比方说，动舌头)，从而希望用磁共振来找到负责该任务的脑功能区或网络。
体素，又称立体像素(voxel)，是体积像素(volume pixel)的简称。体素从概念上类似二维空间的最小单位——像素，像素用在二维电脑图像的影像数据上。体素是数字数据于三维空间分割上的最小单位，应用于三维成像、科学数据与医学影像等领域。
BOLD(英文全称：Blood Oxygen Level Dependency)信号是功能磁共振成像所采集的血氧水平信号。
体素的时间序列信号：功能磁共振成像时，一般对每个体素会采集一段时间的BOLD信号(比方说两分钟)。这个时间段的BOLD信号为该体素的时间序列信号；
感兴趣区域(英文全称：region of interest，简称ROI)，是机器视觉、图像处理中，从被处理的图像以方框、圆、椭圆、不规则多边形等方式勾勒出的需要处理的区域。
皮尔森(pearson)相关系数，是用来衡量两个数据集合是否在一条线上，即衡量定距变量间的线性关系的系数。其计算公式为：
Figure PCTCN2020121817-appb-000001
公式定义为：两个连续变量(X,Y)的pearson相关性系数(ρ x,y)等于它们之间的协方差cov(X,Y)除以它们各自标准差的乘积(σ X,σ Y)。系数的取值总是在-1.0到1.0之间，接近0的变量被成为无相关性，接近1或者-1被称为具有强相关性。
下面通过具体实施例，对本发明进行详细的说明。
请参考图1，本发明实施例提供一种脑功能图谱的绘制方法，可包括步骤：
S1、数据获取步骤：获得被摄者个体的大脑部位的磁共振扫描数据(例如功能磁共振扫描数据，具体例如BOLD信号，其与大脑功能区相关；又例如结构磁共振扫描数据，具体例如磁共振T1信号，其与大脑结构相关)，以及，预先选择或生成一个群体脑功能图谱作为脑功能图谱模板。
功能磁共振是一种新兴的神经影像学方式，其原理是利用磁振造影来测量神经元活动所引发之血氧水平的改变，采集BOLD信号，目前主要是运用在研究人及动物的脑或其它神经系统的功能活动。静息态功能磁共振扫描是指被摄者个体在安静状态下躺在磁共振扫描仪中，全身放松，不做任何任务或系统的思考进行扫描。本发明尤其可利用静息态功能磁共振扫描数据作为原始数据来绘制脑功能图谱。功能磁共振扫描数据是基于体素的四维成像数据，包括所有体素的时间序列信号(BOLD信号)。
预先获取一个群体脑功能图谱是必要的，它作为脑功能图谱模板，用来对个体的脑功能图谱进行初始化。脑功能图谱模板将大脑划分为多个功能区，例如56个或112个或其它数量。可选的，脑功能图谱模板可根据结构信息将大脑分为若干个大区，优选地，将大脑分为左、右皮层两个大区，优选地，将大脑按照高级皮层和低级皮层分为两个大区；更优选地，将大脑按照左右皮层按照高级皮层和低级皮层分为四个大区；更优选地，将大脑的左、右皮层按照结构信息各分为五个大区，即额叶、顶叶、枕叶、颞叶和泛中央沟区域，这样将大脑分成了十个大区。所有的功能区分别归属于这些大区，每个大区包括若干个功能区。
S2、初始化步骤：利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将大脑分为多个功能区。
初始化步骤中，可针对大脑结构的每个大区，将脑功能图谱模板的功能分区投射到个体重构的大脑皮层上来获得初始化的个体脑功能图谱。这种“投射”是通过对同一个体的不同图像之间在点级别的映射关系，用插值等数学方法完成一对多，多对多，多对一等进行配准。不同的配准方法(刚性，放射，非线性等)所用的公式和约束条件不同。例如，Douglas Greve等人提供的BBR算法是一个常用的“投射”方法(Douglas Greve and Bruce Fishl,Accurate and robust brain image alignment using boundary based registration,NeuroImage 2009)。
由于个体差异，这种初始的个体脑功能图谱是不反映该个体的真正的脑功能分区的，需要经后续的迭代计算来提高准确度和个体精度。
S3、预处理步骤：将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区。其中，若干个是指至少两个。
例如，可通过将大脑的左、右皮层各分为额叶、顶叶、枕叶、颞叶和泛中央沟区域五个大区，将初始的个体脑功能图谱分为十个大区。再例如，可按大脑的高级皮层和低级皮层分区，左右脑共分4个区域。
该预处理步骤中，还可以对每个体素的可信度进行初始化，该可信度表示该体素在功能上属于当前所在功能区的可信程度。可选的，可通过初始化将每个体素的可信度的初始值设为1。后续迭代计算中，该可信度将被不断的更新。
S4、迭代计算步骤：进入迭代，每一次迭代过程包括：通过依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。并且，可基于最终的个体脑功能图谱，生成、输出相应的显示脑功能分区的报告和/或图像。
该迭代计算步骤是一个反复迭代计算的过程，每一次迭代中都对所有体素进行调整，该次迭代过程结束后，判断是否满足终结条件，若不满足，则以上一次迭代过程输出的结果作为输入，进入下一次迭代，再次对所有体素进行调整；直至满足终结条件，输出最终的个体脑功能图谱。
最终的个体脑功能图谱中，每个体素已经经过多次的调整，被调整至最可能的功能区中，具有较高的准确度。通过设置合适的迭代终结条件，可以在精度和计算时间之间达成较好的平衡，得到极高精度的最终的个体脑功能图谱。
可选的，可以以是否满足收敛标准或者是否达到预设的迭代次数，作为迭代终结条件。在迭代次数达到预设次数或者达到收敛标准时，终止迭代。
可选的，一种收敛标准是：前后两次迭代变化非常小：比如前后两次迭代的变化小于1％，优选小于0.1％或0.5％等；可选的，另一种收敛标准是：所有可信度达到一定阈值，根据经验，该阈值例如可以是2～10，优选为3，4或5；可选的，预设次数例如可以是5～1000次，优选地，预设测试为100次，150次，或200次。
可选的，每个大区中的每个体素与该大区中的每个功能区的连接度，可以采用该体素的时间序列信号与功能区的参考时间序列信号的相关值来计算。相关值表示体素的时间序列信号与功能区的参考时间序列信号的相关性、相关系数、或者说相似程度，两者相似程度越高，则相关值越高，表示两者越接近，连接度越高(或者说越大、越强)。
进一步的，可采用多种相关关系的计算方法来计算所述相关值，本文不予限制。例如，可以计算体素的时间序列信号与功能区的参考时间序列信号的皮尔森相关值，来表示该体素与该功能区的连接度。
请参考图2，一些实施例中，每一次的迭代过程可具体包括：
S41、根据每个功能区中的所有体素的时间序列信号，计算每个功能区的参考时间序列信号。例如，对于每个功能区，可计算该功能区的所有可信度不低于预设阈值的体素的时间序列信号的平均值或中位数，作为该功能区的参考时间序列信号。可选的，在首次迭代时，对于每个功能区，可计算该功能区的所有体素的时间序列信号的平均值或中位数，作为该功能区的参考时间序列信号。
S42、确定一个未经调整的体素为当前体素。
S43、计算当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值，该相关值作为当前体素与当前大区中的各个功能区的连接度，所述当前大区是当前体素所属的大区。
S44、将当前体素调整到与当前体素的连接度最高的功能区。
通过将体素调整到与其连接度最高(即相关值最大)的功能区，来提高功能区划分的准确度，实现对个体脑功能图谱的更新。可选的，如果有多个相等的最大相关值，则可以将当前体素调整或者说重新分配到任意一个最大相关值的功能区。
S45、更新当前体素的可信度。
可选的，可以从计算得到的当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值中，选出最大相关值和第二大相关值；计算和更新当前体素的可信度时，令当前体素的可信度等于最大相关值与第二大相关值的比值。
S46、判断是否所有体素都已经完成一次调整，若否，则进入步骤42，选择下一个体素，继续执行上述步骤43和44；若是，进入步骤47。
S47、若所有体素都已经完成一次调整，则结束该次迭代过程。
本次迭代过程结束，表示对个体脑功能图谱完成了一轮调整和更新；接下来，若尚未满足迭代终结条件，则基于本次迭代过程得到的更新后的个体脑功能图谱，进入下一次迭代过程。
请参考图3，一些实施例中，采集群体的功能磁共振扫描数据后，可通过如下方法生成一个群体脑功能图谱作为脑功能图谱模板，步骤包括：
A1.根据结构信息将大脑进行分为若干大区，可选的，可将大脑的左右皮层各分为五个大区，即额叶、顶叶、枕叶、颞叶和泛中央沟区域，即将大脑分成十个大区。
A2.在每个大区内计算出群体功能连接图(functional connectivity profile)，即，区域内的体素和该大区内若干个感兴趣区域(region of interest，ROI)的功能连接矩阵，具体可通过如下步骤完成：
①对于每个个体的每个大区中的体素，计算该体素到该大区内每个ROI的功能连接度，得到一个多维向量，例如1000维向量，或者4098维向量，优选地，ROI可通过均匀取样获得；
②对群体中所有个体的每个大区中每个体素的多向量求平均，得到每个体素的群体功能连接向量，将所有体素的功能连接向量组成矩阵，得到群体功能连接图(或称群体功能连接矩阵)。
A3.基于这些功能连接图作为特征，利用聚类算法，把每个大区内的每个体素进行归类，最后将每个大区分成数十个到几十个数量不等的“类”(细颗粒度功能分区)，聚类算法包括但不限于K-means及其相关算法，谱聚类算法(spectral clustering)，高斯混合模型(Gaussian Mixed Model)。
A4.综合评价聚类算法的指标和最大化功能同质性的指标，确定多个局部最优的分区数，例如，聚类算法的指标利用体素之间在1000维空间里的“距离”，把“距离”近的体素分到同一个细颗粒度功能分区从而达到“局部最优”。
A5.利用合并算法将多个大区的细颗粒度功能分区进行合并，从而创建出全脑的群体水平功能分区的模板。合并方法包括但不限于直接拼接合并，拼接后进行边缘修整和边缘体素重新分配。得到的群体脑功能图谱模板可以有不同的分辨率，分辨率为几十个功能区到几百个功能区不等。
可以理解，本发明实施例上述方案例如可以在本地服务器、云计算平台等本地或云端的数据处理系统具体实施。
为便于更好的理解本发明实施例提供的技术方案，下面通过一个具体场景下的实施方式为例进行介绍。
请参考图4，是一个具体应用场景实施例中，脑功能图谱的绘制方法的流程图，步骤包括：
B1.获得个体的功能磁共振扫描数据，包括所有体素的时间序列信号。
B2.选择一个脑功能图谱模板。
B3.用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱。
B4.把每个体素的可信度初始化为最大可信度1。
B5.把个体脑图谱分为多个局部区域，即多个大区，每个大区涵盖多个功能区。
B6.进入迭代。
B7.判断是否满足迭代终结条件，若否，进入步骤B8；若是，进入步骤B16。
B8.计算所有功能区的参考信号(全称：参考时间序列信号)：对于每个功能区，选用可信度不低于一定阈值的体素来计算该功能区的参考时间序列信号。
B9.依次对每个体素进行一次以下步骤的计算。
B10.判断是否所有体素都已经完成一次计算，若是，返回步骤B7；若否，确定下一个尚未计算和调整的体素为当前体素，进入步骤B11。
B11.对每个与当前体素位于同一大区的功能区，计算当前体素与该功能区的连接度。
B12.对于当前体素位于同一大区的所有功能区的连接度进行排序。
B13.调整个体脑功能图谱：把当前体素分配到与其连接度最强的功能区。
B14.计算当前体素属于所调整到的功能区的可信度。
B15.返回步骤B10。
B16.若满足迭代终结条件，结束迭代，生成相应的报告和/或图像。
由上可见，在本发明的一些可行的实施方式中，提供了一种脑功能图谱的绘制方法，该方法对初始的个体脑功能图谱进行迭代计算，通过迭代过程持续调整其中每个体素所属的功能区，最终可以绘制出高精度的个体脑功能图谱，在多种不同的精度或者说分辨率下，都能实现高稳定性、高可靠性和低噪声。
需要说明的是，本发明通过首先划分大区，然后以大区为单位，在每个大区内进行体素的调整或者说重新分配；该种大区个体化的方式，相对于现有技术直接在全脑范围内进行体素分配的全脑个体化的方式，一方面，在每个大区内进行体素调整自然而然的引入了如“同一细颗粒度功能分区的体素应该属于同一大区”这些神经科学的一般规律作为分区的额外信息，有效提高了抗噪音抗干扰能力；一方面，由于在大区这样一个较小的范围内进行体素调整，可以有效提高处理精度和提高计算速度。
本发明实践中，可把全脑分成56个功能区，112个功能区，213个功能区等多种不同精度，且结果稳定可靠，并成功临床验证。
验证方法一，请参考图8，图8中左边一列是一个人的大脑皮层，右边一列是另一个人的大脑皮层：第一行是采用术中直接皮层电刺激(electrical cortical stimulation,ECS)技术定位功能区(也就是常用定位功能区的临床方法)，第二行是本发明的方法定位功能区，可见，本发明的方法定位出的功能区的位置和ECS技术定位出的位置一致，即验证了本发明的定位准确。
验证方法二：请参考图9，验证运动网络，不同色彩部位都是运动网络的一部分，例如某一部位是控制手部运动，另一部位控制躯体运动(脸部，头部，舌头)。
图9中，右边一列是采用传统方法(任务态功能核磁)经大量实验采样得到的运动网络图，具体定位方法是让健康人在核磁功能扫描下做任务，例如做舌头运动或捏手指的同时进行扫描，运动1.5 小时，连续10天的扫描后定位出相关运动的脑功能区域。由于有大量实验采样，该方法可准确的找出相应的功能区。
图9中，左边一列是本发明的方法，可以看到左边的定位区域和右边的定位区域基本一致，右边图中黄色边界是本发明和右边的图比较的结果，显示出两种方法定位得到的运动功能网络重合度非常高。另外，本发明不要求做任何任务，而且采样时间短(在3T磁共振机器上二十分钟即可)，相对于任务态功能核磁优势非常明显，短时间确认个体大脑功能区更有利于临床应用；
请参考图10，是采用本发明一个具体的绘制方法得到的脑功能图谱。具体绘制步骤如下：对一个年轻正常被试者进行了结构和静息态功能磁共振扫描并进行信号预处理。选择一个群体水平的213个功能区的脑图谱模板进行初始化，即将群体水平分区经过非线性配准投射到个体大脑皮层上。把个体大脑皮层按结构分为左右各五个大区：额叶、顶叶、枕叶、颞叶和泛中央沟区域区，然后在个体皮层上对每个大区分别进行个体化计算迭代。以左侧额叶的17个功能分区为例：在个体化的第一次迭代的时候，将初始化的每个功能分区内的所有体素的时间序列信号进行平均。得到的平均信号作为该分区的参考信号，共17个参考信号。然后对左侧额叶中每个体素的时间序列信号和17个参考信号通过计算皮尔森系数作为相关值，并对17个相关值进行降序排序。将该体素重新分配给相关值最大的功能分区并计算可信度。可信度为最大相关值和第二大相关值的比值，可信度的取值范围为[1,+∞)。设定可行度≥3时可将体素分配给功能区是高可信度，比如，第110个体素和分区10的相关值为0.78且该相关值为最大相关值，其与分区8的相关值为0.18且该相关值为第二大相关值，则其可信度为0.78/0.18＝4.3，意味着我们有极大信心将该体素分配到分区10。再比如，第125个体素最强的功能连接度是分区8的0.35，其与分区4的功能连接度0.29为第二强，其可信度为0.35/0.29＝1.2，意味着我对将该体素分配到分区8比较缺乏信心。对所有体素重复以上操作以完成第一次个体化更新。从第二次迭代开始，每次重新计算17个参考信号，此时的参考信号由每个分区中可信度大于3的体素的时间序列信号平均得到，如果分区中无可信度大于3的体素则选择分区中可信度最高的前5％的体素代替。大区中每个体素再根据和参考信号的相关值的大小进行分区的重新分配和可信度的估计。完成大区内所有体素的更新后进入下一次迭代，直到前后两次迭代的个体分区结果相似度达到99％以上或者完成110次迭代后停止迭代。最后在所有大区完成个体化之后，利用直接合并的方法将10个大区进行合并。至此，个体化的功能分区计算完成，得到的213个功能区的个体脑功能图谱。在个体脑功能图谱各个功能区的边界加上边界线，以更加清晰地区分各个功能区。
请参考图11，是采用本发明方法绘制的脑功能图谱与现有技术绘制的脑功能图谱的效果对比图。
如图11中的左边的(a)所示，采用本发明的方法绘制功能区，通过分大区个体化的方式，对大脑划分功能区。如图11中的右边的(b)所示，采用专利US 9,662,039B2的技术方案绘制功能区，其采用全脑同时个体化的方式，对大脑划分功能区。
从图中可以看出，右边的(b)有很多明显的噪音，如图中101和102所示。噪音为图中违反神经科学原则的小区域。神经科学原则包括：在同一分辨率下，神经系统的一个功能区内部一般不应该有其它功能区，如果一个红色的区域的内部出现绿色，则表明这个出理结果有问题；每个功能区的边界通常有一定光滑度的，如果边界犬牙交错，或者一个功能区出现多个不连续的区域，通常是处理方法被噪音干扰造成的。
为了更好的实施本发明实施例的上述方案，下面还提供用于配合实施上述方案的相关装置。
请参考图5，本发明实施例提供一种脑功能图谱的绘制系统，可包括：
初始化模块51，用于利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将大脑分为多个功能区；
预处理模块52，用于将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；
迭代处理模块53，用于进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。
可以理解，本发明实施例的脑功能图谱的绘制系统的各个功能模块的功能可根据上述方法实施例中的方法具体实现，其具体实现过程可参照上述方法实施例中的相关描述，此处不再赘述。
可以理解，本发明实施例的脑功能图谱的绘制系统，可以采用不同的实现形式。
一方面，可以由本地计算机设备例如服务器来实现。服务器从磁共振设备获取功能磁共振扫描数据，进行计算处理，得到最终的个体脑功能图谱，并可进一步生成、输出相应的显示大脑功能分区的报告和/或图像。
另一方面，也可以由云计算平台来实现。首先，可以由服务器从磁共振设备获取功能磁共振扫描数据，并将数据上传至云平台。然后，由云平台进行计算处理，得到最终的个体脑功能图谱，并可进一步生成、输出相应的显示大脑功能分区的报告和/或图像。云平台还可以将最终的个体脑功能图谱、报告和/或图像传回服务器。
再一方面，也可以由服务器、云计算平台共同构成的数据处理系统来实现。首先，可以由服务器从磁共振设备获取功能磁共振扫描数据，并与云平台协作进行计算处理，得到最终的个体脑功能图谱，并可进一步生成、输出相应的显示大脑功能分区的报告和/或图像。
由上可见，在本发明的一些可行的实施方式中，提供了一种脑功能图谱的绘制系统，该系统用于对初始的个体脑功能图谱进行迭代计算，通过迭代过程持续调整其中每个体素所属的功能区，最终可以绘制出高精度的个体脑功能图谱，在多种不同的精度或者说分辨率下，都能实现高稳定性、高可靠性和低噪声。本发明实践中，可把全脑分成56个功能区，112个功能区，213个功能区等多种不同精度，且结果稳定可靠，并成功临床验证。
请参考图6，本发明实施例还提供一种数据处理系统60，可包括：
处理器61，存储器62，通信接口63，总线64，
所述处理器61，存储器62，通信接口63通过所述总线64相互的通信；所述通信接口63，用于接收和发送数据；所述存储器62用于存储计算机执行指令，当所述数据处理系统运行时，所述处理器61执行所述存储器62存储的所述计算机执行指令，使所述数据处理系统执行如上文方法实施例所述的脑功能图谱的绘制方法。
该数据处理系统60可以由本地计算机设备例如本地服务器或云计算平台组成，也可以由两者共同组成。
请参考图7-1至图7-3，本发明实施例还提供一种医学图像处理系统，包括：数据采集系统71，数据输出交互系统72，以及，如上文所述的数据处理系统；所述数据处理系统包括本地服务器73和/或云计算平台74；
所述数据采集系统71，包括磁共振设备等，用于采集个体的功能磁共振扫描数据，将采集的数据上传到所述数据处理系统；
所述数据处理系统，用于执行如上文如图1实施例所述的脑功能图谱的绘制方法，得到最终的个体脑功能图谱，并基于最终的个体脑功能图谱，生成相应的数据处理结果例如报告和/或图像；
所述数据输出交互系统72，包括显示设备、输入输出设备等，用于从所述数据处理系统获取例如报告和/或图像形式的数据处理结果，并显示所述数据处理结果；
其中，所述数据处理系统可以仅包括本地服务器，如图7-1所示；也可以仅包括云计算平台，如图7-2所示；或者，也可以同时包括本地服务器和云计算平台，如图7-3所示。
当所述数据处理系统同时包括所述本地服务器和所述云计算平台时，所述本地服务器和所述云计算平台基于负载均衡策略和/或共享策略，共同执行本发明提供的脑功能图谱的绘制方法。
本地服务器和云计算平台的交互关系例如可以包括：
1.本地服务器无法完成的工作给云计算平台做；云计算平台和本地服务器的一种负载平衡策略例如可以是：本地服务器的负载超过某一设定值例如70％时，新工作进入云计算平台进行处理。
2.当用户具有共享要求或其他位置需要时，可将工作交给云计算平台。
本发明实施例还提供一种计算机存储介质，一种存储一个或多个程序的计算机可读存储介质，所述一个或多个程序包括计算机执行指令，所述计算机执行指令被包括处理器的数据处理系统执行时，使所述数据处理系统执行如上文方法实施例所述的脑功能图谱的绘制方法。
需要说明的是，对于前述的各方法实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本发明并不受所描述动作顺序的限制，因为依据本发明，某些步骤可以采用其它顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作和模块并不一定是本发明所必须的。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装置的具体工作过程，可以参考前述方法实施例中的对应过程。某个实施例中没有详细描述的部分，可以参见其它实施例的相关描述
在本申请所提供的几个实施例中，应该理解到，所揭露的系统，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本发明各个实施例方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
以上对本发明实施例所提供的脑功能图谱的绘制方法和系统进行了详细介绍，本文中应用了具体个例对本发明的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本发明的方法及其核心思想；同时，对于本领域的一般技术人员，依据本发明的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本发明的限制。

Claims (12)

  1. 一种个体脑功能图谱的绘制方法，其特征在于，包括：
    利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将大脑分为多个功能区；
    将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；
    进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；
    在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。
  2. 根据权利要求1所述的方法，其特征在于，
    每一次迭代过程具体包括：
    根据每个功能区中的所有体素的时间序列信号，计算每个功能区的参考时间序列信号；
    确定一个未经调整的体素为当前体素；
    计算当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值，该相关值作为当前体素与当前大区中的各个功能区的连接度，所述当前大区是当前体素所属的大区；
    将当前体素调整到与当前体素的连接度最高的功能区；
    判断是否所有体素都已经完成一次调整，若否，则返回所述确定一个未经调整的体素为当前体素的步骤；
    若所有体素都已经完成一次调整，则结束该次迭代过程。
  3. 根据权利要求2所述的方法，其特征在于，
    进入迭代之前还包括：对每个体素的可信度进行初始化；
    将当前体素调整到与当前体素的连接度最高的功能区之后，还包括：更新当前体素的可信度；
    计算每个功能区的参考时间序列信号，包括：对于每个功能区，计算该功能区的所有可信度不低于预设阈值的体素的时间序列信号的平均值或中位数，作为该功能区的参考时间序列信号。
  4. 根据权利要求3所述的方法，其特征在于，所述更新当前体素的可信度，包括：
    从计算得到的当前体素的时间序列信号和当前大区中的各个功能区的参考时间序列信号的相关值中，选出最大相关值和第二大相关值；
    计算和更新当前体素的可信度，当前体素的可信度等于最大相关值与第二大相关值的比值。
  5. 根据权利要求1-4中任一所述的方法，其特征在于，在满足终结条件时，终止迭代，包括：
    在迭代次数达到预设次数或者达到收敛标准时，终止迭代。
  6. 根据权利要求1-4中任一所述的方法，其特征在于，还包括：预先选择或生成一个群体脑功能图谱作为脑功能图谱模板，步骤包括：
    将大脑分为若干个大区；
    在每个大区内计算出群体功能连接图，该功能连接图为大区内的体素和该大区内的N个感兴趣区域的功能连接矩阵，其中N为自然数；
    基于所述功能连接图作为特征，利用聚类算法，将每个大区分成多个细颗粒度功能分区；
    综合评价聚类算法的指标和最大化功能同质性的指标，确定多个局部最优的分区数；
    利用合并算法将大区内的细颗粒度功能分区进行合并，创建出全脑的群体脑功能图谱作为所需要的脑功能图谱模板。
  7. 一种群体脑功能图谱的绘制方法，其特征在于，包括：
    获取一群人的脑部磁共振扫描数据；
    将大脑分为若干个大区；
    在每个大区内计算出群体功能连接图，该群体功能连接图为大区内的所有体素和该大区内的N个感兴趣区域的功能连接矩阵，其中，N为自然数；
    基于所述群体功能连接图作为特征，利用聚类算法，将每个大区分成多个细颗粒度功能分区；
    综合评价聚类算法的指标和最大化功能同质性的指标，确定多个局部最优的分区数；
    利用合并算法将大区内的细颗粒度功能分区进行合并，创建出全脑的群体脑功能图谱。
  8. 根据权利要求7所述的绘制方法，其特征在于，所述计算出群体功能连接图，包括：
    在每个大区内计算出个体功能连接图，该群体功能连接图为大区内的体素和该大区内的N个感兴趣区域的功能连接矩阵，其中，N为自然数；
    计算每个大区中的所有个体功能连接图的平均值，得到群体功能连接图。
  9. 一种个体脑功能图谱的绘制系统，其特征在于，包括：
    初始化模块，用于利用脑功能图谱模板对个体的脑功能图谱进行初始化，得到初始的个体脑功能图谱，该初始的个体脑功能图谱将大脑分为多个功能区；
    预处理模块，用于将初始的个体脑功能图谱分为若干个大区，每个大区包括若干个功能区；
    迭代处理模块，用于进入迭代，每一次迭代过程包括：依次计算每个大区中的每个体素与该大区中的每个功能区的连接度，将每个体素调整到与该体素的连接度最高的功能区，直至完成对所有体素的调整；在满足终结条件时，终止迭代，得到最终的个体脑功能图谱。
  10. 一种数据处理系统，其特征在于，包括处理器和存储器；
    所述存储器用于存储计算机执行指令，当所述数据处理系统运行时，所述处理器执行所述存储器存储的所述计算机执行指令，使所述数据处理系统执行如权利要求1-6中任一所述的脑功能图谱的绘制方法。
  11. 一种医学图像处理系统，其特征在于，包括：数据采集系统，数据输出交互系统，以及，如权利要求10所述的数据处理系统；
    所述数据采集系统，用于采集个体的磁共振扫描数据，将采集的数据上传到所述数据处理系统；优选地，所述磁共振扫描数据包括功能磁共振扫描数据和/或结构磁共振扫描数据；
    所述数据处理系统，用于执行如权利要求1-7中任一所述的脑功能图谱的绘制方法，得到最终的个体脑功能图谱，并基于最终的个体脑功能图谱，生成相应的报告和/或图像；
    所述数据输出交互系统，用于从所述数据处理系统获取所述报告和/或图像，并显示所述报告和/或图像；
    其中，所述数据处理系统包括本地服务器和/或云计算平台；当所述数据处理系统同时包括所述本地服务器和所述云计算平台时，所述本地服务器和所述云计算平台基于负载均衡策略和/或共享策略，共同执行如权利要求1-6中任一所述的脑功能图谱的绘制方法。
  12. 一种存储一个或多个程序的计算机可读存储介质，所述一个或多个程序包括计算机执行指令，所述计算机执行指令被包括处理器的数据处理系统执行时，使所述数据处理系统执行如权利要求1-6中任一所述的脑功能图谱的绘制方法。
PCT/CN2020/121817 2019-12-02 2020-10-19 脑功能图谱的绘制方法和系统 WO2021109727A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP20895505.4A EP4071765A4 (en) 2019-12-02 2020-10-19 METHOD AND SYSTEM FOR DRAWING A BRAIN FUNCTION ATLAS
JP2022533142A JP7356590B2 (ja) 2019-12-02 2020-10-19 脳機能地図を描画するための方法およびシステム
US17/781,595 US20220414979A1 (en) 2019-12-02 2020-10-19 Method and system for drawing brain functional atlas

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911214999.9A CN111081351B (zh) 2019-12-02 2019-12-02 脑功能图谱的绘制方法和系统
CN201911214999.9 2019-12-02

Publications (1)

Publication Number Publication Date
WO2021109727A1 true WO2021109727A1 (zh) 2021-06-10

Family

ID=70312476

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/121817 WO2021109727A1 (zh) 2019-12-02 2020-10-19 脑功能图谱的绘制方法和系统

Country Status (5)

Country Link
US (1) US20220414979A1 (zh)
EP (1) EP4071765A4 (zh)
JP (1) JP7356590B2 (zh)
CN (2) CN111081351B (zh)
WO (1) WO2021109727A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114187227A (zh) * 2021-09-13 2022-03-15 北京银河方圆科技有限公司 脑肿瘤累及区域的功能区确定方法及装置

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111081351B (zh) * 2019-12-02 2024-02-20 北京优脑银河科技有限公司 脑功能图谱的绘制方法和系统
CN112002428B (zh) * 2020-08-24 2022-03-08 天津医科大学 以独立成分网络为参照的全脑个体化脑功能图谱构建方法
CN113571142B (zh) * 2021-06-07 2023-04-25 四川大学华西医院 精神影像一体化系统
CN113450893B (zh) * 2021-06-11 2023-04-11 北京银河方圆科技有限公司 脑功能区定位和定侧方法、装置、设备及存储介质
CN114140377B (zh) * 2021-09-13 2023-04-07 北京银河方圆科技有限公司 脑肿瘤患者的脑功能图谱确定方法及装置
CN115082586B (zh) * 2022-07-12 2023-04-04 中国科学院自动化研究所 群组先验引导的基于深度学习的丘脑个体化图谱绘制方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105117731A (zh) * 2015-07-17 2015-12-02 常州大学 一种大脑功能网络的社团划分方法
US9662039B2 (en) 2014-03-31 2017-05-30 The General Hospital Corporation System and method for functional brain organization mapping
CN107330948A (zh) * 2017-06-28 2017-11-07 电子科技大学 一种基于流行学习算法的fMRI数据二维可视化方法
CN111081351A (zh) * 2019-12-02 2020-04-28 北京优脑银河科技有限公司 脑功能图谱的绘制方法和系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007043462A1 (ja) * 2005-10-12 2007-04-19 Tokyo Denki University 脳機能解析方法および脳機能解析プログラム
WO2012041364A1 (en) * 2010-09-28 2012-04-05 Brainlab Ag Advanced fiber tracking and medical navigation in a brain
JP2015054218A (ja) 2013-09-13 2015-03-23 株式会社東芝 磁気共鳴イメージング装置及び画像処理装置
CN106055881A (zh) * 2016-05-25 2016-10-26 中国科学院苏州生物医学工程技术研究所 用于医疗数据处理的云平台
CN108305279B (zh) * 2017-12-27 2019-07-05 东南大学 一种迭代空间模糊聚类的大脑磁共振图像超体素生成方法
CN111583181A (zh) * 2020-04-08 2020-08-25 深圳市神经科学研究院 个体化脑功能图谱构建方法及系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9662039B2 (en) 2014-03-31 2017-05-30 The General Hospital Corporation System and method for functional brain organization mapping
CN105117731A (zh) * 2015-07-17 2015-12-02 常州大学 一种大脑功能网络的社团划分方法
CN107330948A (zh) * 2017-06-28 2017-11-07 电子科技大学 一种基于流行学习算法的fMRI数据二维可视化方法
CN111081351A (zh) * 2019-12-02 2020-04-28 北京优脑银河科技有限公司 脑功能图谱的绘制方法和系统

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
GLASSER ET AL.: "A multi-modal parcellation of human cerebral cortex", NATURE, 2016
MUELLER ET AL.: "Individual Variability in Functional Connectivity Architecture of the Human Brain", NEURON, 2013

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114187227A (zh) * 2021-09-13 2022-03-15 北京银河方圆科技有限公司 脑肿瘤累及区域的功能区确定方法及装置
CN114187227B (zh) * 2021-09-13 2023-01-24 北京银河方圆科技有限公司 脑肿瘤累及区域的功能区确定方法及装置

Also Published As

Publication number Publication date
EP4071765A4 (en) 2023-01-04
JP2023504662A (ja) 2023-02-06
EP4071765A1 (en) 2022-10-12
CN111081351B (zh) 2024-02-20
CN117954055A (zh) 2024-04-30
JP7356590B2 (ja) 2023-10-04
US20220414979A1 (en) 2022-12-29
CN111081351A (zh) 2020-04-28

Similar Documents

Publication Publication Date Title
WO2021109727A1 (zh) 脑功能图谱的绘制方法和系统
JP6947759B2 (ja) 解剖学的対象物を自動的に検出、位置特定、及びセマンティックセグメンテーションするシステム及び方法
CN111127441B (zh) 一种基于图节点嵌入的多模态脑影像抑郁识别方法和系统
CN104700120B (zh) 一种基于自适应熵投影聚类算法的fMRI特征提取及分类方法
Liu et al. Automatic whole heart segmentation using a two-stage u-net framework and an adaptive threshold window
US20220036561A1 (en) Method for image segmentation, method for training image segmentation model
US9042629B2 (en) Image classification based on image segmentation
CN112508902B (zh) 白质高信号分级方法、电子设备及存储介质
KR20210067913A (ko) 학습 모델을 이용한 데이터 처리 방법
WO2023125828A1 (en) Systems and methods for determining feature points
Guru Prasad et al. Glaucoma detection using clustering and segmentation of the optic disc region from retinal fundus images
CN109949288A (zh) 肿瘤类型确定系统、方法及存储介质
CN111402278A (zh) 分割模型训练方法、图像标注方法及相关装置
Guo et al. Statistical shape analysis of brain arterial networks (BAN)
CN114863225A (zh) 图像处理模型训练方法、生成方法、装置、设备及介质
Wang et al. Left ventricle landmark localization and identification in cardiac MRI by deep metric learning-assisted CNN regression
Li et al. BrainK for structural image processing: creating electrical models of the human head
Roy et al. An accurate and robust skull stripping method for 3-D magnetic resonance brain images
CN108597589B (zh) 模型生成方法、目标检测方法及医学成像系统
CN107330948B (zh) 一种基于流行学习算法的fMRI数据二维可视化方法
Pujadas et al. Shape-based normalized cuts using spectral relaxation for biomedical segmentation
KR20220133834A (ko) 학습 모델을 이용한 데이터 처리 방법
Lila et al. Functional random effects modeling of brain shape and connectivity
Pandian et al. Brain Tumor Segmentation Using Deep Learning
ElBedoui ECG Classifiction Based on Federated Unlearning

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20895505

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022533142

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020895505

Country of ref document: EP

Effective date: 20220704