CN117593306B - Functional magnetic resonance brain cortex partitioning method and system - Google Patents
Functional magnetic resonance brain cortex partitioning method and system Download PDFInfo
- Publication number
- CN117593306B CN117593306B CN202410078121.1A CN202410078121A CN117593306B CN 117593306 B CN117593306 B CN 117593306B CN 202410078121 A CN202410078121 A CN 202410078121A CN 117593306 B CN117593306 B CN 117593306B
- Authority
- CN
- China
- Prior art keywords
- brain
- partition
- cortex
- individual
- magnetic resonance
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 210000004556 brain Anatomy 0.000 title claims abstract description 95
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000000638 solvent extraction Methods 0.000 title claims abstract description 53
- 238000005192 partition Methods 0.000 claims abstract description 93
- 238000002599 functional magnetic resonance imaging Methods 0.000 claims abstract description 31
- 238000013507 mapping Methods 0.000 claims abstract description 20
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 30
- 230000004927 fusion Effects 0.000 claims description 23
- 238000004364 calculation method Methods 0.000 claims description 21
- 230000001054 cortical effect Effects 0.000 claims description 15
- 210000003710 cerebral cortex Anatomy 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 10
- 210000001175 cerebrospinal fluid Anatomy 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 210000004885 white matter Anatomy 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 210000004884 grey matter Anatomy 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000002759 z-score normalization Methods 0.000 claims description 3
- 230000004886 head movement Effects 0.000 claims 1
- 230000003925 brain function Effects 0.000 abstract description 3
- 238000002610 neuroimaging Methods 0.000 abstract description 2
- 230000006872 improvement Effects 0.000 description 8
- 230000000875 corresponding effect Effects 0.000 description 7
- 238000011156 evaluation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000002490 cerebral effect Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000014644 Brain disease Diseases 0.000 description 1
- 208000016285 Movement disease Diseases 0.000 description 1
- 208000018737 Parkinson disease Diseases 0.000 description 1
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 208000010877 cognitive disease Diseases 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000000337 motor cortex Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0033—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0033—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room
- A61B5/004—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part
- A61B5/0042—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part for the brain
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- General Health & Medical Sciences (AREA)
- Radiology & Medical Imaging (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Heart & Thoracic Surgery (AREA)
- Biomedical Technology (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- High Energy & Nuclear Physics (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Neurology (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention relates to the technical field of neuroimaging, and discloses a functional magnetic resonance brain cortex partitioning method and system, wherein the method comprises the following steps: s1, acquiring magnetic resonance data of a group of tested individuals, and preprocessing fMRI images in the magnetic resonance data of each individual to obtain the whole brain signals of the tested individuals. S2, dividing the subareas by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of the group level. S3, carrying out sub-region combination on the rough partition result based on the hierarchical clustering thought to generate a final partition result at the group level; the final partition result at the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas. S4, guiding individual level partitioning by using the final partitioning result of the group level. The invention reserves the advantages of the boundary mapping method and the clustering method, generates the homogeneous brain function subregion with continuous space, and reserves the hierarchical structure of brain region division under different scales.
Description
Technical Field
The invention relates to the technical field of neuroimaging, in particular to a functional magnetic resonance brain cortex partitioning method and system.
Background
Brain partitioning the human brain into different subregions based on cell type, anatomy and functional characteristics is a goal that has been pursued for over 100 years. The generated brain partition map is not only helpful for fairly comparing different experimental results and promoting the communication among researchers, but also reveals the organization principle of the brain and is helpful for better understanding the brain functions and diseases. At the same time, neuroscientists are able to measure unique topographical (topographic) features of the brain, such as sub-region size and number, etc. Furthermore, as a data reduction strategy, brain partitioning is critical for subsequent data analysis and physiological measurements, such as extracting more uniform brain signals, measuring accurate brain connections/networks, and guiding accurate surgical targets.
Brain-based connection mode partitioning of functional homojunction regions is a common brain partitioning approach, generally employing two strategies: 1) Local partitioning: a typical boundary mapping (boundary mapping) method depicts brain region boundaries by locating the location of sudden spatial changes in local connection patterns by edge detection techniques; 2) Global partition: a typical clustering (clustering) method divides voxels/vertices in global space with similar connected patterns into the same sub-regions. Boundary mapping algorithms can produce spatially contiguous subregions that conform to histological definition, but typically require setting boundary thresholds, but the difference in local boundaries cannot explicitly guarantee functional similarity inside the subregions. The clustering algorithm can generate more uniform and homogeneous subareas due to the clear objective function of the clustering algorithm, but cannot guarantee the continuity of subareas, and the number of subareas needs to be set. The two most commonly used brain partitioning methods described above have been widely used for brain cognition and brain disease studies, including individual level partitioning and group level partitioning based on average functional linkage or two-step (2-level) methods, however, alignment of sub-region tags across individuals and between groups and individuals remains a significant challenge.
In view of the above, the following drawbacks in the prior art need to be overcome: 1) The subarea generated by the boundary mapping method is continuous in space but low in homogeneity, and the boundary threshold value to be set is a continuous variable; the sub-regions generated by the clustering method have higher homogeneity but cannot guarantee the space continuity. 2) The boundary mapping method typically calculates gradients (first order spatial derivatives) of all connected modes, which is time-consuming and unsuitable for partition calculation for a large number of individuals. 3) The common brain partitioning method cannot obtain the sub-region corresponding relation among different partition objects.
Disclosure of Invention
In order to avoid and overcome the technical problems in the prior art, the invention provides a functional magnetic resonance brain cortex partitioning method and a functional magnetic resonance brain cortex partitioning system. The invention divides the brain into successive subregions with high homogeneity by combining boundary mapping and clustering for brain partition division.
In order to achieve the above purpose, the present invention provides the following technical solutions:
The invention discloses a partitioning method of a functional magnetic resonance cerebral cortex, which comprises the following steps of S1-S4.
S1, acquiring magnetic resonance data of a group of tested individuals, and preprocessing fMRI images in the magnetic resonance data of each tested individual to obtain the whole brain signals of each tested individual.
S2, dividing the subareas by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of the group level.
S3, carrying out sub-region combination on the rough partition result based on the hierarchical clustering thought so as to generate a final partition result at the group level; the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas.
S4, guiding individual level partitioning by using the final partitioning result of the group level.
As a further improvement of the above scheme, step S2 includes the following specific steps:
S21, calculating functional connection of the whole brain signals of each tested individual and the signals of the tested individual after dimension reduction, and calculating a whole brain connection mode matrix of each tested individual according to the functional connection.
S22, averaging all brain connection mode matrixes of all tested individuals to obtain a group average brain connection mode matrix.
S23, carrying out principal component decomposition on the group average whole brain connection mode matrix, reserving the first three principal components and calculating respective spatial gradients.
S24, fusing boundaries of the spatial gradients of the first three main components, and converting the fusion result into brain partitions.
S25, performing boundary filling on the fused brain partition according to the label of the nearest neighbor vertex to obtain a rough partition result of a group level.
As a further improvement to the above-described solution,
In step S21, the calculation formula of the functional connection between the brain signal of the tested individual and the signal thereof after the dimension reduction is:
Wherein i=1, 2, …, N; n is the number of cortex vertexes; k=1, 2, …, N'; n' is the number of vertexes of the dimension-reduced cortex; x ik is the functional connection of cortex vertex i and dimension-reduced cortex vertex k; s i is the signal of cortex vertex i; Signals of vertex k of the dimension-reduced cortex; representing the vector inner product;
the calculation formula of the whole brain connection mode matrix of each tested individual is as follows:
Wherein, J=1, 2, …, N for the connected mode of cortex vertices i and j; q k＝(xik+xjk)/2; x jk represents the functional connection of cortical vertex j and reduced dimension cortical vertex k.
As a further improvement of the above scheme, in step S24, the method for fusing boundaries of the first three principal component spatial gradients includes the following specific steps:
s241, calculating the respective boundaries of the spatial gradients of the first three main components by using a watershed algorithm, and further obtaining a corresponding cortex partition map
S242, converting the cortex partition map of the three main components into partition adjacency maps respectivelyAnd a boundary adjacency graph a -m, the calculation formula is as follows:
Wherein, For/>Is the i-th row j column element of (a); /(I)An i-th row j column element of A -m; /(I)Is the partition of the cortex vertex i in the m-th main component; /(I)Partitioning cortex vertex j in the m-th principal component; m=1, 2,3.
S243, calculating to obtain primary fusion results A + and A - according to the two adjacency graphs in the step S242, wherein the calculation formula is as follows:
Wherein, An i-th row j column element of A +; /(I)An i-th row j column element of A -;
S244, further fusing the primary fusion results A + and A - to obtain a secondary fusion result A, wherein the calculation formula is as follows:
Wherein A ij is the ith row and j column elements of A; or represents or, & represents;
S245, obtaining a final fusion adjacent matrix A f according to the secondary fusion result A ij:
Af＝A*Aadj
Wherein, is the Hadamard product of the matrix; a adj is a first order adjacency matrix of cortical vertices.
As a further improvement of the above-described scheme, in step S241, the process of calculating the boundary of the principal component spatial gradient is:
(1) Finding out the local minimum value top point of each principal component space gradient, and respectively giving different sub-region labels.
(2) Starting from the minimum gradient value, vertices smaller than the threshold are assigned to adjacent subregions each time a traversal is performed.
(3) And traversing all gradient values, wherein voxels adjacent to the subinterval are boundaries.
As a further improvement of the above scheme, step S3 includes the following specific steps:
s31, extracting average fMRI signals of each subarea under rough subarea result and adjacency matrix of subareas Wherein the spatially adjacent sub-regions i 'and j' satisfy/>N' is the number of subregions;
S32, calculating the functional connection P of each subarea and the adjacent subarea, and merging the two subareas with the strongest connection; the calculation formula of the functional connection P is as follows:
Where P i′j′ is the functional connection of the sub-region i 'and the sub-region j', And/>Average fMRI signal for subregion i 'and subregion j', respectively;
S33, updating average fMRI signals of subareas Adjacency matrix/>
S34, repeating the steps S32 and S33 until the number of the subareas meets the requirement, obtaining a final subarea result of a group level, and reserving a subarea tree diagram structure of the group.
As a further improvement of the above scheme, step S4 includes the following specific steps:
s41, carrying out principal component decomposition on the whole brain mode connection matrix of each tested individual, and calculating and fusing gradients of the first three principal components of the individual.
S42, acquiring an individual rough partition by adopting a watershed algorithm based on gradients of the first three main components of the individual and a component partition result.
S43, conducting boundary filling on the individual rough partitions, and conducting sub-region merging according to the component partition tree diagram structure to obtain individual horizontal brain partitions corresponding to the component partition results.
As a further improvement of the scheme, the magnetic resonance data of each tested individual also comprises a T1 structural image; in step S1, the following preprocessing is also performed on the T1 structural image of each subject: extracting a brain region from the T1 structural image; carrying out grey matter, white matter and cerebrospinal fluid segmentation on the extracted brain region T1 structural image; cortical reconstruction was performed on the brain region T1 structural image and sampled into the standard cortical space fs_lr32k.
As a further improvement of the above-described arrangement, the method of preprocessing fMRI images in magnetic resonance data of each subject includes the steps of:
and step1, performing time-layer correction on the fMRI image with the repetition time being longer than 1 second.
And 2, performing head motion correction on the fMRI image by adopting 6-degree-of-freedom rigid transformation.
And 3, aligning the fMRI image with the T1 structure and sampling the fMRI image to fs_LR32k space.
And 4, carrying out disturbance regression treatment on the head motion parameters, white matter and cerebrospinal fluid signals in the image.
And 5, sequentially carrying out filtering treatment of 0.01-0.1 HZ and smoothing treatment of 6mm full width half maximum kernel on the image.
And step 6, performing Z-score normalization on the sequence signals so as to obtain the whole brain signals of the tested individuals.
The invention also discloses a partition system of the functional magnetic resonance cerebral cortex, which is used for realizing the partition method of the functional magnetic resonance cerebral cortex; the partition system includes: and the data acquisition module and the data processing module.
The data acquisition module is used for acquiring magnetic resonance data of a group of tested individuals.
The data processing module is used for preprocessing the fMRI image in the magnetic resonance data of each tested individual to obtain the whole brain signal of each tested individual; the data processing module is also used for carrying out subarea division by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of group level; the method is also used for carrying out sub-region combination on the rough partition result based on the hierarchical clustering thought so as to generate a final partition result at the group level; the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas; the data processing module is also configured to direct individual level partitioning using the group level final partitioning results.
Compared with the prior art, the invention has the beneficial effects that:
1. The invention discloses a partition method of a functional magnetic resonance cerebral cortex, which simultaneously reserves the advantages of a boundary mapping method and a clustering method, and can generate a homogeneous cerebral functional subregion with continuous space and reserve the hierarchical structure of cerebral region partition under different scales by firstly acquiring a large number of subregions based on a boundary mapping algorithm and then combining based on a hierarchical clustering thought.
2. The partitioning method disclosed by the invention is used for carrying out boundary mapping based on principal component analysis, so that the running speed can be greatly improved, and a large number of continuous fine subregions (more than 1000 subregions of the hemispherical cortex) in space can be generated. The former three main components are captured to replace tens of thousands of connection attributes, so that the subsequent gradient calculation time can be greatly shortened, and the information redundancy and noise influence can be reduced; the adopted multi-component boundary fusion strategy can acquire more detailed subareas.
3. The partitioning method also provides a group constraint individual partitioning algorithm, adopts a group horizontal partitioning and a tree diagram as priori, simultaneously performs boundary partitioning according to individual gradients, can generate sub-regions with individual variability, and simultaneously reserves the corresponding relation of the sub-regions among individuals.
4. The partition system disclosed by the invention has the same beneficial effects as the partition method, and is not repeated here.
Drawings
Fig. 1 is a flow chart of a method for partitioning a functional magnetic resonance cerebral cortex in example 1 of the present invention.
Fig. 2 is a flowchart of a group-level rough partition acquisition method in embodiment 1 of the present invention.
FIG. 3 is a flow chart of a final partition acquisition method at the group level in embodiment 1 of the present invention.
FIG. 4 is a flow chart of the individual level partitioning guided by the group level final partitioning result in embodiment 1 of the present invention.
Fig. 5 is a schematic diagram of the effect of boundary fusion policy in the partitioning method in embodiment 1 of the present invention during evaluation.
FIG. 6 is a schematic diagram showing the individualizing rough partitioning effect of the partitioning method in embodiment 1 of the present invention at the time of evaluation.
FIG. 7 is a diagram showing the individual final partitioning effect of the partitioning method in example 1 of the present invention at the time of evaluation.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
Referring to fig. 1, the present embodiment provides a method for partitioning a functional magnetic resonance cerebral cortex, which includes the following steps, namely four parts S1-S4.
S1, acquiring magnetic resonance data of a group of tested individuals, and preprocessing fMRI images in the magnetic resonance data of each tested individual to obtain the whole brain signals of each tested individual.
The magnetic resonance data of each tested person further comprises a T1 structural image, and the step S1 further carries out the following preprocessing on the T1 structural image of each tested person: extracting a brain region from the T1 structural image; carrying out grey matter, white matter and cerebrospinal fluid segmentation on the extracted brain region T1 structural image; cortical reconstruction was performed on the brain region T1 structural image and sampled into the standard cortical space fs_lr32k.
The method for preprocessing fMRI images in magnetic resonance data of each subject includes the steps of step 1-step 6.
And step1, performing time-layer correction on the fMRI image with the repetition time being longer than 1 second.
And 2, performing head motion correction on the fMRI image by adopting 6-degree-of-freedom rigid transformation.
And 3, aligning the fMRI image with the T1 structure and sampling the fMRI image to fs_LR32k space.
And 4, carrying out disturbance regression treatment on the head motion parameters, white matter and cerebrospinal fluid signals in the image.
And 5, sequentially carrying out filtering treatment of 0.01-0.1 HZ and smoothing treatment of 6mm full width half maximum kernel on the image.
Step 6, timing sequence signalZ-score normalization/>Thereby obtaining the whole brain signal S of the tested individual; wherein N is the number of vertexes to be partitioned, and T is the number of time points; mean (·) represents average, std (·) represents standard deviation.
S2, carrying out fine subarea division (rough subarea) of a group level by adopting a boundary mapping method based on principal component analysis according to the whole brain signals of each tested individual to obtain a rough subarea result of the group level.
Referring to fig. 2, step S2 specifically includes steps S21 to S25 shown below.
S21, calculating the whole brain signal S of each tested individual and the whole brain signal after dimension reductionFunctional connections (about 8k vertices per hemisphere) from which a full brain connection pattern matrix is calculated for each subject.
The calculation formula of the functional connection of the whole brain signal of the tested individual and the signal after the dimension reduction is as follows:
Wherein i=1, 2, …, N; n is the number of cortex vertexes; k=1, 2, …, N'; n' is the number of vertexes of the dimension-reduced cortex; x ik is the functional connection of cortex vertex i and dimension-reduced cortex vertex k; s i is the signal of cortex vertex i; Signals of vertex k of the dimension-reduced cortex; representing the vector inner product;
the calculation formula of the whole brain connection mode matrix of each tested individual is as follows:
Wherein, J=1, 2, …, N for the connected mode of cortex vertices i and j; q k＝(xik+xjk)/2; x jk represents the functional connection of cortical vertex j and reduced dimension cortical vertex k.
S22, averaging all brain connection mode matrixes of all tested individuals to obtain a group average brain connection mode matrix.
S23, carrying out main component decomposition on the group average whole brain connection mode matrix, and reserving the first three main componentsAnd calculate the respective spatial gradients. To maintain consistency of the symbols, the first feature of each component may be set to be a positive number, and the spatial first derivative of each component calculated as gradient/>Where Nei i is the neighbor vertex direction of vertex i,/>Representing the partial derivative.
S24, fusing boundaries of the spatial gradients of the first three main components, and converting the fusion result into brain partitions. The method for fusing the boundaries of the spatial gradients of the first three principal components comprises the following specific steps, namely S241-S245.
S241, calculating the respective boundaries of the spatial gradients of the first three main components by using a watershed algorithm, and further obtaining a corresponding cortex partition mapThe process for calculating the spatial gradient boundary of the main component is as follows: (1) Finding out the local minimum value top point of each principal component space gradient, and respectively giving different sub-region labels. (2) Starting from the minimum gradient value, vertices smaller than the threshold are assigned to adjacent subregions each time a traversal is performed. (3) And traversing all gradient values, wherein voxels adjacent to the subinterval are boundaries. Partition map of the mth component is/>Where a value of 0 represents a boundary and other values represent different partitions.
S242, converting the cortex partition map of the three main components into partition adjacency maps respectivelyAnd a boundary adjacency graph a -m, the calculation formula is as follows:
Wherein, For/>Is the i-th row j column element of (a); /(I)An i-th row j column element of A -m; /(I)Is the partition of the cortex vertex i in the m-th main component; /(I)Partitioning cortex vertex j in the m-th principal component; m=1, 2,3.
S243, calculating to obtain primary fusion results A + and A - according to the two adjacency graphs in the step S242, wherein the calculation formula is as follows:
Wherein, An i-th row j column element of A +; /(I)Is the i-th row j column element of A -.
S244, further fusing the primary fusion results A + and A - to obtain a secondary fusion result A, wherein the calculation formula is as follows:
Wherein A ij is the ith row and j column elements of A; or represents or, & represents; other wise means else, meaning other conditions independent of the conditions when a ij =1.
S245, obtaining a final fusion adjacent matrix A f according to the secondary fusion result A ij:
Af＝A*Aadj
Wherein, is Hadamard product of matrix; a adj is a first order adjacency matrix of cortical vertices.
S25, performing boundary filling on the fused brain partition according to the label of the nearest neighbor vertex to obtain a rough partition result of a group level.
The boundary mapping algorithm based on principal component analysis, provided by the invention, has the advantages that on one hand, the calculation speed of the spatial gradient is greatly improved, and on the other hand, the noise influence is reduced; the adopted multi-component boundary fusion strategy can acquire more detailed subareas.
S3, carrying out sub-region combination on the rough partition result based on the hierarchical clustering thought so as to generate a final partition result at the group level; the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas.
Referring to fig. 3, step S3 specifically includes steps S31 to S34 as follows.
S31, extracting average fMRI signals of each subarea under rough subarea result and adjacency matrix of subareasWherein the spatially adjacent sub-regions i 'and j' satisfy/>N' is the number of subregions.
S32, calculating the functional connection P of each subarea and the adjacent subarea; the calculation formula of the functional connection P is as follows:
Where P i′j′ is the functional connection of the sub-region i 'and the sub-region j', And/>Average fMRI signal for subregion i 'and subregion j', respectively.
And the two sub-areas i 'j' with the strongest connection are combined, i 'j' takes argmax {i′j′}Pi′j′.
S33, updating average fMRI signals of subareasAdjacency matrix/>
S34, repeating the steps S32 and S33 until the number of the subareas meets the requirement, obtaining a final subarea result of a group level, and reserving a subarea tree diagram structure of the group.
S4, guiding individual level partitioning by using the final partitioning result of the group level.
Referring to fig. 4, step S4 specifically includes steps S41 to S43 as follows.
S41, decomposing main components of the whole brain mode connection matrix of each tested individual, and calculating and fusing gradients of the first three main components of the individualWherein/>And maximally preserving boundary information.
S42, gradient-basedAnd obtaining an individualized rough partition by adopting a watershed algorithm. Wherein the initial point of expansion of the subarea i' is the vertex/>, where the minimum value of the individual gradients in the subarea of the component subareas is located
S43, conducting boundary filling on the individuation rough partition, and conducting sub-region merging according to the component partition tree diagram structure to obtain an individuation final partition corresponding to the component partition result, namely an individuation horizontal brain partition.
And (3) sequentially passing a group of magnetic resonance data through the four parts, and processing according to each part of steps to obtain a group of horizontal brain partitions and corresponding individual horizontal brain partitions.
The brain partition dividing method combining boundary mapping and clustering provided by the invention divides the brain into continuous subareas with high homogeneity. On the basis, the invention provides an individual partition strategy under the guidance of the group partition, so that partition labels are consistent between groups and individuals.
The present embodiment also performs the following evaluation on the above-described partitioning method:
(1) As shown in fig. 5 to 7, fig. 5 shows the boundary fusion result after the step S24 in this embodiment, which is shown as the zoning result of the local region of the motor cortex, and a finer subregion is obtained after fusion; FIG. 6 is a group-level rough partitioning result after step S25 of the present example, dividing the entire cortex into thousands of sub-regions; fig. 7 shows the final partitioning result at the group level after step S34 in this embodiment, where the entire cortex is partitioned into hundreds of spatially consecutive sub-regions of functional homogeneity. It can be seen that the group level brain partition is of good visual quality and the space is completely continuous.
(2) Compared with the partition time of the boundary mapping algorithm (> 8000 seconds), the partition time of the partition method (< 200 seconds) greatly improves the running speed.
(3) The obtained individual partition topological features (sub-region sizes) and homogeneity are correlated with the severity of parkinson's disease, and the results are shown in table 1.
TABLE 1 Individual partition characteristics and Parkinson's movement disorder correlation
| Visual V1 subregion size | Functional homogeneity of cortex | |
| R (relativity) | -0.274 | -0.285 |
| P (significance) | 1.7*10-4 | 9*10-5 |
As can be seen from table 1, the individual partition results obtained in this example can significantly capture disease-related brain function topology changes.
Example 2
The present embodiment provides a partition system of a functional magnetic resonance cerebral cortex for implementing the partition method of the functional magnetic resonance cerebral cortex in embodiment 1; the partition system includes: and the data acquisition module and the data processing module.
The data acquisition module is used for acquiring magnetic resonance data of a group of tested individuals.
The data processing module is used for preprocessing the fMRI image in the magnetic resonance data of each tested individual to obtain the whole brain signal of each tested individual; the data processing module is also used for carrying out subarea division by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of group level; the method is also used for carrying out sub-region combination on the rough partition result based on the hierarchical clustering thought so as to generate a final partition result at the group level; the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas; the data processing module is also configured to direct individual level partitioning using the group level final partitioning results.
The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art, who is within the scope of the present invention, should make equivalent substitutions or modifications according to the technical scheme of the present invention and the inventive concept thereof, and should be covered by the scope of the present invention.
Claims (8)
1. A method for partitioning a functional magnetic resonance cerebral cortex, comprising the steps of:
S1, acquiring magnetic resonance data of a group of tested individuals, and preprocessing fMRI images in the magnetic resonance data of each tested individual to obtain whole brain signals of each tested individual;
s2, dividing subareas by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of a group level;
S3, carrying out sub-region combination on the rough partition result based on a hierarchical clustering idea so as to generate a final partition result at a group level; wherein the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas;
S4, guiding individual level partitioning by utilizing the final partitioning result of the group level;
the step S2 comprises the following specific steps:
S21, calculating functional connection of the whole brain signals of each tested individual and the signals of the tested individual after dimension reduction, and calculating a whole brain connection mode matrix of each tested individual according to the functional connection;
S22, averaging all brain connection mode matrixes of all tested individuals to obtain a group average brain connection mode matrix;
S23, carrying out principal component decomposition on the group average whole brain connection mode matrix, reserving the first three principal components and calculating respective spatial gradients;
s24, fusing boundaries of the spatial gradients of the first three main components, and converting fusion results into brain partitions;
s25, performing boundary filling on the fused brain partition according to the label of the nearest neighbor vertex to obtain a rough partition result of the group level;
Step S3 comprises the following specific steps:
s31, extracting average fMRI signals of each subarea under rough subarea result and adjacency matrix of subareas Wherein the spatially adjacent sub-regions i 'and j' satisfy/>I ', j' =1, 2, …, N "; n' is the number of subregions;
S32, calculating the functional connection P of each subarea and the adjacent subarea, and merging the two subareas with the strongest connection; the calculation formula of the functional connection P is as follows:
Where P i′j′ is the functional connection of the sub-region i 'and the sub-region j', And/>Average fMRI signal for subregion i 'and subregion j', respectively;
S33, updating average fMRI signals of subareas Adjacency matrix/>
S34, repeating the steps S32 and S33 until the number of the subareas meets the requirement, obtaining a final subarea result of a group level, and reserving a subarea tree diagram structure of the group.
2. The method according to claim 1, wherein in step S21, the calculation formula of the functional connection between the whole brain signal of the tested individual and the signal after the dimension reduction is:
Wherein i=1, 2, …, N; n is the number of cortex vertexes; k=1, 2, …, N'; n' is the number of vertexes of the dimension-reduced cortex; x ik is the functional connection of cortex vertex i and dimension-reduced cortex vertex k; s i is the signal of cortex vertex i; signals of vertex k of the dimension-reduced cortex; representing the vector inner product;
the calculation formula of the whole brain connection mode matrix of each tested individual is as follows:
Wherein, J=1, 2, …, N for the connected mode of cortex vertices i and j; q k＝(xik+xjk)/2; x jk represents the functional connection of cortical vertex j and reduced dimension cortical vertex k.
3. A method according to claim 2, wherein in step S24, the method of fusing the boundaries of the spatial gradients of the first three principal components comprises the following steps:
s241, calculating the respective boundaries of the spatial gradients of the first three main components by using a watershed algorithm, and further obtaining a corresponding cortex partition map
S242, converting the cortex partition map of the three main components into partition adjacency maps respectivelyAnd boundary adjacency graph/>The calculation formula is as follows:
Wherein, For/>Is the i-th row j column element of (a); /(I)For/>Is the i-th row j column element of (a); /(I)Is the partition of the cortex vertex i in the m-th main component; /(I)Partitioning cortex vertex j in the m-th principal component; m=1, 2,3;
s243, calculating to obtain primary fusion results A + and A - according to the two adjacency graphs in the step S242, wherein the calculation formula is as follows:
Wherein, An i-th row j column element of A +; /(I)An i-th row j column element of A -;
S244, further fusing the primary fusion results A + and A - to obtain a secondary fusion result A, wherein the calculation formula is as follows:
Wherein A ij is the ith row and j column elements of A; or represents or, & represents;
S245, obtaining a final fusion adjacent matrix A f according to the secondary fusion result A ij:
Af＝A*Aadj
Wherein, is the Hadamard product of the matrix; a adj is a first order adjacency matrix of cortical vertices.
4. A method according to claim 3, wherein in step S241, the process of calculating the boundary of the principal component spatial gradient is:
(1) Finding out the local minimum value top point of each main component space gradient, and respectively giving different sub-region labels;
(2) Starting from the minimum gradient value, each time traversing vertices smaller than the threshold value, assigning to adjacent subregions;
(3) And traversing all gradient values, wherein voxels adjacent to the subinterval are boundaries.
5. A method of partitioning a functional magnetic resonance cerebral cortex according to claim 1, wherein step S4 comprises the specific steps of:
S41, carrying out principal component decomposition on the whole brain mode connection matrix of each tested individual, and calculating and fusing gradients of the first three principal components of the individual;
S42, acquiring an individual rough partition by adopting a watershed algorithm based on gradients of three main components and a component partition result before an individual;
s43, carrying out boundary filling on the individuation rough partition, and carrying out sub-region merging according to the component partition tree diagram structure to obtain an individual horizontal brain partition corresponding to the component partition result.
6. The method of claim 1, wherein the magnetic resonance data of each subject further comprises a T1 structural image; in step S1, the following preprocessing is also performed on the T1 structural image of each subject: extracting a brain region from the T1 structural image; carrying out grey matter, white matter and cerebrospinal fluid segmentation on the extracted brain region T1 structural image; cortical reconstruction was performed on the brain region T1 structural image and sampled into the standard cortical space fs_lr32k.
7. A method of partitioning a functional magnetic resonance cerebral cortex according to claim 6, wherein the method of preprocessing fMRI images in the magnetic resonance data of each subject comprises the steps of:
step1, performing time layer correction on an fMRI image with repetition time longer than 1 second;
step 2, performing head motion correction on the fMRI image by adopting 6-degree-of-freedom rigid transformation;
Step 3, aligning the fMRI image with the T1 structure and sampling fs_LR32k space according to the fMRI image;
step 4, carrying out disturbance regression treatment on the head movement parameters, white matter and cerebrospinal fluid signals in the image;
Step 5, sequentially carrying out filtering treatment of 0.01-0.1 HZ and smoothing treatment of 6mm full width half maximum kernel on the image;
And step 6, performing Z-score normalization on the sequence signals so as to obtain the whole brain signals of the tested individuals.
8. A compartmentalization system of functional magnetic resonance cerebral cortex, characterized by a compartmentalization method for achieving a functional magnetic resonance cerebral cortex according to any one of claims 1 to 7; the partition system includes:
the data acquisition module is used for acquiring magnetic resonance data of a group of tested individuals;
The data processing module is used for preprocessing the fMRI image in the magnetic resonance data of each tested individual to obtain the whole brain signal of each tested individual; the data processing module is also used for carrying out subarea division by adopting a boundary mapping method according to the whole brain signals of each tested individual to obtain a rough subarea result of group level; the method is also used for carrying out sub-region combination on the rough partition result based on hierarchical clustering thought so as to generate a final partition result at a group level; wherein the final partition result of the group level comprises homogeneous functional subareas which are continuous in space, and hierarchical structures of brain area division under different scales are reserved among the subareas; the data processing module is also configured to direct individual level partitioning using the set of level final partitioning results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078121.1A CN117593306B (en) | 2024-01-19 | 2024-01-19 | Functional magnetic resonance brain cortex partitioning method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078121.1A CN117593306B (en) | 2024-01-19 | 2024-01-19 | Functional magnetic resonance brain cortex partitioning method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117593306A CN117593306A (en) | 2024-02-23 |
| CN117593306B true CN117593306B (en) | 2024-05-03 |
Family
ID=89912003
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410078121.1A Active CN117593306B (en) | 2024-01-19 | 2024-01-19 | Functional magnetic resonance brain cortex partitioning method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117593306B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106023194A (en) * | 2016-05-18 | 2016-10-12 | 西安交通大学 | Amygdaloid nucleus spectral clustering segmentation method based on resting state function connection |
| CN107392907A (en) * | 2017-09-01 | 2017-11-24 | 上海理工大学 | Parahippocampal gyrus function division method based on tranquillization state FMRI |
| CN111161226A (en) * | 2019-12-20 | 2020-05-15 | 西北工业大学 | Method for uniformly segmenting cerebral cortex surface based on spectral clustering algorithm |
| CN112614126A (en) * | 2020-12-31 | 2021-04-06 | 中国科学院自动化研究所 | Magnetic resonance image brain region dividing method, system and device based on machine learning |
| CN116401889A (en) * | 2023-04-21 | 2023-07-07 | 电子科技大学 | Cerebellum partitioning method based on functional connection optimization and spectral clustering |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE202017007366U1 (en) * | 2016-12-16 | 2021-02-17 | The Brigham And Women's Hospital, Inc. | Sensor array for recording a relative protein concentration |
| US11705226B2 (en) * | 2019-09-19 | 2023-07-18 | Tempus Labs, Inc. | Data based cancer research and treatment systems and methods |
-
2024
- 2024-01-19 CN CN202410078121.1A patent/CN117593306B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106023194A (en) * | 2016-05-18 | 2016-10-12 | 西安交通大学 | Amygdaloid nucleus spectral clustering segmentation method based on resting state function connection |
| CN107392907A (en) * | 2017-09-01 | 2017-11-24 | 上海理工大学 | Parahippocampal gyrus function division method based on tranquillization state FMRI |
| CN111161226A (en) * | 2019-12-20 | 2020-05-15 | 西北工业大学 | Method for uniformly segmenting cerebral cortex surface based on spectral clustering algorithm |
| CN112614126A (en) * | 2020-12-31 | 2021-04-06 | 中国科学院自动化研究所 | Magnetic resonance image brain region dividing method, system and device based on machine learning |
| CN116401889A (en) * | 2023-04-21 | 2023-07-07 | 电子科技大学 | Cerebellum partitioning method based on functional connection optimization and spectral clustering |
Non-Patent Citations (3)
| Title |
|---|
| Imaging-based parcellations of the human brain;Simon B. Eickhoff.et.;《Nature Reviews Neuroscience》;20181009;第19卷;672–686 * |
| 基于边界映射的大脑功能分区研究;张朕;《中国优秀硕士学位论文全文数据库医药卫生科技辑》;20230215(第2期);E060-462 * |
| 基于静息态功能磁共振成像的脑区划分研究及其在脑疾病中的应用;李煜;《万方》;20230427;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117593306A (en) | 2024-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106340021B (en) | Blood vessel extraction method | |
| CN107242873B (en) | Brain network construction method based on functional magnetic resonance psychophysiological interaction | |
| Haak et al. | Objective analysis of the topological organization of the human cortical visual connectome suggests three visual pathways | |
| Cover et al. | Computational methods for corpus callosum segmentation on MRI: A systematic literature review | |
| CN107392907A (en) | Parahippocampal gyrus function division method based on tranquillization state FMRI | |
| CN107203997B (en) | Left and right semi-brains segmentation method | |
| EP3605464A1 (en) | Quantitative calculation method for level of brain atrophy based on automatic segmentation | |
| Bayrak et al. | The impact of ischemic stroke on connectivity gradients | |
| CN113947157B (en) | Dynamic brain effect connection network generation method based on hierarchical clustering and structural equation model | |
| US7764816B2 (en) | Method for analyzing medical image data | |
| CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
| CN113610808A (en) | Individual brain atlas individualization method, system and equipment based on individual brain connection atlas | |
| Zhu et al. | Discovering dense and consistent landmarks in the brain | |
| CN107093190A (en) | A kind of Medical Image Registration Algorithm based on multichannel chromatogram tag fusion | |
| CN114708263B (en) | Individual brain functional region positioning method, device, equipment and storage medium | |
| Jiao et al. | Module partitioning for multilayer brain functional network using weighted clustering ensemble | |
| CN117593306B (en) | Functional magnetic resonance brain cortex partitioning method and system | |
| Ramana | Alzheimer disease detection and classification on magnetic resonance imaging (MRI) brain images using improved expectation maximization (IEM) and convolutional neural network (CNN) | |
| Van Leemput et al. | Model-based segmentation of hippocampal subfields in ultra-high resolution in vivo MRI | |
| Kozlov et al. | Selection of functionally homogeneous brain regions based on correlation-clustering analysis | |
| CN116401889A (en) | Cerebellum partitioning method based on functional connection optimization and spectral clustering | |
| CN109994204A (en) | A kind of cerebral function network key node searching method | |
| CN115841472A (en) | Method, device, equipment and storage medium for identifying high-density characteristics of middle cerebral artery | |
| CN115496743A (en) | Cerebrovascular lesion segmentation method, cerebrovascular lesion segmentation device, storage medium and electronic device | |
| US20210307610A1 (en) | Methods and systems for precise quantification of human sensory cortical areas |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |