Post-stroke social cognitive function impairment analysis system based on brain structure connection iconography
Through an analysis system based on brain connectivity imaging, combined with multimodal craniocerebral MRI data, a dynamic brain connectivity model is established to predict the social cognitive rehabilitation mechanism of ischemic stroke patients, solving the difficulty in analyzing and predicting poststroke in the existing technology. The problem of social cognitive function damage has realized the disclosure of the internal laws of the social cognitive rehabilitation process and the clinical application value.
Patent Information
- Application Number
- CN202411099511.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is difficult to effectively analyze and predict social cognitive function injury and its recovery mechanism in patients after ischemic stroke. Only studying focal damage is not enough to understand the impact of cerebral structural ischemia on social cognitive function.
Using an analysis system based on brain connectivity imaging, the image acquisition module, poststroke evaluation module, brain structure model establishment module and prediction model learning module are adopted, combined with multimodal cranial MRI (sMRI, DWI, DTI, rs-fMRI) data , establish a dynamic model of brain connections, and conduct deep learning to predict social cognitive rehabilitation.
It reveals the dynamic changes in the brain structure and functional network of patients with ischemic stroke, provides a theoretical basis for the neural mechanism of social cognitive rehabilitation, helps improve patients' quality of life and return to society as soon as possible.
Smart Images

Figure CN118986281A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of post-stroke cognitive function rehabilitation, and in particular is an analysis system for post-stroke social cognitive function damage based on brain connection imaging. Background Art
[0002] Stroke has become the leading cause of death and disability among Chinese residents. According to the 2018 Global Burden of Disease Report published by Lancet magazine, the lifetime prevalence of stroke in the Chinese population is as high as 39.4%, ranking first in the world. Among them, ischemic stroke is the main type of stroke. Most ischemic stroke patients have varying degrees of physical dysfunction; up to 80.97% of patients have varying degrees of cognitive dysfunction and depression symptoms.
[0003] According to existing technologies and research, social cognitive processes are related to structures such as the amygdala, superior temporal gyrus, inferior frontal gyrus, anterior medial frontal cortex, temporal-parietal junction, anterior cingulate cortex, and premotor cortex. However, these structures do not act alone, but form a complex network that collaborates to form social cognitive processes. Brain structural ischemia not only causes focal damage, but also destroys or compensatory changes in brain connections. Therefore, it is not enough to study the impact of focal damage on social cognitive function. Instead, it is necessary to explore imaging mapping and brain network connections and their mechanisms after brain structural ischemic damage. Summary of the invention
[0004] The purpose of the present invention is to provide an analysis system for social cognitive function impairment after stroke based on brain connectivity imaging, to establish the neural mechanism of social cognitive rehabilitation after stroke, and to provide a basis for predicting social cognitive rehabilitation in patients with ischemic stroke.
[0005] In order to achieve the above object, the technical solution of the present invention is as follows:
[0006] An analysis system for social cognitive impairment after stroke based on brain connectivity imaging, including an image acquisition module, a post-stroke evaluation module, a brain structure model building module and a prediction model learning module;
[0007] The image acquisition module is used to collect the patient's cranial MRI examination images and send the cranial MRI examination images to the brain connection dynamic model establishment module;
[0008] The post-stroke evaluation module is used to obtain and input the patient's post-stroke evaluation data, and send the post-stroke evaluation data to the brain connection dynamic model building module;
[0009] The brain structure model building module is used to build a brain connection dynamic model based on the cranial MRI examination image, and the brain connection dynamic model includes a brain base model, a brain structure atrophy area, a lesion area related to social cognition, a microscopic brain white matter damage area, and a functional brain network change area; and the brain connection dynamic model is sent to the prediction model learning module;
[0010] The prediction model learning module is used to establish a connection based on the brain connection dynamic model and post-stroke assessment data, obtain brain volume, multi-parameter injury mapping statistics of different brain regions, FiCD values and DMN parameters corresponding to each cortical partition, and substitute them into the neural network learning model for deep learning to obtain a social cognitive rehabilitation prediction model.
[0011] Furthermore, the brain MRI examination images include 3D-T1W, DWI, rs-fMRI, DTI and T2-FLAIRT2.
[0012] Furthermore, 3D-T1W is used to calculate the quantitative parameters of brain structure volume and brain structure atrophy; DWI is used to segment the infarct area; rs-fMRI is used to calculate the brain functional network connectivity; DTI is used to establish the cortical mapping of white matter fiber connection density; T2-FLAIRT2 is used to segment the white matter highlight area.
[0013] Furthermore, the post-stroke assessment data is used to determine post-stroke social cognitive dysfunction based on theory of mind, emotional cognition, empathy and behavioral norms.
[0014] Furthermore, the theory of mind was assessed using the Cartoon test; emotion cognition was assessed using the Ekman60-Faces test; empathy was assessed using the Faux Pas test; and behavioral norms were assessed using the Hayling sentence test.
[0015] Furthermore, areas of brain structural atrophy were determined based on 3D-T1W; areas of lesions related to social cognition were determined based on DWI; areas of microscopic white matter damage were determined based on DTI; and areas of functional brain network changes were determined based on rs-fMRI.
[0016] Furthermore, the steps for establishing the brain structure atrophy area are as follows: first, the automatic brain volume quantification tool AccuBrain is used to segment the brain structure of 3D-T1W, the volume and atrophy degree in the brain connectivity dynamic model are calculated, and the brain structure atrophy area is marked in the brain connectivity dynamic model.
[0017] Furthermore, the steps of extracting the lesion area related to social cognition are as follows: first, segment the infarct area based on DWI images, segment the white matter highlight area based on T2-FLAIR, and mark the infarct area and the white matter highlight area in the standard space;
[0018] The VLSM method was used to identify brain regions significantly associated with social cognition as seed feature regions. Each infarct area or white matter highlight area was then used to predict the patient's social cognitive status, and the lesion areas that made a significant contribution to the prediction were screened out.
[0019] Furthermore, when the microscopic white matter damage areas were determined using DTI, the density differences of cortical fiber connections between different groups were also calculated based on the FiCD method to measure the relationship between social cognition and white matter atrophy in the cerebral cortex;
[0020] The specific steps of the FiCD method are as follows: first, the basic brain model is divided into several cortical units as regions of interest, and then the FiCD map corresponding to each cortical unit is established based on protein fiber tracking technology. Then, the FiCD map is mapped to the standard space to establish the relationship between social cognition and cerebral cortical white matter atrophy.
[0021] Furthermore, the steps for determining the functional brain network change area are as follows: first, rs-fMRI is used to establish DMN, and the basic brain model is divided into 210 cortices and 36 regions of interest according to the Brainnetome Atlas and FiCD method. The regions of interest are correlated with the seed feature areas in the VLSM method, and the functional brain network change areas that play a predictive role are determined by comparing the differences in brain connection dynamic models between people with progressive disease and people with stable cognitive functions.
[0022] Feature selection was performed using regional lesions in regions of interest; candidate ROIs were damaged in at least four patients and had Pearson correlations with social cognition scores in univariate models; correction was made for total lesion volume by including it as an independent variable (along with the regional volume of candidate ROIs from feature selection).
[0023] The following beneficial effects were achieved by adopting the above scheme: (1) Through multimodal cranial MRI (sMRI, DWI, DTI, rs-fMRI) and post-stroke assessment data, the dynamic evolution data of multimodal neuroimaging parameters were obtained, revealing the internal laws of the evolution of fine structural atrophy of the brain and the dynamic changes of cross-modal joint quantitative parameters of brain structure and functional network connectomics in patients with ischemic stroke during their social cognitive rehabilitation process, providing a theoretical basis for establishing the neural mechanism of social cognitive rehabilitation after stroke.
[0024] (2) The neural mechanism of social cognitive rehabilitation after stroke has great clinical application value and can be applied in clinical practice, helping to improve the quality of life of stroke patients and their early return to society.
[0025] (3) Various MRIs are processed through different methods such as the FiCD method to determine the characteristic parameters of the image data and to derive the relationship between the stroke area and social cognitive changes, thereby helping doctors to further understand the development patterns of this type of disease in clinical practice. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a schematic diagram of an analysis system for social cognitive function impairment after stroke based on brain connection imaging according to an embodiment of the present invention.
[0027] Figure 2 Schematic diagram of the quantitative brain segmentation results of AccuBrain according to an embodiment of the present invention.
[0028] Figure 3 This is a flow chart of a method for jointly analyzing symptom correlations of multiple injuries according to an embodiment of the present invention.
[0029] Figure 4 Schematic diagram of a flow chart of a method for analyzing white matter fiber connection density according to an embodiment of the present invention. DETAILED DESCRIPTION
[0030] The following is further described in detail through specific implementation methods:
[0031] The embodiment is basically as shown in the attached Figures 1 to 4 Shown: An analysis system for social cognitive function impairment after stroke based on brain connectivity imaging, including an image acquisition module, a post-stroke evaluation module, a brain structure model building module and a prediction model learning module.
[0032] The image acquisition module is used to collect the patient's cranial MRI examination images and send the cranial MRI examination images to the brain connection dynamic model establishment module; wherein the cranial MRI examination images include 3D-T1W (three-dimensional T1 weighted image), DWI (diffusion weighted image), rs-fMRI (resting state functional MRI image), DTI (diffusion tensor imaging) and T2-FLAIRT2 (weighted fluid attenuated inversion recovery image), and 3D-T1W is used to calculate the quantitative parameters of brain structure volume and brain structure atrophy; DWI is used to segment the infarct area; rs-fMRI is used to calculate the brain functional network connection; DTI is used to establish a white matter fiber connection density cortical mapping map; T2-FLAIRT2 is used to segment the white matter highlight area (WMH).
[0033] The post-stroke evaluation module is used to obtain and input the patient's post-stroke evaluation data, and send the post-stroke evaluation data to the brain connection dynamic model establishment module; wherein the post-stroke evaluation data is used to determine post-stroke social cognitive dysfunction based on theory of mind, emotional cognition, empathy and behavioral norms.
[0034] Theory of mind was assessed using the Cartoon Test, in which patients describe 12 cartoons of humorous situations. Half of the pictures simply required the subject to describe the attributes of the cartoonist's mental state to understand the humorous situation, while the other half required the subject to understand the meaning of the picture, which was based on the patient's belief understanding of the cartoon character. The total score on the Cartoon Test ranged from 0 to 36, with higher scores indicating better theory of mind.
[0035] Emotional recognition was assessed using the Ekman 60-Faces test, in which patients were asked to identify 60 faces, each showing 10 basic emotions: fear, disgust, anger, happiness, sadness, or surprise. Each face was identified for 3 seconds. The total score ranged from 0 to 60, with higher scores reflecting better emotion recognition ability.
[0036] Empathy was assessed using the Faux Pas test, in which patients were asked to measure their feelings for victims who failed in a planned birth test. The Faux Pas test included 10 items, and patients rated their answers on a scale of 0 to 5, with higher scores indicating more empathy.
[0037] Behavioral regulation was assessed using the Hayling Sentence Test; in the Hayling Sentence Test, patients are asked to think the last word of a total of 30 sentences. For the first 15 sentences, patients were asked to simply complete the sentences, with the goal of measuring their response initiation speed. For the second 15 sentences, patients were asked to complete the sentences using words that were unrelated to the expected word, reflecting the patient's inhibition ability. Thinking time and response inhibition ability were measured on a scale from 1 (severely impaired) to 10 (excellent); higher total scores indicate better behavioral regulation ability. Comprehensive scores were obtained using different test methods to facilitate the determination of the assessment of social cognitive dysfunction after stroke.
[0038] The brain structure model building module is used to establish a brain connection dynamic model based on cranial MRI examination images, and the brain connection dynamic model includes a brain basic model, brain structure atrophy areas, lesion areas related to social cognition, microscopic white matter damage areas and functional brain network change areas; and the brain connection dynamic model is sent to the prediction model learning module.
[0039] The brain structure atrophy area is determined based on 3D-T1W. The steps for establishing the brain structure atrophy area are as follows: First, the automatic brain volume quantification tool AccuBrain is used to segment the 3D-T1W brain structure, calculate the volume and atrophy degree in the brain connection dynamic model, and mark the brain structure atrophy area in the brain connection dynamic model. The automatic brain volume quantification tool AccuBrain and the use of Figure 2 shown.
[0040] The lesion areas related to social cognition were determined based on DWI, and the extraction steps were as follows: first, the infarct area was segmented based on the DWI image, and the white matter highlight area was segmented based on the T2-FLAIR, and the infarct area and the white matter highlight area were marked correspondingly in the standard space; then the VLSM method was used to determine the brain area significantly related to social cognition as the seed feature area, and then each infarct area or white matter highlight area was used to predict the patient's social cognitive status, and the lesion area that made a significant contribution to the prediction was screened out.
[0041] For example, the VLSM method is a fully automatic hypothesis-free analysis method that can point-by-point link the spatial location of lesions in the brain with symptom or behavioral data. It is of great value for the study of the distribution of lesions in groups of functional defects (such as language and motor defects, such as aphasia or motor dysfunction). In this embodiment, the infarct area is segmented from the DWI image using a deep learning method, and the white matter highlight area (related to post-stroke social cognition) is segmented from the T2-FLAIR, and the lesions on different modalities are unified into a standard space through spatial registration. In the standard space, the social cognitive status of the patient is predicted based on the lesion volume of each brain region or the voxel-based lesion status (normal / damaged), and the lesion area that contributes significantly to the prediction is screened out, as shown in the following example. Figure 3 shown.
[0042] The area of microscopic white matter damage is determined based on DTI; the density difference of cortical fiber connections between different groups is also calculated based on the FiCD method (white matter fiber connection density algorithm) to measure the relationship between social cognition and cerebral cortical white matter atrophy; the specific steps of the FiCD method are as follows: first, the brain base model is divided into several cortical units as regions of interest, and then the FiCD map corresponding to each cortical unit is established based on protein fiber tracking technology, and then the FiCD map is mapped to the standard space to establish the relationship between social cognition and cerebral cortical white matter atrophy.
[0043] For example, since the cortex plays a key role in brain function and is connected to several subcortical structures (such as the thalamus and basal ganglia) through dense fiber connection pathways, the FiCD method has good intra- and inter-individual reproducibility by determining the brain's cortical structural connections and can accurately reflect the cortical regions of post-stroke patients, such as Figure 3 shown.
[0044] The validation of the FiCD method is as follows: the intergroup comparison of whole-brain FiCD identified the cortical areas with changes in FiCD values, and there was spatial consistency between the cortical clusters with low FiCD values and the subcortical lesions of the patients. The FiCD method has good intra-individual and inter-individual reproducibility and can accurately reflect the cortical areas of patients after stroke. By applying the FiCD method to patients with small vessel disease (CSVD), the relationship between imaging FiCD and clinical observation was observed, and it was found that compared with CSVD patients with normal cognitive ability, CSVD patients with small vessel disease with MCI had higher lesion loads of WMH and lacunar infarctions, and the corresponding overall FiCD values were also lower (P<0.01). Lacunar infarcts (r=-0.318, P<0.01) and WMH (acute ischemic) (r=-0.400, P<0.01) were significantly correlated with overall atrophy and overall FiCD values, and it was found that white matter microstructural connectivity was disrupted even in cognitively normal CSVD patients, and the degree and intensity of connectivity disruption in the frontal and parietal regions were the basis of cognitive dysfunction in CSVD, so there was a correlation.
[0045] The functional brain network change areas were determined based on rs-fMRI. The steps for determining the functional brain network change areas are as follows: first, rs-fMRI was used to establish the DMN (default state network). The basic brain model was divided into 210 cortices and 36 regions of interest according to the Brainnetome Atlas and FiCD methods. The regions of interest were correlated with the seed feature areas in the VLSM method. By comparing the differences in the dynamic models of brain connections between people with disease progression and people with stable cognitive functions, fMRI imaging markers that predict the progression of social cognitive rehabilitation-related diseases were found to determine the functional brain network change areas that play a predictive role.
[0046] For example, based on the region of interest (ROI) acquisition method, feature selection was performed on these ROIs before their regional lesion volume was entered as an independent variable in the linear SVR model. Candidate ROIs were damaged in at least 4 patients and had Pearson correlation measurements with social cognition scores in the univariate model. Total lesion volume was corrected by including it as an independent variable (as well as the regional volume of the candidate ROIs from feature selection).
[0047] The prediction model learning module is used to establish a connection based on the brain connection dynamic model and post-stroke assessment data, obtain brain volume, multi-parameter injury mapping statistics of different brain regions, FiCD values and DMN parameters corresponding to each cortical partition, and substitute them into the neural network learning model for deep learning to obtain a social cognitive rehabilitation prediction model.
[0048] At the same time, before deep learning, the principal component analysis method is used for feature extraction. Then AdamOptimizer is used as the optimization algorithm and Cross Entropy is used as the loss function to let the model learn to predict the degree of social cognitive rehabilitation. This embodiment will use a small filter in VGG. Its deep learning network modular hierarchical model has high computing efficiency and can also abstract features at different spatial scales. And after the first two layers, batch homogenization is used to ensure that the network can converge and avoid overfitting.
[0049] By using multimodal cranial MRI (sMRI, DWI, DTI, rs-fMRI) and post-stroke assessment data, we obtained dynamic evolution data of multimodal neuroimaging parameters, revealing the inherent laws of the evolution of fine structural atrophy of the brain and the dynamic changes of cross-modal joint quantitative parameters of brain structure and functional network connectomics in patients with ischemic stroke during their social cognitive rehabilitation process, providing a theoretical basis for establishing the neural mechanism of social cognitive rehabilitation after stroke. At the same time, using machine learning methods, establishing the neural mechanism of social cognitive rehabilitation after stroke has great clinical application value, can be used in clinical practice, and help improve the quality of life of stroke patients and return to society as soon as possible.
[0050] The above is only an embodiment of the present invention, and the common knowledge such as the known specific structure and / or characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.
Claims
1. An analysis system for social cognitive impairment after stroke based on brain connectivity imaging. Features: It includes image acquisition module, post-stroke evaluation module, brain structure model building module and prediction model learning module; The image acquisition module is used to collect the patient's cranial MRI examination images and send the cranial MRI examination images to the brain connection dynamic model establishment module; The post-stroke evaluation module is used to obtain and input the patient's post-stroke evaluation data, and send the post-stroke evaluation data to the brain connection dynamic model building module; The brain structure model building module is used to build a brain connection dynamic model for cranial MRI examination images, and the brain connection dynamic model includes a brain base model, a brain structure atrophy area, a lesion area related to social cognition, a microscopic brain white matter damage area, and a functional brain network change area; And send the brain connection dynamic model to the prediction model learning module; The prediction model learning module is used to establish a connection based on the brain connection dynamic model and post-stroke assessment data, obtain brain volume, multi-parameter injury mapping statistics of different brain regions, FiCD values and DMN parameters corresponding to each cortical partition, and substitute them into the neural network learning model for deep learning to obtain a social cognitive rehabilitation prediction model.
2. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 1, Features: The brain MRI examination images include 3D-T1W, DWI, rs-fMRI, DTI and T2-FLAIRT2.
3. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 2, Features: 3D-T1W is used to calculate the quantitative parameters of brain structure volume and brain structure atrophy; DWI is used to segment the infarct area; rs-fMRI is used to calculate the brain functional network connectivity; DTI is used to establish the cortical mapping of white matter fiber connection density; T2-FLAIRT2 is used to segment the white matter highlight area.
4. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 3, Features: The post-stroke assessment data were used to determine post-stroke social cognitive dysfunction based on theory of mind, emotion cognition, empathy, and behavioral norms.
5. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 4, Features: Theory of mind was assessed using the Cartoon test; emotion cognition was assessed using the Ekman60-Faces test; empathy was assessed using the Faux Pas test; and behavioral norms were assessed using the Hayling sentence test.
6. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 5, Features: The areas of brain structural atrophy were determined based on 3D-T1W; the areas of lesions related to social cognition were determined based on DWI; the areas of microscopic white matter damage were determined based on DTI; and the areas of functional brain network changes were determined based on rs-fMRI.
7. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 6, Features: The steps for establishing the brain structure atrophy area are as follows: first, the automatic brain volume quantification tool AccuBrain is used to segment the 3D-T1W brain structure, calculate the volume and atrophy degree in the brain connection dynamic model, and mark the brain structure atrophy area in the brain connection dynamic model.
8. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 7, Features: The steps of extracting lesion areas related to social cognition are as follows: first, segment the infarct area based on DWI images, segment the white matter highlight area based on T2-FLAIR, and mark the infarct area and white matter highlight area in the standard space; The VLSM method was used to identify brain regions significantly associated with social cognition as seed feature regions. Each infarct area or white matter highlight area was then used to predict the patient's social cognitive status, and the lesion areas that made a significant contribution to the prediction were screened out.
9. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 8, Features: When the microscopic white matter damage areas were determined using DTI, the density differences of cortical fiber connections between different groups were also calculated based on the FiCD method to measure the relationship between social cognition and white matter atrophy in the cerebral cortex; The specific steps of the FiCD method are as follows: first, the basic brain model is divided into several cortical units as regions of interest, and then the FiCD map corresponding to each cortical unit is established based on protein fiber tracking technology. Then, the FiCD map is mapped to the standard space to establish the relationship between social cognition and cerebral cortical white matter atrophy.
10. The analysis system for post-stroke social cognitive impairment based on brain connectivity imaging according to claim 9, Features: The steps for determining the functional brain network change area are as follows: first, rs-fMRI is used to establish DMN, and the basic brain model is divided into 210 cortices and 36 regions of interest according to the Brainnetome Atlas and FiCD method. The regions of interest are correlated with the seed feature areas in the VLSM method, and the differences in the brain connection dynamic models between the people with progressive disease and the people with stable cognitive function are compared to determine the functional brain network change areas that play a predictive role.
