ChangesoftheLocalBrainNetworkPropertyRegionsinAlzheimersDiseaseBasedonGraphTheory
.pdf
School
Ashford University**We aren't endorsed by this school
Course
PSY 625
Subject
Biology
Date
Dec 18, 2024
Pages
5
Uploaded by ColonelLightning14848
Changes of the Local Brain Network Property Regions in Alzheimer's Disease Based on Graph Theory Bing Zhu School of Computer Science and Technology Changchun University of Science and Technology Changchun, China baulina@163.com Chunjie Guo*Department of Radiology The First Hospital of Jilin University Changchun, China *Corresponding author: guocj@jlu.edu.cn Qi Li*School of Computer Science and Technology Changchun University of Science and Technology Changchun, China *Corresponding author: liqi@cust.edu.cn Yu Yang Department of Neurology and Neuroscience Center The First Hospital of Jilin University Changchun, China yang_yu@jlu.edu.cn Jinglong Wu Research Center for Medical Artificial Intelligence Shenzhen Institute of Advanced Technology, Chinese Academy of Science Shenzhen, China jl.wu@siat.ac.cn Zhilin Zhang Research Center for Medical Artificial Intelligence Shenzhen Institute of Advanced Technology, Chinese Academy of Science Shenzhen, China zhangzhilin@siat.ac.cn Abstract—Alzheimer's disease (AD) is an irreversible degenerative disease of the nervous system. Early diagnosis of Alzheimer's disease is the key to treatment. This study adopted clinical neuropsychological examinations and functional Magnetic Resonance Imaging brain network properties constructed by graph theory. The results of clinical neuropsychological examinations and five local network topological properties with significant differences among the AD, Mild Cognitive Impairment and normal control groups were analyzed. Twenty-seven local brain network property regions were extracted in the anatomical automatic marking (AAL) 90 template. In short, the changes in local brain network property regions can be used as important markers to distinguish AD patients. It also plays a reference effect on AD's early diagnosis. Keywords-Alzheimer's disease; clinical neuropsychological examination; functional Magnetic Resonance Imaging; brain network; graph theory I.INTRODUCTIONAlzheimer's disease (AD) is an irreversible, age-related neurodegenerative disease marked by progressive memory loss and cognitive decline [1]. According to AD, it is the fifth leading cause of death worldwide. The increase in nursing demand has caused patients and family members to withstand great pain and economic pressure [2]. As an important indicator for evaluating the progress of AD, the clinical neuropsychological examination is widely used in clinical practice. The neuropsychological assessment of AD is mainly aimed at cognitive dysfunction, social and daily abilities, and mental behavior symptoms. To evaluate the changes in potential mechanisms of the AD brain more objectively, imaging examination has also become an indispensable part. Resting-state functional Magnetic Resonance Imaging (fMRI) is one of the most widely used methods to study neurological and psychiatric diseases, and its main advantage is that it requires minimal compliance from participants. Participants only need to stay awake, without any additional coordination. Therefore, it is easy to implement for AD patients. For fMRI, graph theory analysis allows quantification of the degree of integration and separation of brain network topological properties. Graph theory can not only display different functional connectivity properties, but also combine information from the entire network. The nodal properties can represent the local information of the local region [3]. Combining the local properties of the nodes in graph theory with the resting-state fMRI can better observe the changes in the local brain regions, and obtain the functional abnormalities in specific brain regions of AD. Therefore, we calculated the clinical neuropsychological examinations and local network topological properties of resting-state fMRI results of AD, Mild Cognitive Impairment (MCI) and normal controls (NC) participants. By analyzing the local network topological properties with significant differences among the three groups, the specific brain function change regions of AD patients were obtained. To provide ideas for the pathogenesis, early diagnosis and intervention treatment of AD patients. 2022 16th ICME International Conference on Complex Medical Engineering (CME)978-1-6654-9699-5/22/$31.00©2022 IEEE1512022 16th ICME International Conference on Complex Medical Engineering (CME) | 978-1-6654-9699-5/22/$31.00 ©2022 IEEE | DOI: 10.1109/CME55444.2022.10063306Authorized licensed use limited to: UAGC. Downloaded on December 19,2024 at 04:37:28 UTC from IEEE Xplore. Restrictions apply.
II.MATERIALS AND METHODSA.Clinical neuropsychological examinations Forty-nine patients (23 AD and 26 MCI) and 18 NC were recruited for this experiment. With the approval of the Institutional Research Review Committee, the participants or their guardians have fully explained the purpose of the experiment and the basic procedures of fMRI scanning, and signed the Ethics Committee of The First Hospital of Jilin University. At present, data cannot be disclosed. All participants received six clinical neuropsychological examinations. Three AD participants could not complete the examinations, and the results were not included. These examinations were conducted by professional appraisers in neurological research on each participant. The neuropsychological examinations include the Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment (MoCA), Memory and Executive Screening (MES), Clinical Dementia Rating (CDR), Activity of Daily Living Scale (ADL), Verbal Fluency Test (VFT). B.fMRI data acquisition and preprocessing All participants underwent a 3.0T Philips resting-state fMRI scanner at The First Hospital of Jilin University. The specific parameters were as follows: TR = 2,000 ms, TE = 30 ms, flip angle = 90°, slice thickness = 3.5 mm, FOV = 224×224×138 mm3, matrix = 64×63, slice thickness = 3.5 mm, and 33 slices. All participants remained awake during the scan. Considering that the data contains the patient, the head movement of all the participants was less than 3 mm or the rotation was less than 3.0°. Data conforming to the above standards were preprocessed in Statistical Parametric Mapping (SPM12). After removing the 10 time points, participants’ data were corrected for slice timing. The images were realigned to the first volume for head-motion correction, normalized to standard anatomical space (3 × 3 × 3 mm3), and smoothed with an 8 mm FWHM. C.Functional network construction Based on the anatomical automatic marking (AAL) brain template, the whole brain of each participant was divided into 90 regions of interest. At the same time, extracted a series from 90 brain regions, and calculated the Pearson correlation coefficients to estimate the connectivity. Correspondingly, a matrix of 90 × 90 was generated for each participant. Table 1 was the abbreviation in some AAL90 brain regions. TABLE I. PART OF THE AAL90BRAIN REGIONS ABBREVIATIONAbbreviation Region name ACG Anterior cingulate and paracingulate gyri AMYG Amygdala ANG Angular gyrus CAL Calcarine fissure and surrounding cortex CAU Caudate nucleus HIP Hippocampus IFGoperc Inferior frontal gyrus, opercular part INS Insula ITG Inferior temporal gyrus LING Lingual gyrus MCG Median cingulate and paracingulate gyri MTG Middle temporal gyrus OLF Olfactory cortex PHG Parahippocampal gyrus PreCG Precentral gyrus PUT Lenticular nucleus, putamen REC Gyrus rectus SPG Superior parietal gyrus TPOmid Temporal pole: middle temporal gyrus D.Local Network Analysis We used the Graph Theoretical Network Analysis Toolbox (GRETNA) to explore the local topological properties of each participant based on graph theory. For each AAL90 region, five local properties were calculated. They included: (1) Node local efficiency refers to the efficiency of information propagation over the node's direct neighbors. (2) Node efficiency indicates the average difficulty of the node to other nodes in the network. (3) Node clustering coefficient is used to describe the degree of network clustering and measures the likelihood that the neighbors of a node are neighbors to each other. (4) Degree describes the centrality of a node in the network, and the node with the largest degree is considered to be the core node of the network. (5) Betweenness centrality is the number of times the shortest path connected between the remaining pairs of nodes in the network passes through that node. E.Statistical analysis The basic information and clinical neuropsychological examination scores of the three groups were analyzed using a one-way analysis of variance (ANOVA). The within-group sex ratio was calculated by the chi-square test. The five local property differences in the three groups of participants were also computed by one-way ANOVA. The value of p< 0.05 was considered to be statistically significantly different. III.RESULTSA.Demographical and clinical characteristics There were no significant differences in the age and education level of the three groups. The scores of MMSE (p< 0.001), MoCA (p= 0.001), MES (p< 0.001), CDR (p< 0.001), ADL (p< 0.001) and VFT (p< 0.001) measured by clinical neuropsychological examinations were significantly different among the three groups. In the post hoc multiple comparisons results, except for MMSE, MoCA and ADL, there was no significant difference between MCI and NC groups. There were significant differences between the other two groups' comparisons. The obtained results were presented in Table 2. 152Authorized licensed use limited to: UAGC. Downloaded on December 19,2024 at 04:37:28 UTC from IEEE Xplore. Restrictions apply.
TABLE II. DEMOGRAPHIC AND NEUROPSYCHOLOGICAL EXAMINATIONS FOR STUDY PARTICIPANTSAD(n=23) MCI(n=26) NC(n=18) F/χ² p Post hoc multiple comparisons AD vs. MCI AD vs. NC MCI vs. NC Age 66.69±7.03 64.81 ± 7.02 63.44 ± 7.94 0.87 0.425 NS NS NS Gender(M/F) 8/15 8/18 4/14 0.68* 0.678 NS NS NS Education 10.95±4.18 10.11±4.28 10.94±3.78 0.26 0.773 NS NS NS MMSE 18.76±5.82 25.3±2.42 27.82±2.18 24.22 p<0.001 p<0.001 p<0.001 0.003 MoCA 13.47±5.21 23.84±15.11 23.29±3.23 7.65 0.001 0.002 0.009 NS MES 45.47±17.94 75.65±19.38 85.70±13.89 31.08 p<0.001 p<0.001 p<0.001 NS CDR 1.59±0.78 0.53±0.24 0.20±0.25 32.29 p<0.001 p<0.001 p<0.001 p<0.001 ADL 30.77±9.44 21.76±2.15 20.58±1.22 14.20 p<0.0010.001p<0.001NSVFT 21.1±14.84 38.57±11.79 50.41±12.93 23.84 p<0.001 p<0.001 p<0.001 0.016 *, chi-square test. All others are one-way ANOVA methods. p< 0.05. NS, not significant. B.Local network topological properties Fig. 1 shows the location of five local brain regions with significant differences in the cerebral cortex. The volume of each region ball represents the local property value of the AD group. There were significant differences in 27 brain regions, including PreCG.R, bilateral IFGoperc, OLF.R, REC.L, INS.L, bilateral ACG, bilateral MCG, HIP.L, PHG.R, AMYG.L, CAL.L, bilateral LING, SPG.R, ANG.R, CAU.R, bilateral PUT, bilateral MTG, bilateral TPOmid, and bilateral ITG. Table 3 represented the results of brain regions with significant differences among the five local brain network properties in AAL90. Among them, IFGoperc.L, IFGoperc.R and TPOmid.R presented a more obvious significant difference in statistics of local network topological properties. Figure 1. Location of local properties with significant differences in the cerebral cortex: one-way ANOVA was used to calculate the regions with significant differences in local network topological properties, and different color markers were used, p< 0.05. The size ratio of the balls in each region was calculated based on the local property value of AD patients. TABLE III. LOCAL NETWORK TOPOLOGICAL PROPERTIES WITH SIGNIFICANT DIFFERENCES IN THE FIVE LOCAL PROPERTIESRegions Node local efficiency Node efficiency Node clustering coefficient Degree Betweenness PreCG.R √√IFGoperc.L √√√153Authorized licensed use limited to: UAGC. Downloaded on December 19,2024 at 04:37:28 UTC from IEEE Xplore. Restrictions apply.
IFGoperc.R √√√√OLF.R √REC.L √√INS.L √√√ACG.L √ACG.R √MCG.L √√MCG.R √√HIP.L √PHG.R √AMYG.L √CAL.L √√√LING.L √LING.R √√SPG.R √ANG.R √√CAU.R √PUT.L √√PUT.R √MTG.L √MTG.R √TPOmid.L √√TPOmid.R √√√√ITG.L √√ITG.R √One-way ANOVA, p< 0.05. IV.DISCUSSIONA.Clinical neuropsychological examinations We have chosen the examinations commonly used in clinical practice and a few directions of interest for research. The results showed that there were significant differences among the three groups in six examinations. The results of the ANOVA post hoc multiple comparisons revealed that the scores of the AD group were significantly different from MCI and NC groups. It indicated that AD patients had obvious decline in cognition, attention, executive ability, memory, language, abstract thinking, visual-spatial structure, daily activities, etc. B.Changes in local brain regions Through the calculation of the five local properties, there were significant differences in 27 local regions. The changes in these regions have caused abnormal function. The IFGoperc is involved in memory maintenance and is involved in the integration of information, and the encoding and decoding of situational memories. The HIP is responsible for encoding long-term memory, and the PHG is involved in memory creation and the recall of visual scenes. The INS is involved in emotional processing and body state awareness. It is also thought to guide the regulation of brain circulation, thus helping in memory maintenance. There were significant differences exist in the MTG, TPOmid, and ITG of the temporal lobe. These regions involve higher-level cognitive functions, such as episodic memory, attention, motivation, and self-awareness. For olfaction, the OLF is a part of the olfactory system. The cingulate gyrus is thought to be involved in working memory during odor discrimination, and also plays an important role in cognitive and emotional regulation. The CAL and LING located in the occipital lobe are closely related to visual recognition and have an important role in situational memory. The SPG is involved in sensory discrimination. The ANG is associated with processing language, mathematics and other cognitive skills. The PreCG, located in the frontal lobe, is in charge of the movement of the contralateral half of the body. The CAU is part of the striatum of the basal ganglia and is important for both cognitive and motor functions. The PUT is traditionally associated with reinforcement learning and motor control, including speech intelligibility. In many 154Authorized licensed use limited to: UAGC. Downloaded on December 19,2024 at 04:37:28 UTC from IEEE Xplore. Restrictions apply.
neuropsychiatric and neurodegenerative diseases, the PUT is characterized by motor deficits, behavioral impulses, and cognitive deficits [4, 5]. V.CONCLUSIONSIn this study, we found that AD was abnormal in various clinical neuropsychological examinations. The changes in local brain network properties in graph theory were of great significance to the study of specific functional regions, and provide a basis for the pathogenesis and disease progression of AD. ACKNOWLEDGMENTThis study was supported by the Jilin Scientific and Technological Development Program (No. 20200802004GH), the National Natural Science Foundation of China (No. 81600923), the Natural Science Foundation of Jilin Province (No. 20210101273JC), Bethune Project of Jilin University (No. 2020B47), and the Science and Technology Achievement Transformation Fund of the First Hospital of Jilin University (No. JDYY2021-A0010). REFERENCES[1]L. Serra et al., "Behavioral psychological symptoms of dementia and functional connectivity changes: a network-based study," Neurobiol Aging, vol. 94, pp. 196-206, Oct 2020. [2]G. Livingston et al., "Dementia prevention, intervention, and care: 2020 report of the Lancet Commission," The Lancet, vol. 396, no. 10248, pp. 413-446, 2020. [3]L. Kucikova et al., "Resting-state brain connectivity in healthy young and middle-aged adults at risk of progressive Alzheimer's disease," Neurosci Biobehav Rev, vol. 129, pp. 142-153, Oct 2021. [4]S. Y. Lin et al., "Multiparametric graph theoretical analysis reveals altered structural and functional network topology in Alzheimer's disease," Neuroimage Clin, vol. 22, p. 101680, 2019. [5]T. Li et al., "APOE ε4 and cognitive reserve effects on the functional network in the Alzheimer's disease spectrum," (in eng), Brain Imaging Behav, vol. 15, no. 2, pp. 758-771, Apr 2021. 155Authorized licensed use limited to: UAGC. Downloaded on December 19,2024 at 04:37:28 UTC from IEEE Xplore. Restrictions apply.