Share：
Share this content in WeChat
X
[Chinese] [PDF] 10 1
• Clinical Article •
Cross-attention fusion of static-dynamic graph convolutional networks for Parkinson's disease diagnosis
TANG Yueshan  ZHANG Xiaofei  LIU Xuejun  YU Mengmeng  CHEN Xue  REN Yande 

DOI:10.12015/issn.1674-8034.2025.12.002.


[Abstract] Objective Based on resting-state functional magnetic resonance imaging (rs-fMRI) data, the cross attention mechanism (CAM) combined with static-dynamic graph convolutional network (GCN) technology was utilized to evaluate the classification efficacy of this method for patients with Parkinson's disease (PD), and to explore potential imaging biomarkers, providing a new perspective for the clinical diagnosis and pathological mechanism analysis of PD.Materials and Methods A total of 32 patients with PD were prospectively recruited from the outpatient department of the Affiliated Hospital of Qingdao University, and 30 healthy controls (HC), matched for gender, age and education years were recruited from the community health management center of the Affiliated Hospital of Qingdao University. Resting-state functional magnetic resonance imaging was collected from both groups of subjects. After image preprocessing, static-graph convolutional networks (static-GCN) and dynamic-graph convolutional networks (dynamic-GCN) were constructed for each subject based on the AAL atlas and GCN. Through multi-scale feature extraction and CAM, the complementary information of static-GCN and dynamic-GCN was fused. The performance was evaluated using the accuracy of five-fold cross-validation and the area under the receiver operating characteristic (ROC) curve (AUC). The attention weight coefficients obtained during the process were combined with statistical analysis to identify the abnormal brain regions and static-dynamic functional connections (static-dynamic FC) most related to PD. Two independent sample t-tests were used for inter-group comparisons, and Pearson correlation analysis was used to explore the correlation between the statistically significant static-dynamic FC and clinical scales.Results The method based on CAM combined with static-dynamic graph convolution network has excellent classification performance (with an accuracy of 79.84%, a sensitivity of 80.47%, and a specificity of 78.47%). The ROC curve analysis results show that the AUC for diagnosing PD is 0.814 (95% CI: 0.727 to 0.902, P < 0.001). Five PD high-weight brain regions were identified: the right supplementary motor area, the left posterior cingulate gyrus, the left postcentral gyrus, cerebellar Lobule Ⅵ, and vermis 10. At the same time, two most relevant static-dynamic FC were discovered. Compared with the HC group, the static-dynamic FC in the following two pairs of brain regions was significantly enhanced in the PD group (P < 0.05): (1) the left posterior cingulate gyrus - cerebellar Lobule Ⅵ; (2) the right supplementary motor area - vermis 10 - cerebellar Lobule Ⅵ/the left postcentral gyrus. Moreover, both of these enhanced static-dynamic FC were significantly positively correlated with the UPDRS-Ⅲ score (r = 0.432, P = 0.017; r = 0.420, P = 0.021).Conclusions The method combining CAM with static-dynamic graph convolution networks has excellent diagnostic performance, and has discovered abnormal enhanced patterns of specific static-dynamic FC between the cerebellum and the cerebral cortex in patients with PD, providing a new basis for the objective imaging diagnosis of PD.
[Keywords] Parkinson's disease；magnetic resonance imaging；static functional connectivity；dynamic functional connectivity；graph convolutional network；cross-attention mechanism

TANG Yueshan1   ZHANG Xiaofei2   LIU Xuejun1   YU Mengmeng1   CHEN Xue3   REN Yande1*  

1 Department of Radiology, the Affiliated Hospital of Qingdao University, Qingdao 266000, China

2 College of Electronic and Information Engineering, Shandong University of Science and Technology, Qingdao 266000, China

3 Institute for Digital Medicine and Computer-assisted Surgery in Qingdao University, Qingdao 266000, China

Corresponding author: REN Y D, E-mail: 8198458ryd@qdu.edu.cn

Conflicts of interest   None.

Received  2025-10-01
Accepted  2025-12-05
DOI: 10.12015/issn.1674-8034.2025.12.002
DOI:10.12015/issn.1674-8034.2025.12.002.

[1]
ZEMEL D, GRITTON H, CHEUNG C, et al. Dopamine depletion selectively disrupts interactions between striatal neuron subtypes and LFP oscillations[J/OL]. Cell Rep, 2022, 38(3): 110265 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/35045299/. DOI: 10.1016/j.celrep.2021.110265.
[2]
LIU T T, KONG X R, QIAO J B, et al. Decoding Parkinson's Disease: The interplay of cell death pathways, oxidative stress, and therapeutic innovations[J/OL]. Redox Biol, 2025, 85: 103787 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/40712453/. DOI: 10.1016/j.redox.2025.103787.
[3]
JOZA S, HU M T, JUNG K Y, et al. Progression of clinical markers in prodromal Parkinson's disease and dementia with Lewy bodies: a multicentre study[J]. Brain, 2023, 146(8): 3258-3272. DOI: 10.1093/brain/awad072.
[4]
SHIMA A, INANO R, TABU H, et al. Altered functional connectivity associated with striatal dopamine depletion in Parkinson's disease[J/OL]. Cereb Cortex Commun, 2023, 4(1): tgad004 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/36949935/. DOI: 10.1093/texcom/tgad004.
[5]
LU Y Q, GE Y G, CHEN W Q, et al. The effectiveness of virtual reality for rehabilitation of Parkinson disease: an overview of systematic reviews with meta-analyses[J/OL]. Syst Rev, 2022, 11(1): 50 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/35305686/. DOI: 10.1186/s13643-022-01924-5.
[6]
TOLOSA E, GARRIDO A, SCHOLZ S W, et al. Challenges in the diagnosis of Parkinson's disease[J]. Lancet Neurol, 2021, 20(5): 385-397. DOI: 10.1016/S1474-4422(21)00030-2.
[7]
ZHAO L. Advances in functional magnetic resonance imaging-based brain function mapping: a deep learning perspective[J/OL]. Psychoradiology, 2025, 5: kkaf007 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/40401160/. DOI: 10.1093/psyrad/kkaf007.
[8]
NAKANO Y, NIETHAMMER M, EIDELBERG D. Imaging of disease-related networks in Parkinson's disease[J/OL]. Cold Spring Harb Perspect Med, 2025: a041841 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/40097189/. DOI: 10.1101/cshperspect.a041841.
[9]
WANG S X, XIE Q Y, LIANG Q M, et al. Study on brain age of patients with chronic consciousness disorder based on low frequency amplitude of resting functional magnetic resonance imaging[J]. Chin J Neuromed, 2025, 24(5): 449-455. DOI: 10.3760/cma.j.cn115354-20250124-00046.
[10]
XIN Z Y, ZHOU Q C, QI Y P, et al. Resting-state f MRI study on altered spontaneous brain activity of Parkinson's disease patients with and without rapid eye movement sleep behavior disorder[J]. Chin J Magn Reson Imag, 2024, 15(4): 1-8. DOI: 10.12015/issn.1674-8034.2024.04.001.
[11]
LI Z H, MIAO X X, GE S Y, et al. Study on the changes of brain dynamic function network connection mode in Parkinson's disease patients with postural instability/gait disorder[J]. Chin J Neuromed, 2024, 23(4): 348-356. DOI: 10.3760/cma.j.cn115354-20240329-00195.
[12]
NI J, HUANG Z H, YU C, et al. Comparative convolutional dynamic multi-attention recommendation model[J]. IEEE Trans Neural Netw Learn Syst, 2022, 33(8): 3510-3521. DOI: 10.1109/TNNLS.2021.3053245.
[13]
REN C X, XU G X, DAI D Q, et al. Cross-site prognosis prediction for nasopharyngeal carcinoma from incomplete multi-modal data[J/OL]. Med Image Anal, 2024, 93: 103103 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/38368752/. DOI: 10.1016/j.media.2024.103103.
[14]
LI X M, HUANG Y, LIU M Q, et al. Childhood trauma is linked to abnormal static-dynamic brain topology in adolescents with major depressive disorder[J/OL]. Int J Clin Health Psychol, 2023, 23(4): 100401 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/37584055/. DOI: 10.1016/j.ijchp.2023.100401.
[15]
HUANG J S, WANG M L, JU H R, et al. SD-CNN: a static-dynamic convolutional neural network for functional brain networks[J/OL]. Med Image Anal, 2023, 83: 102679 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/36423466/. DOI: 10.1016/j.media.2022.102679.
[16]
JI J, ZOU A, LIU J, et al. A survey on brain effective connectivity network learning[J]. IEEE Trans Neural Netw Learn Syst, 2023, 34(4): 1879-1899. DOI: 10.1109/tnnls.2021.3106299.
[17]
XIA J, CHAN Y H, GIRISH D, et al. Disentangling shared and unique brain functional changes associated with clinical severity and cognitive phenotypes in schizophrenia via deep learning[J/OL]. Commun Biol, 2025, 8(1): 1215 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/40804294/. DOI: 10.1038/s42003-025-08637-0.
[18]
YANG J, LIU N, XIONG L C, et al. Study on the differential diagnosis model of Alzheimer's disease and frontotemporal dementia with abnormal behavior based on convolutional neural network[J]. Chinese Journal of Psychiatry, 2018, 51(4): 228-234. DOI: 10.3760/cma.j.issn.1006-7884.2018.04.003.
[19]
LI X L, WANG J X, LI A, et al. Imaging research progress in mild cognitive impairment using convolutional neural networks[J]. Chin J Magn Reson Imag, 2021, 12(9): 88-90, 94. DOI: 10.12015/issn.1674-8034.2021.09.022.
[20]
ZHANG X, HE L F, CHEN K, et al. Multi-view graph convolutional network and its applications on neuroimage analysis for Parkinson's disease[J/OL]. AMIA Annu Symp Proc, 2018, 2018: 1147-1156 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/30815157/.
[21]
HATHALIYA J, MODI H, GUPTA R, et al. Stacked model-based classification of Parkinson's disease patients using imaging biomarker data[J/OL]. Biosensors (Basel), 2022, 12(8): 579 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/36004975/. DOI: 10.3390/bios12080579.
[22]
GORE S, DHOLE A, KUMBHAR S, et al. Radiomics for Parkinson's disease classification using advanced texture-based biomarkers[J/OL]. MethodsX, 2023, 11: 102359 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/37791007/. DOI: 10.1016/j.mex.2023.102359.
[23]
WANG M, TAN C L, SHEN Q, et al. Altered functional-structural coupling may predict Parkinson's patient's depression[J]. Brain Struct Funct, 2024, 229(4): 897-907. DOI: 10.1007/s00429-024-02780-w.
[24]
Parkinson's Disease and Movement Disorders Study Group of Neurology Branch of Chinese Medical Association, Parkinson's disease and Movement Disorders Professional Committee of neurophysicians branch of Chinese Medical Doctor Association. Diagnostic criteria of Parkinson's disease in China[J]. Chin J Neurol, 2016, 49(4): 268-271. DOI: 10.3760/cma.j.issn.1006‐7876.2016.04.002.
[25]
LIU Y J, YUAN J Y, TAN C L, et al. Exploring brain asymmetry in early-stage Parkinson's disease through functional and structural MRI[J/OL]. CNS Neurosci Ther, 2024, 30(7): e14874 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/39056398/. DOI: 10.1111/cns.14874.
[26]
HOLTGRAVES T, MCNAMARA P, CAPPAERT K, et al. Linguistic correlates of asymmetric motor symptom severity in Parkinson's Disease[J]. Brain Cogn, 2010, 72(2): 189-196. DOI: 10.1016/j.bandc.2009.08.004.
[27]
VERREYT N, NYS G M S, SANTENS P, et al. Cognitive differences between patients with left-sided and right-sided Parkinson's disease. A review[J]. Neuropsychol Rev, 2011, 21(4): 405-424. DOI: 10.1007/s11065-011-9182-x.
[28]
WANG M, JIANG S M, YUAN Y S, et al. Alterations of functional and structural connectivity of freezing of gait in Parkinson's disease[J]. J Neurol, 2016, 263(8): 1583-1592. DOI: 10.1007/s00415-016-8174-4.
[29]
JI G J, HU P P, LIU T T, et al. Functional connectivity of the corticobasal Ganglia-thalamocortical network in parkinson disease: a systematic review and meta-analysis with cross-validation[J]. Radiology, 2018, 287(3): 973-982. DOI: 10.1148/radiol.2018172183.
[30]
MIJALKOV M, VOLPE G, PEREIRA J B. Directed brain connectivity identifies widespread functional network abnormalities in Parkinson's disease[J]. Cereb Cortex, 2022, 32(3): 593-607. DOI: 10.1093/cercor/bhab237.
[31]
MAFFIA M, DE MICCO R, PETTORINO M, et al. Speech rhythm variation in early-stage Parkinson's disease: a study on different speaking tasks[J/OL]. Front Psychol, 2021, 12: 668291 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/34194369/. DOI: 10.3389/fpsyg.2021.668291.
[32]
DICKSON D W. Parkinson's disease and Parkinsonism: neuropathology[J/OL]. Cold Spring Harb Perspect Med, 2012, 2(8): a009258 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/22908195/. DOI: 10.1101/cshperspect.a009258.
[33]
GONZALEZ-RODRIGUEZ P, ZAMPESE E, SURMEIER D J. Selective neuronal vulnerability in Parkinson's disease[J/OL]. Prog Brain Res, 2020, 252: 61-89 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/32247375/. DOI: 10.1016/bs.pbr.2020.02.005.
[34]
WANG Q, YU M, YAN L, et al. Altered functional connectivity of the primary motor cortex in tremor dominant and postural instability gait difficulty subtypes of early drug-naive Parkinson's disease patients[J/OL]. Front Neurol, 2023, 14: 1151775 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/37251215/. DOI: 10.3389/fneur.2023.1151775.
[35]
TUOVINEN N, SEPPI K, DE PASQUALE F, et al. The reorganization of functional architecture in the early-stages of Parkinson's disease[J/OL]. Parkinsonism Relat Disord, 2018, 50: 61-68 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/29449186/. DOI: 10.1016/j.parkreldis.2018.02.013.
[36]
LENCH D H, EMBRY A, MALAKOUTI N, et al. Reduced automaticity in freezing of gait is associated with elevated cortico-cerebellar connectivity[J]. Brain Imag Behav, 2025, 19(3): 637-646. DOI: 10.1007/s11682-025-00996-w.
[37]
MAITI B, RAWSON K S, TANENBAUM A B, et al. Functional connectivity of vermis correlates with future gait impairments in Parkinson's disease[J]. Mov Disord, 2021, 36(11): 2559-2568. DOI: 10.1002/mds.28684.
[38]
STOODLEY C J, SCHMAHMANN J D. Evidence for topographic organization in the cerebellum of motor control versus cognitive and affective processing[J]. Cortex, 2010, 46(7): 831-844. DOI: 10.1016/j.cortex.2009.11.008.
[39]
LOPEZ A M, TRUJILLO P, HERNANDEZ A B, et al. Structural correlates of the sensorimotor cerebellum in Parkinson's disease and essential tremor[J]. Mov Disord, 2020, 35(7): 1181-1188. DOI: 10.1002/mds.28044.
[40]
LEECH R, SHARP D J. The role of the posterior cingulate cortex in cognition and disease[J]. Brain, 2014, 137(pt 1): 12-32. DOI: 10.1093/brain/awt162.
[41]
MORADI VASTEGANI S, NASROLAHI A, GHADERI S, et al. Mitochondrial dysfunction and Parkinson's disease: pathogenesis and therapeutic strategies[J]. Neurochem Res, 2023, 48(8): 2285-2308. DOI: 10.1007/s11064-023-03904-0.
[42]
PLASCENCIA-VILLA G, PERRY G. Exploring molecular targets for mitochondrial therapies in neurodegenerative diseases[J/OL]. Int J Mol Sci, 2023, 24(15): 12486 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/37569861/. DOI: 10.3390/ijms241512486.
[43]
KOEHLER R C. Regulation of the cerebral circulation during development[J]. Compr Physiol, 2021, 11(4): 2371-2432. DOI: 10.1002/cphy.c200028.
[44]
MANN K, DENY S, GANGULI S, et al. Coupling of activity, metabolism and behaviour across the Drosophila brain[J]. Nature, 2021, 593(7858): 244-248. DOI: 10.1038/s41586-021-03497-0.
[45]
GAO X Y, YANG T, GU Y, et al. Mitochondrial dysfunction in Parkinson's disease: from mechanistic insights to therapy[J/OL]. Front Aging Neurosci, 2022, 14: 885500 [2025-09-30]. https://pubmed.ncbi.nlm.nih.gov/35795234/. DOI: 10.3389/fnagi.2022.885500.
[46]
ZHOU M, ZHENG M, LIANG S Y, et al. Inherent potential of mitochondria-targeted interventions for chronic neurodegenerative diseases[J]. Neural Regen Res, 2026, 21(4): 1409-1427. DOI: 10.4103/NRR.NRR-D-24-01507.

PREV Multimodal MRI combined with canonical correlation analysis reveals the coupling relationship between brain functional abnormalities and clinical symptoms in parents who have lost their only child
NEXT Default mode network functional connectivity changes in adolescent bipolar disorder patients after short-term treatment
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn