Share：
Share this content in WeChat
X
[Chinese] [PDF] 8 1
• Clinical Article •
Research on diagnostic and staging models for Parkinson's disease patients using T1 images based on interpretable machine learning
HAO Lu  ZHU Minghui  ZHU Yutong  KALIBUNUER·Mahemuti   WANG Yunling  HANJIAERBIEKE·Kukun   GUAN Yangtai 

DOI:10.12015/issn.1674-8034.2025.12.004.


[Abstract] Objective To construct a three-classification model based on T1-weighted imaging (T1WI) radiomics for the diagnosis and early, middle and late stage classification of Parkinson's disease (PD), and to explore the diagnostic value of different nuclear features.Materials and Methods A prospective analysis was conducted on the T1WI data of 146 patients, including 86 cases from the Second Affiliated Hospital of Xinjiang Medical University (development cohort), with 26 cases in the normal group, 35 cases in the early PD group, and 25 cases in the middle and late PD group; and 60 cases from Affiliated First Hospital of Xinjiang Medical University (external validation cohort), with 18 cases in the normal group, 22 cases in the early PD group, and 20 cases in the middle and late PD group. Six nuclei, namely the caudate nucleus (CN), putamen (PUT), globus pallidus (GP), red nucleus (RN), substantia nigra (SN), and nucleus accumbens (NAC), were segmented from all data, and 1688 radiomics features (including first-order statistics, shape, texture, and filter features) were extracted. Logistic regression (LR) algorithm was used to construct 6 single-nucleus models and 1 combined model. Key features were selected through variance threshold method, univariate selection method, and least absolute shrinkage and selection operator (LASSO) algorithm. Model performance and interpretability were analyzed using receiver operator characteristic (ROC) curve, confusion matrix, and SHapley Additive exPlanations (SHAP) value.Results The macro area under the curve (AUC) of the combined model in the training set, internal validation set, and external validation set were 0.93 (95% CI: 0.87 to 1.00), 0.88 (95% CI: 0.69 to 1.00), and 0.84 (95% CI: 0.75 to 0.92), respectively, which were significantly better than most single-nucleus models. Among the 21 key features selected, 9 were related to GP (e.g., wavelet-HLL_glcm_Correlation_GP), and the absolute value of the coefficient of the lbp-3D-k_firstorder_Minimum_RN feature of RN was the largest (0.12). SHAP analysis showed that the classification of the normal group relied on the texture symmetry of GP and SN, early PD focused on the local structural changes of PUT and GP, and middle and late PD was characterized by signal abnormalities of RN and NAc.Conclusions The multi-nucleus combined model based on T1WI has high performance in the diagnosis and staging of PD. Nucleus-specific features can reflect the spatial heterogeneity of the pathological process of PD, with GP and SN playing a core role in diagnosis, providing an imaging basis for precise clinical staging.
[Keywords] Parkinson's disease；magnetic resonance imaging；T1-weighted imaging；explainable machine learning；nucleus features；diagnostic staging

HAO Lu1   ZHU Minghui1   ZHU Yutong1   KALIBUNUER·Mahemuti 1   WANG Yunling2   HANJIAERBIEKE·Kukun 2   GUAN Yangtai3*  

1 Department of Medical Imaging Center, the Second Affiliated Hospital of Xinjiang Medical University, Urumqi 830000, China

2 Department of Radiology, Affiliated First Hospital of Xinjiang Medical University, Urumqi 830000, China

3 Department of Neurology, The Affiliated Renji Hospital of Shanghai Jiao Tong University, Shanghai 200000, China

Corresponding author: GUAN Y T, E-mail: yangtaiguan@sina.com

Conflicts of interest   None.

Received  2025-07-21
Accepted  2025-10-21
DOI: 10.12015/issn.1674-8034.2025.12.004
DOI:10.12015/issn.1674-8034.2025.12.004.

[1]
ZHANG S, GAO J, JIANG L G. Current Epidemiological Research Status on the Prevalence and Associated Factors of Parkinson's Disease[J]. Journal of Jilin Medical University, 2021, 42(6): 437-439. DOI: 10.13845/j.cnki.issn1673-2995.2021.06.016.
[2]
LIN K J, CHEN S D, LIN K L, et al. Iron Brain Menace: The Involvement of Ferroptosis in Parkinson Disease[J/OL]. Cells, 2022, 11(23): 3829 [2025-07-21]. https://pmc.ncbi.nlm.nih.gov/articles/PMC9735800/. DOI: 10.3390/cells11233829.
[3]
JIAN Y J, PENG J X, WANG W, et al. Prediction of cognitive decline in Parkinson's disease based on MRI radiomics and clinical features: A multicenter study[J/OL]. CNS Neurosci Ther, 2024, 30(6): e14789 [2025-07-21]. https://doi.org/10.1111/CNS.14789. DOI: 10.1111/CNS.14789.
[4]
JIANG H, DU Y, Mok G S P, et al. Radiomics incorporating deep features for predicting Parkinson's disease in 123I-Ioflupane SPECT[J/OL]. EJNMMI Phys, 2024, 11(1): 60 [2025-07-21]. https://doi.org/10.1186/S40658-024-00651-1. DOI: 10.1186/S40658-024-00651-1.
[5]
KURIBARA T, ENATSU R, KITAGAWA M, et al. Neuroimaging and neurophysiological evaluation of severity of Parkinson's disease[J]. J Clin Neurosci, 2020, 74: 135-140. DOI: 10.1016/j.jocn.2020.02.006.
[6]
HUANG Y H, YANG M L, Li Y Z, et al. Differentiating idiopathic Parkinson's disease from multiple system atrophy-P using brain MRI-based radiomics: a multicenter study[J/OL]. Ther Adv Neurol Disord, 2025, 18: 17562864251318865 [2025-07-21]. https://doi.org/10.1177/17562864251318865. DOI: 10.1177/17562864251318865.
[7]
WANG Q H, ZENG W, DAI X K. Gait classification for early detection and severity rating of Parkinson's disease based on hybrid signal processing and machine learning methods[J]. Cogn Neurodyn, 2024, 18(1): 109-132. DOI: 10.1007/S11571-022-09925-9.
[8]
KEMPSTER P. Prodromal and advanced non-motor features of Parkinson's disease[J/OL].BMJ Neurol Open, 2021, 3(1): e000168 [2025-07-21]. https://doi.org/10.1136/BMJNO-2021-000168. DOI: 10.1136/BMJNO-2021-000168.
[9]
MORRIS H R, SPILLANTINI M G, SUE C M, et al. The pathogenesis of Parkinson's disease[J]. Lancet, 2024, 403(10423): 293-304. DOI: 10.1016/S0140-6736(23)01478-2.
[10]
YANG C, LIAO Q, HAO L, et al. Advances in the Application of Three Novel Magnetic Resonance Imaging Techniques and Radiomics in Parkinson's Disease[J]. China Modern Doctor, 2024, 62(34): 108-111. DOI: 10.3969/j.issn.1673-9701.2024.34.024.
[11]
CHEN Y Y, WANG Y X, WEN M, et al. Application of magnetic resonance imaging-based radiomics in the diagnosis of depression in Parkinson's disease[J]. Anhui Medical and Pharmaceutical Journal, 2024, 28(12): 2407-2411. DOI: 10.3969/j.issn.1009-6469.2024.12.015.
[12]
KANG J J, CHEN Y, XU G D, et al. Combining quantitative susceptibility mapping to radiomics in diagnosing Parkinson's disease and assessing cognitive impairment[J]. Eur Radiol, 2022, 32(10): 6992-7003. DOI: 10.1007/s00330-022-08790-8.
[13]
TIAN Y T. Preliminary exploration of the α-synuclein pathology in Parkinson's disease based on 3D-T1WI, 3D chemical exchange saturated transfer imaging and DTI free water imaging[D]. Beijing: Peking Union Medical College, 2024. DOI: 10.27648/d.cnki.gzxhu.2024.000048.
[14]
CHANG B, GENG Z, MEI J M, et al. Application of multimodal deep learning and multi-instance learning fusion techniques in predicting STN-DBS outcomes for Parkinson's disease patients[J/OL]. Neurotherapeutics, 2024, 21(6): e00471 [2025-07-21]. https://doi.org/10.1016/J.NEUROT.2024.E00471. DOI: 10.1016/J.NEUROT.2024.E00471.
[15]
FU J Y, CHEN H Y, XU C L, et al. Harnessing routine MRI for the early screening of Parkinson's disease: a multicenter machine learning study using T2-weighted FLAIR imaging[J/OL]. Insights Imaging, 2025, 16(1): 92 [2025-07-21]. https://doi.org/10.1186/S13244-025-01961-3. DOI: 10.1186/S13244-025-01961-3.
[16]
ACCIOLY A R M, MENEZES V O, CALIXTO L H, et al. Machine Learning-Based Diagnostic Prediction Model Using T1-Weighted Striatal Magnetic Resonance Imaging for Early-Stage Parkinson's Disease Detection[J]. Acad Radiol, 2025, 32(7): 4177-4187. DOI: 10.1016/J.ACRA.2025.04.001.
[17]
JIA F L, LIAO Y, LI X S, et al. Preliminary study on quantitative assessment of the fetal brain using MOLLI T1 mapping sequence[J]. J Magn Reson Imaging, 2022, 56(5): 1505-1512. DOI: 10.1002/jmri.28195.
[18]
EMINIAN S, HAJDU S D, MEULI R A, et al. Rapid high resolution T1 mapping as a marker of brain development:Normative ranges in key regions of interest[J/OL]. PloS One, 2018, 13(6): e0198250 [2025-07-21]. https://pubmed.ncbi.nlm.nih.gov/29902203/. DOI: 10.1371/|journal.pone.0198250.
[19]
SCHMARANZER F, AFACAN O, LERCH T D, et al. Magnetization-prepared 2 Rapid Gradient-Echo MRI for B1 Insensitive 3D T1 Mapping of Hip Cartilage: An Experimental and Clinical Validation[J]. Radiology, 2021, 299(1): 150-158. DOI: 10.1148/radiol.2021200085.
[20]
YANG C, LIAO Q, HAO L, et al. Preliminary study on different stages of Parkinson's Disease combined with amide proton transfer imaging and quantitative susceptibility mapping[J]. Chin J Magn Reson Imaging, 2024, 15(12): 79-86. DOI: 10.12015/issn.1674-8034.2024.12.012.
[21]
ZHANG Z H, PENG J X, SONG Q W, et al. Identification of Depression Subtypes in Parkinson's Disease Patients via Structural MRI Whole-Brain Radiomics: An Unsupervised Machine Learning Study[J/OL]. CNS Neurosci Ther, 2025, 31(2): e70182 [2025-07-21]. https://doi.org/10.1111/CNS.70182. DOI: 10.1111/CNS.70182.
[22]
LUO Y, XIANG Y Q, LIU J B, et al. A Multi-omics Framework Based on Machine Learning as a Predictor of Cognitive Impairment Progression in Early Parkinson's Disease[J]. Neurol Ther, 2025, 14(2): 643-658. DOI: 10.1007/S40120-025-00716-Y.
[23]
LING R H, CEN X X, WU S Y, et al. Explainable graph neural network based on metabolic brain imaging for differential diagnosis of parkinsonism[J/OL]. Front Aging Neurosci, 2025, 17: 1580910 [2025-07-21]. https://doi.org/10.3389/FNAGI.2025.1580910. DOI: 10.3389/FNAGI.2025.1580910.
[24]
FATOUROS P P, MARMAROU A, KRAFT K A, et al. In vivo brain water determination by T1 measurements: effect of total water content, hydration fraction, and field strength[J]. Magn Reson Med, 1991, 17(2): 402-413. DOI: 10.1002/mrm.1910170212.
[25]
LI Z, SUN J X, LIN H P, et al. Predicting motor function improvement following deep brain stimulation of the subthalamic nucleus for Parkinson's disease based on STN-T2MRI radiomics[J]. J Parkinsons Dis, 2025, 15(3): 561-572. DOI: 10.1177/1877718X251319697.
[26]
LIAN Y D, XU Y B, HU L L, et al. Machine learning-based brain magnetic resonance imaging radiomics for identifying rapid eye movement sleep behavior disorder in Parkinson's disease patients[J/OL]. BMC Med Imaging, 2025, 25(1): 227 [2025-07-21]. https://doi.org/10.1186/s12880-025-01748-4. DOI: 10.1186/s12880-025-01748-4.
[27]
ZHANG N G, ZHAO W, SHANG S A, et al. Diffusion Kurtosis Imaging in Diagnosing Parkinson's Disease: A Preliminary Comparison Study Between Kurtosis Metric and Radiomic Features[J]. Acad Radiol. 2025, 32(2): 922-929. DOI: 10.1016/J.ACRA.2024.07.001.
[28]
BU S T, PANG H Z, LI X L, et al. Multi-parametric radiomics of conventional T1 weighted and susceptibility-weighted imaging for differential diagnosis of idiopathic Parkinson's disease and multiple system atrophy[J/OL]. BMC Med Imaging, 2023, 23(1): 204 [2025-07-21]. https://doi.org/10.1186/S12880-023-01169-1. DOI: 10.1186/S12880-023-01169-1.
[29]
PANG H Z, YU Z Y, YU H M, et al. Multimodal striatal neuromarkers in distinguishing parkinsonian variant of multiple system atrophy from idiopathic Parkinson's disease[J]. CNS Neuroscience & Therapeutics, 2022, 28(12): 2172-2182. DOI: 10.1111/CNS.13959.
[30]
CUI J Q, YU Q E, ZHANG T J. Research progress of functional MRI radiomics in Parkinson's disease[J]. Chin J Magn Reson Imaging, 2025, 16(6): 132-138. DOI: 10.12015/issn.1674-8034.2025.06.020.
[31]
GORGES M, MULLER H P, LIEPELT-SCARFONE I, et al. Structural brain signature of cognitive decline in Parkinson's disease: DTI-based evidence from the LANDSCAPE study[J/OL]. Ther Adv Neurol Disord, 2019, 12: 1756286419843447 [2025-07-21]. https://doi.org/10.1177/1756286419843447. DOI: 10.1177/1756286419843447.
[32]
WEI F L, LIU D, LI X J, et al. Assessment of Parkinson's disease severity based on T2-weighted magnetic resonance imaging radiomic modeling[J]. Chin J Magn Reson Imaging, 2024, 15(11): 12-16, 38. DOI: 10.12015/issn.1674-8034.2024.11.003.
[33]
SU D N, ZHANG Z J, ZHANG Z, et al. Distinctive Pattern of Metal Deposition in Neurologic Wilson Disease: Insights From 7T Susceptibility-Weighted Imaging[J/OL]. Neurology, 2024, 102(12): e209478 [2025-07-21]. https://doi.org/10.1212/wnl.0000000000209478. DOI: 10.1212/wnl.0000000000209478.
[34]
CARANOVA M, SOARES J F, BATISTA S, et al. A systematic review of microstructural abnormalities in multiple sclerosis detected with NODDI and DTI models of diffusion-weighted magnetic resonance imaging[J]. Magn Reson Imaging, 2023, 104: 61-71. DOI: 10.1016/j.mri.2023.09.010.
[35]
WU H T, HONG H, WU C Q, et al. Regional white matter hyperintensity volume in Parkinson's disease and associations with the motor signs[J]. Ann Clin Transl Neurol, 2023, 10(9): 1502-1512. DOI: 10.1002/acn3.51839.

PREV Default mode network functional connectivity changes in adolescent bipolar disorder patients after short-term treatment
NEXT Glymphatic system alterations and cognitive-emotional impairment in bilateral sudden sensorineural hearing loss: A structural MRI study
  


LINK


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