Objective: To investigate whether the multiparametric combination of structural and functional (fMRI) neuroimaging parameters improves the prediction of mild cognitive impairment (MCI) in Parkinson’s disease (PD) patients compared to single parametric structural or fMRI measures.

Background: Cognitive deficits, including MCI, often accompany motor impairments in PD [1]. While neuroimaging is commonly used for MCI prediction [1-3], we integrated Cortical Myelination (CM) from T1-weighted/T2-weighted (T1w/T2w) ratios with Kendall’s Coefficient of Concordance-Regional Homogeneity (KCC-ReHo) from resting-state fMRI, hypothesizing their combination would improve PD-MCI prediction accuracy over single-parameter approaches.

Method: We recruited 30 PD patients, 18 PD-MCI and 12 cognitively normal (PD-NC), and divided them into training (n=18) and independent testing (n=12) sets. Structural MRI (T1w MPRAGE and T2w SPACE) and resting-state BOLD MRI scans were acquired. Data augmentation (n=450, Figure 1) was applied to the training set to enhance model robustness. After preprocessing [4] and dimensionality reduction [5], Random Forest (RF) and Support Vector Machine (SVM) models were trained on CM alone, KCC-ReHo alone, and their combination and evaluated on the independent testing set.

Results: Figure 2A highlights similar accuracy in identifying PD-FOG patients across all three approaches but differences in classifying PD-nFOG patients. The ROC curves [Figure 2B] show the combined approach to achieving the highest area under the curve (AUC=0.85), followed by KCC-ReHo (0.78) and CM (0.65). Performance metrics [Figure 2C] indicate that KCC-ReHo had the highest precision (85%) and specificity (67%), while all approaches showed similar recall (78%). Figure 3 illustrates the brain regions contributing to prediction, with CM showing sparse frontal and parietal involvement, while KCC-ReHo displayed more widespread predictive regions.

Conclusion: Our findings suggest that KCC-ReHo is a more reliable parameter for PD-MCI prediction than CM or multiparametric approaches. The superior precision of KCC-ReHo suggests functional connectivity changes are more sensitive to cognitive impairment in PD than structural alterations. Future studies should explore combining KCC-ReHo with diffusion MRI parameters to investigate the early detection of cognitive decline in PD.

Data Augmentation Techniques

Machine Learning Performance Metrics

Brain Regions Contributing to the Prediction Model

References: [1] Litvan, I.; Aarsland, D.; Adler, C. H.; Goldman, J. G.; Kulisevsky, J.; Mollenhauer, B.; Rodriguez-Oroz, M. C.; Troster, A. I.; Weintraub, D. MDS Task Force on mild cognitive impairment in Parkinson’s disease: critical review of PD-MCI. Mov Disord 2011, 26 (10), 1814-1824. DOI: 10.1002/mds.23823 From NLM Medline.
[2] Beheshti, I.; Ko, J. H. Predicting the occurrence of mild cognitive impairment in Parkinson’s disease using structural MRI data. Front Neurosci 2024, 18, 1375395. DOI: 10.3389/fnins.2024.1375395 From NLM PubMed-not-MEDLINE.
[3] Lin, H.; Liu, Z.; Yan, W.; Zhang, D.; Liu, J.; Xu, B.; Li, W.; Zhang, Q.; Cai, X. Brain connectivity markers in advanced Parkinson’s disease for predicting mild cognitive impairment. Eur Radiol 2021, 31 (12), 9324-9334. DOI: 10.1007/s00330-021-08086-3 From NLM Medline.
[4] Abdennour, N. O., T.; Amor N. B. The importance of signal pre-processing for machine learning: The influence of Data scaling in a driver identity classification. 2021 IEEE/ACS 18th International Conference on Computer Systems and Applications (AICCSA) 2021, pp. 1-6. DOI: 10.1109/AICCSA53542.2021.9686756.
[5] Howley, T.; Madden, M. G.; O’Connell, M. L.; Ryder, A. G. The effect of principal component analysis on machine learning accuracy with high-dimensional spectral data. Knowl-Based Syst 2006, 19 (5), 363-370. DOI: 10.1016/j.knosys.2005.11.014.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/multiparametric-prediction-of-mild-cognitive-impairment-in-parkinsons-disease-using-a-combination-of-structural-and-functional-mri-measures/