Abstract
The presence of iron is an important factor for normal brain functions, whereas excessive deposition of iron may impair normal cognitive function in the brain and lead to Alzheimer's disease (AD). MRI has been widely applied to characterize brain structural and functional changes caused by AD. However, the effectiveness of using susceptibility-weighted imaging (SWI) for the analysis of brain iron deposition is still unclear, especially within the context of early AD diagnosis. Thus, in this study, we aim to explore the relationship between brain iron deposition measured by SWI with the progression of AD using various feature selection and classification methods. The proposed model was evaluated on a 69-subject SWI imaging dataset consisting of 24 AD patients, 21 mild cognitive impairment patients, and 24 normal controls. The identified AD progression-related regions were then compared with the regions reported from previous genetic association studies, and we observed considerable overlap between these two. Further, we have identified a new potential AD-related gene (MEF2C) closely related to the interaction between iron deposition and AD progression in the brain.
Introduction
Alzheimer’s Disease (AD) remains a global health crisis with no known cure, making early diagnosis essential for managing progression. While traditional MRI typically focuses on brain atrophy (shrinkage), recent research highlights the role of iron deposition as a potential biomarker. Although iron is necessary for brain function, excessive deposits can trigger neurotoxicity and are closely linked to the amyloid plaques and tau tangles characteristic of AD. However, using iron as a diagnostic tool is challenging because its deposition patterns are complex. This study investigates whether Susceptibility-Weighted Imaging (SWI)—an MRI technique highly sensitive to iron—can be combined with machine learning to accurately distinguish between healthy aging, Mild Cognitive Impairment (MCI), and AD.
Methods
The researchers developed a multi-stage pipeline to analyze iron deposition patterns:
- Imaging & Preprocessing: SWI data was collected and mapped onto three distinct brain atlases—AAL, Harvard-Oxford, and MMP—to extract iron concentration features from specific anatomical regions.
- Feature Selection: To manage the high dimensionality of brain imaging data, the study employed Lasso and Adaptive Lasso algorithms. These "shrinkage" methods identify the most predictive brain regions while filtering out irrelevant noise.
- Classification: Four machine learning models—AdaBoost, LinearSVC, Random Forest, and XGBoost—were tested to categorize subjects into Normal, MCI, or AD groups.
- Genomic Integration: Findings were cross-referenced with the Allen Human Brain Atlas. The team calculated a "gene presence frequency" to determine which risk-factor genes were most active in the iron-heavy regions identified by the AI model.
Results
The study yielded several key findings regarding the diagnostic utility of SWI and the genetic underpinnings of iron deposition:
- Classification Performance: The best-performing model (LinearSVC using the Lasso method on the AAL atlas) achieved an accuracy of 73.88% in the three-class classification task.
- Regional Identification: The model identified 20 key brain regions associated with AD progression, including the amygdala, hippocampus, and putamen. These findings align with major meta-analyses that highlight the frontal lobe, parietal lobe, and caudate nucleus as critical areas for iron accumulation.
- Genetic Association: There was significant overlap between regions of high iron deposition and regions where AD-risk genes (such as APOE, MAPT, and CLU) are highly expressed. Notably, the study highlighted MEF2C as a top-tier gene associated with these iron-heavy regions. MEF2C is known to regulate memory-forming structures, suggesting it may play a crucial role in how iron-induced damage leads to cognitive decline.
Discussion and Conclusion
This research demonstrates that iron deposition follows a specific regional "spatial fingerprint" that shifts as patients progress from MCI to AD. The discovery of the MEF2C gene's association with these patterns suggests that iron deposition is not merely a byproduct of the disease but is tied to the underlying genetic vulnerability of specific brain areas.
While the results are promising, the study's sample size of 69 subjects necessitates further validation with larger datasets. Additionally, while the AAL atlas provided the best results, its relatively coarse resolution suggests that future work using finer-grained parcellation or voxel-level analysis could pinpoint iron deposits with even higher precision. Further investigation into the biological mechanisms by which MEF2C influences iron deposition will be essential for translating these findings into clinical practice.
Summary updated