Current Alzheimer research · 2026
Multitask Sparse Canonical Correlation Analysis and Regression with Parameter Decomposition based on Deep Subspace Reconstruction for Alzheimer's Disease.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Kong Wei, Su Pengfei, Xu Yufang, Wang Shuaiqun, Wei Kai, Ke Fengchun, Wen Gen, Yu Yaling
Аннотация
INTRODUCTION: Multimodal imaging genomics overcomes the limitations of singlemodality analyses, offering a more comprehensive understanding of brain pathophysiology. Traditional methods like sparse canonical correlation analysis (SCCA) and its improved versions have been widely used to identify key brain regions and single nucleotide polymorphisms (SNPs) associated with neurodegenerative diseases, such as Alzheimer's disease (AD). However, the fusion of complex and heterogeneous multimodal neuroimages, along with the linear formulation of these methods, limits their ability to capture complex, nonlinear relationships in imaging-genetics data.
METHODS: To address this limitation, a novel framework, DSR-PDMTSCCAR, is introduced for multimodal AD data. This framework combines deep subspace reconstruction for nonlinear mapping and parameter decomposition to extract modality-consistent and modality-specific features across structural MRI (sMRI) and positron emission tomography (PET) data. Additionally, multitask sparse canonical correlation analysis (MTSCCA) is incorporated to capture correlations across multiple tasks, facilitating modeling of heterogeneous multimodal associations.
RESULTS: Evaluations on simulated datasets and the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort show that DSR-PDMTSCCAR outperforms conventional methods, including SCCA, partial least squares (PLS), and deep canonical correlation analysis (DCCA). The method identifies biomarkers such as hippocampal and amygdalar alterations in sMRI/PET and APOErelated genetic variants, all consistent with AD pathology.
DISCUSSION: The maximum canonical correlation coefficient (CCC) achieved was 0.1759, compared to 0.1525 for MTSCCA. The PET-SNP correlation (0.1896) was more than double that of MTSCCA (0.0864).
CONCLUSION: The DSR-PDMTSCCAR framework offers robust performance in multimodal imaging- genetics analysis, providing enhanced insights into AD pathophysiology and advancing precision medicine.
METHODS: To address this limitation, a novel framework, DSR-PDMTSCCAR, is introduced for multimodal AD data. This framework combines deep subspace reconstruction for nonlinear mapping and parameter decomposition to extract modality-consistent and modality-specific features across structural MRI (sMRI) and positron emission tomography (PET) data. Additionally, multitask sparse canonical correlation analysis (MTSCCA) is incorporated to capture correlations across multiple tasks, facilitating modeling of heterogeneous multimodal associations.
RESULTS: Evaluations on simulated datasets and the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort show that DSR-PDMTSCCAR outperforms conventional methods, including SCCA, partial least squares (PLS), and deep canonical correlation analysis (DCCA). The method identifies biomarkers such as hippocampal and amygdalar alterations in sMRI/PET and APOErelated genetic variants, all consistent with AD pathology.
DISCUSSION: The maximum canonical correlation coefficient (CCC) achieved was 0.1759, compared to 0.1525 for MTSCCA. The PET-SNP correlation (0.1896) was more than double that of MTSCCA (0.0864).
CONCLUSION: The DSR-PDMTSCCAR framework offers robust performance in multimodal imaging- genetics analysis, providing enhanced insights into AD pathophysiology and advancing precision medicine.