Diagnosis of early Alzheimer's disease based on dynamic high order networks.

Machine learning methods have been widely used for early diagnosis of Alzheimer's disease (AD) via functional connectivity networks (FCNs) analysis from neuroimaging data. The conventional low-order FCNs are obtained by time-series correlation of the whole brain based on resting-state functional magnetic resonance imaging (R-fMRI). However, FCNs overlook inter-region interactions, which limits application to brain disease diagnosis. To overcome this drawback, we develop a novel framework to exploit the high-level dynamic interactions among brain regions for early AD diagnosis. Specifically, a sliding window approach is employed to generate some R-fMRI sub-series. The correlations among these sub-series are then used to construct a series of dynamic FCNs. High-order FCNs based on the topographical similarity between each pair of the dynamic FCNs are then constructed. Afterward, a local weight clustering method is used to extract effective features of the network, and the least absolute shrinkage and selection operation method is chosen for feature selection. A support vector machine is employed for classification, and the dynamic high-order network approach is evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Our experimental results demonstrate that the proposed approach not only achieves promising results for AD classification, but also successfully recognizes disease-related biomarkers.

Low Rank Self-calibrated Brain Network Estimation and Autoweighted Centralized Multi-Task Learning for Early Mild Cognitive Impairment Diagnosis.

Detection of mild cognitive impairment (MCI) is important, and appropriate interventions can be taken to delay or prevent its progression to Alzheimer's disease (AD). The construction of brain networks based on brain image data to depict the interaction of brain functions or structures at the level of brain connections has been widely used to identify individuals with MCI/AD from the normal control (NC). Exploring the structural and functional connections and interactions between brain regions is beneficial to detect MCI. For this reason, we propose a new model for automatic MCI diagnosis based on this information. Firstly, a new functional brain network estimation method is proposed. Self-calibration is introduced using quality indicators, and functional brain network estimation is performed at the same time. Then we integrate the functional and structural connected neuroimaging patterns into our multitask learning model to select informative feature. By identifying synergies and differences between different tasks, the most discriminative features are determined. Finally, the most relevant features are sent to the support vector machine classifier for diagnosis and identification of MCI. The experimental results based on the public Alzheimer's disease neuroimaging (ADNI) show that our method can effectively diagnose different stages of MCI and assist the physician to improve the MCI diagnostic accuracy. At the same time, compared with the existing classification methods, the proposed method achieves relatively high classification accuracy. In addition, it can identify the most discriminative brain regions. These findings suggest that our approach not only improves classification performance, but also successfully identifies important biomarkers associated with disease.