Multimodal hyper-connectivity of functional networks using functionally-weighted LASSO for MCI classification

• Published in: Medical image analysis Vol. 52; pp. 80 - 96
• Main Authors: Li, Yang, Liu, Jingyu, Gao, Xinqiang, Jie, Biao, Kim, Minjeong, Yap, Pew-Thian, Wee, Chong-Yaw, Shen, Dinggang
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.02.2019; Elsevier BV
• Subjects: Algorithms; Arterial spin labeling (ASL); Biomarkers; Blood flow; Brain; Brain mapping; Cerebral blood flow; Cerebrovascular Circulation - physiology; Classification; Cognitive Dysfunction - classification; Cognitive Dysfunction - diagnostic imaging; Connectome - methods; Echo-Planar Imaging; Functional magnetic resonance imaging; Humans; Hyper-connectivity network; Image Interpretation, Computer-Assisted - methods; Inference; Least squares; Magnetic Resonance Imaging - methods; Mild cognitive impairment (MCI); Multimodality; Networks; Neural networks; Shrinkage; Software; Spin labeling; Spin Labels; Temporal resolution; Topology; Ultra-least squares (ULS); Weighted LASSO; Arterial spin labeling (ASL); Ultra-least squares (ULS); Weighted LASSO; Multimodality; Hyper-connectivity network; Mild cognitive impairment (MCI)
• ISSN: 1361-8415; 1361-8423; 1361-8423
• DOI: 10.1016/j.media.2018.11.006

•Multimodal hyper-networks are constructed via a novel FW-LASSO method from ASL-fMRI and BOLD-fMRI.•The ASL-derived functional connectivity is constructed by an Ultra-GroupLASSO-UOLS algorithm.•Integration of ASL-fMRI and BOLD-fMRI improves the MCI classification accuracy and the subject consistency in the learned connection structure. Recent works have shown that hyper-networks derived from blood-oxygen-level-dependent (BOLD) fMRI, where an edge (called hyper-edge) can be connected to more than two nodes, are effective biomarkers for MCI classification. Although BOLD fMRI is a high temporal resolution fMRI approach to assess alterations in brain networks, it cannot pinpoint to a single correlation of neuronal activity since BOLD signals are composite. In contrast, arterial spin labeling (ASL) is a lower temporal resolution fMRI technique for measuring cerebral blood flow (CBF) that can provide quantitative, direct brain network physiology measurements. This paper proposes a novel sparse regression algorithm for inference of the integrated hyper-connectivity networks from BOLD fMRI and ASL fMRI. Specifically, a least absolution shrinkage and selection operator (LASSO) algorithm, which is constrained by the functional connectivity derived from ASL fMRI, is employed to estimate hyper-connectivity for characterizing BOLD-fMRI-based functional interaction among multiple regions. An ASL-derived functional connectivity is constructed by using an Ultra-GroupLASSO-UOLS algorithm, where the combination of ultra-least squares (ULS) criterion with a group LASSO (GroupLASSO) algorithm is applied to detect the topology of ASL-based functional connectivity networks, and then an ultra-orthogonal least squares (UOLS) algorithm is used to estimate the connectivity strength. By combining the complementary characterization conveyed by rs-fMRI and ASL fMRI, our multimodal hyper-networks demonstrated much better discriminative characteristics than either the conventional pairwise connectivity networks or the unimodal hyper-connectivity networks. Experimental results on publicly available ADNI dataset demonstrate that the proposed method outperforms the existing single modality based sparse functional connectivity inference methods. [Display omitted]