Label-guided low-rank Approximation for functional brain network learning in identifying subcortical vascular cognitive impairment

• Published in: Biomedical signal processing and control Vol. 98; p. 106766
• Main Authors: Jiang, Xiao, Wang, Guangyu, Zhang, Limei, Xi, Xiaoming, Leone, Renato De, Qiao, Lishan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2024
• Subjects: Functional Brain Network; Identification; Ischemic Vascular Cognitive Impairment; Label-guided; Low-rank; Rest-state Functional Magnetic Resonance Imaging; Low-rank; Rest-state Functional Magnetic Resonance Imaging; Identification; Functional Brain Network; Ischemic Vascular Cognitive Impairment; Label-guided
• ISSN: 1746-8094
• DOI: 10.1016/j.bspc.2024.106766

•A novel label-guided low-rank FBN learning framework is proposed, which explicitly integrates the label information of data samples by the regularization terms into the learning model. This integration ensures that brain network learning is not entirely isolated from subsequent classification tasks. Moreover, the presence of label information typically enhances the discriminative power of the learned network. Guided by the label information, the proposed method implicitly focuses on features relevant to brain disorders. As a result, the resulting network structure aligns with the specific cognitive or disease-related features of interest, leading to more accurate and biologically meaningful representations.•The proposed method concurrently considers FBNs across all the subjects, distinguishing itself from most existing methods that independently learn FBNs for each subject. Concretely, the method considers relationships among different individuals and utilizes their similar topological structures by the low-rank constraint. Moreover, such a constraint transfers the information learned during the optimization of brain networks guided by labels to the testing subject.•The presented method constitutes a versatile framework that can be effectively employed in the analysis of various brain pathologies. Our method addresses the interplay between label information and data characteristics by flexibly adjusting the weighting of regularization terms, thereby enhancing adaptability to diverse tasks and datasets. This flexibility allows for its potential application in studying other brain disorders and exploring the relationships between brain connectivity and various cognitive conditions. The obtained outcomes in the detection of distinct cognitive states of SVCI corroborate this assertion, as elaborated upon in the section on Experiments. Functional brain network (FBN) has played a pivotal role in unraveling the inherent mechanism of cognition/behavior and investigating neuropsychological disorders. Particularly, growing evidence suggests that FBNs have shown great potential in classifying brain disorders such as subcortical vascular cognitive impairment (SVCI). However, learning a high-quality FBN is still challenging since there is no ground truth. Most traditional methods tend to focus on learning FBNs independently of the downstream task. This practice neglects the valuable label information crucial for accurate classification. Besides, the methodology of constructing brain networks in an isolated, individualistic manner sidesteps the common information among individuals. To address these issues, we developed a new FBN joint learning strategy named Label-Guided Low-rank Approximation (LGLA). LGLA integrates label information from training subjects and uses unlabeled testing subjects for auxiliary training, enhancing the discriminative power of FBN features tailored to testing subjects. By enforcing class-specific similarities and differences, and applying a low-rank constraint to capture shared information, LGLA presents an optimization framework that can be effectively solved. Experimental results demonstrate the effectiveness of our proposed FBN learning method for identifying SVCI. The proposed method achieved an accuracy of 75.70% and an area under the receiver operating characteristic curve (AUC) of 0.8233 in the task of identifying patients with amnestic mild cognitive impairment (aMCI). For identifying patients with non-amnestic mild cognitive impairment (naMCI), it achieved an accuracy of 63.64% and an AUC of 0.7857, while the task of identifying patients with no cognitive impairment (NCI), it achieved an accuracy of 63.64% and an AUC of 0.7917. Additionally, we further explored the brain network features and discovered potential biomarkers for personalized diagnosis. The LGLA method enhances FBN construction and classification accuracy for SVCI diagnosis by effectively integrating the label information and shared structure among subjects.