Research on unsupervised cross-domain adaptive based depression recognition algorithm
WANG Chenyu
LI Xiang
WEI Benzheng
Abstract：In order to solve the inconsistent feature distribution between domains of multi-site resting-state functional magnetic resonance imaging(rs-fMRI)data,we proposed an unsupervised cross-domain adaptive depression recognition algorithm to make full use of multi-scale timing information,balance the feature distribution difference between the source domain and the target domain of multi-site rs-fMRI data.First,the inter-domain distribution difference was reduced through generative adversarial network,The do-main alignment mechanism with statistical feature matched with adversarial learning was employed to achieve unsupervised cross-do-main feature alignment between two domains.Then,an attention-guided multi-scale spatiotemporal graph convolution module was de-signed to extract and fuse multi-scale feature information.Based on the analysis of 681 subjects in the REST-meta-MDD dataset,the accuracy,area under the receiver operating characteristic curve,sensitivity,specificity and precision of this algorithm reached 67.57％,65.16％,89.19％,66.22％and 65.00％,respectively.The experimental results show that the algorithm has excellent recog-nition performance and can provide certain technical support for the clinical auxiliary diagnosis of major depressive disorder.
Keywords：Major depressive disorderResting-state functional magnetic resonance imagingDomain adaptationAdversarial learningMulti-scale information
Publication Date：2025-04-30
Online Publishing Date：2026-09-11(First online date of this platform, not the publication date of the document)
Pages：8( 75-82 )