English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Resting-state functional MRI (rs-fMRI) is increasingly employed in multi-siteresearch to aid neurological disorder analysis. Existing studies usually sufferfrom significant cross-site/domain data heterogeneity caused by site effectssuch as differences in scanners/protocols. Many methods have been proposed toreduce fMRI heterogeneity between source and target domains, heavily relying onthe availability of source data. But acquiring source data is challenging dueto privacy concerns and/or data storage burdens in multi-site studies. To thisend, we design a source-free collaborative domain adaptation (SCDA) frameworkfor fMRI analysis, where only a pretrained source model and unlabeled targetdata are accessible. Specifically, a multi-perspective feature enrichmentmethod (MFE) is developed for target fMRI analysis, consisting of multiplecollaborative branches to dynamically capture fMRI features of unlabeled targetdata from multiple views. Each branch has a data-feeding module, aspatiotemporal feature encoder, and a class predictor. A mutual-consistencyconstraint is designed to encourage pair-wise consistency of latent features ofthe same input generated from these branches for robust representationlearning. To facilitate efficient cross-domain knowledge transfer withoutsource data, we initialize MFE using parameters of a pretrained source model.We also introduce an unsupervised pretraining strategy using 3,806 unlabeledfMRIs from three large-scale auxiliary databases, aiming to obtain a generalfeature encoder. Experimental results on three public datasets and one privatedataset demonstrate the efficacy of our method in cross-scanner and cross-studyprediction tasks. The model pretrained on large-scale rs-fMRI data has beenreleased to the public.

Source-Free Collaborative Domain Adaptation via Multi-Perspective Feature Enrichment for Functional MRI Analysis