The abstract is read at the publisher; the summary is JClub's.
Deep learning harmonization merged resting-state fMRI functional connectivity from a tinnitus dataset (n=35) with Lifespan Human Connectome Project Aging controls (n=377), preserving tinnitus-discriminative connectivity.
- Why it matters: Tinnitus research is constrained by limited neuroimaging samples and site-to-site heterogeneity, making it hard to pool data and learn reliable connectivity signatures.
- What they did: The study combined age- and gender-matched resting-state FC maps across datasets using neural network modules that remove dataset-identifying information while reconstructing harmonized FC.
- The result: Dataset artifacts were significantly reduced, and tinnitus-vs-control relationships in reconstructed FC were well-preserved, with default mode, ventral attention, and frontoparietal control networks crucial for classification.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.