Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Harmonizing resting state functional MRI data to increase sample size and to classify tinnitus brain connectivity

Shagun Ajmera¹, Rafay Ali Khan¹, Fatima T. Husain¹; ¹University of Illinois at Urbana-Champaign

Presenter: Shagun Ajmera

The neuroscientific understanding of tinnitus, or ringing in the ears, is limited at present, partly due to a lack of sufficient neuroimaging data. However, data scarcity can be overcome with ‘harmonization’ methods if scans from independent sites and studies are pooled successfully. We achieved data harmonization by merging a modest tinnitus fMRI dataset acquired in our lab (n=35) with another large dataset, the Lifespan Human Connectome Project Aging (n=377). We measured resting-state functional connectivity (FC) between brain regions, and used deep learning architecture to purge dataset-identifying information from FC maps while retaining characteristic patterns of tinnitus-related FC. Hallmark artifactual information of the datasets was significantly reduced while reconstructing individual FC data. Default mode network connectivity was identified to be crucial for distinguishing tinnitus, as masking the network’s connections severely impacted downstream classification on reconstructed FC. The data harmonization model was successful in merging FC matrices from independently acquired fMRI datasets, without losing out on meaningful functional connectivity patterns of interest for the neuropsychological disorder.

Topic Area: Brain Networks & Neural Dynamics

Extended Abstract: Full Text PDF