NeuroStream: an interactive platform for exploratory visualization and harmonization of multicohort brain MRI data

Area of Science:

• Neuroimaging and computational neuroscience.

Background:

• Large-scale neuroimaging studies integrate multicohort brain MRI data, necessitating tools for batch effect identification and harmonization.
• Existing visualization tools are often cohort-specific, static, or separate harmonization from data exploration.

Purpose of the Study:

• To introduce NeuroStream, an interactive application for real-time visualization and harmonization of multicohort brain MRI quantitative data.
• To enable exploration of quantitative imaging features alongside demographic and clinical variables.

Main Methods:

• NeuroStream utilizes the Streamlit framework for interactive exploration of structural MRI quantitative imaging features.
• The platform supports various preprocessing and harmonization techniques, including log transformation, intracranial volume normalization, and ComBat-based batch correction.
• It offers principal component analysis, distributional plots, group comparison statistics, and covariate-adjusted regression for site effect evaluation.

Main Results:

• NeuroStream facilitates interactive comparison of preprocessing and harmonization options through dynamic visualizations.
• The application enables quantitative and visual assessment of cohort-related variability across different preprocessing settings.
• Demonstrates intuitive inspection of batch-related variance and harmonization outcomes using example datasets from BICWALZS and KoGES multicohort MRI data.

Conclusions:

• NeuroStream provides a unified platform for exploring and harmonizing multicohort brain MRI data.
• The tool simplifies the identification and mitigation of batch effects, improving the reliability of large-scale neuroimaging analyses.
• It empowers researchers to perform complex data assessment and harmonization without custom scripting.