Related Experiment Video
Updated: Jul 5, 2026
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
Analysis of Quantitative Susceptibility Mapping Data for Multi-Site and Multi-Modal Brain Imaging Studies: For
Christa Sonderer1, Chidi Ugonna1, Kevin Johnson2
1Department of Biomedical Engineering, University of Arizona, Tucson, Arizona, USA.
Introduction:
Quantitative susceptibility mapping (QSM) is a valuable tool for measuring brain iron and its changes with age. However, multi-site QSM studies face challenges: (1) it is difficult to precisely align brain atlas to an individual's anatomy (for region of interest [ROI]-based analysis) and (2) it is unknown whether the choice of QSM algorithm impacts multi-site analyses. This study leverages data from over 500 individuals across four study sites to investigate the impact of ROI on QSM consistency across sites and evaluate site-to-site variability and bias in 16 QSM processing pipelines.
Methods:
We obtained susceptibility maps using 16 QSM pipelines. ROI-analysis was performed using 14 cortical and sub-cortical ROIs from the AAL3, and refined ROIs based on AAL3 and multi-contrast magnetic resonance imaging data. For each ROI and pipeline, between-site variance was performed via one-way analysis of variance, and correlation between mean susceptibility and age was computed.
Results:
Our data show that the QSM results obtained from multi-site studies vary depending on the QSM processing pipeline. Background field removal algorithm and reference tissue for magnetic susceptibility normalization impacted the consistency of QSM measures across sites and were differentially impacted by site-to-site variability. The consistency of ROI-based QSM measures across study sites was not greatly impacted by ROI. Regardless of QSM processing pipelines, correlation between susceptibility and age was largely consistent within non-cortical brain regions (except the substantia nigra and red nucleus) and was not significantly impacted by site-to-site variability.
Conclusion:
The selection of QSM reconstruction pipelines impacts results obtained from multi-site studies, indicating the importance of harmonizing QSM pipelines when interpreting susceptibility measures across studies.
