Exploration of scanning effects in multi-site structural MRI studies

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Data Science

Background:

• Multi-site MRI studies require data pooling but face systematic differences from varied scanning platforms.
• These scanning differences can confound true biological effects, necessitating methods to improve data integrity.
• Calibration of scanning parameters or inclusion as covariates are potential strategies to mitigate bias.

Purpose of the Study:

• To develop and validate a data-driven correction method for scanning effects in multi-site sMRI studies.
• To explore scanning effects using a source-based morphometry (SBM) model.
• To assess the influence of scanning parameters on neuroimaging data.

Main Methods:

• Utilized a source-based morphometry (SBM) model to analyze multi-site sMRI data.
• Extracted independent components from the data to identify associations with scanning parameters.
• Removed identified scanning-related components to correct the data.

Main Results:

• A small set of SBM components captured the majority of variance related to scanning differences.
• Pronounced scanning effects were observed in the brainstem and thalamus, linked to magnetic field strength, inversion time, and RF-receiving coil in a large cohort.
• The SBM approach effectively corrected scanning effects in a study of schizophrenia patients and healthy controls.

Conclusions:

• Both SBM and GLM (General Linear Model) correction effectively reduced scanning effects.
• SBM-corrected data revealed more significant differences between patient and control groups.
• SBM offers greater flexibility and better handling of collinear effects due to its data-driven nature.