Menu
Ashley Boersma, BS
Pronouns
Rank
Research Staff
Institution
Brigham and Women's Hospital
Department
Psychiatry
Authors
*A. Boersma, S. Bouix, M. Coleman, K.E. Lewandowski, D. Holt, M. Keshavan, D. Ongur, A. Breier, M.E. Shenton, *R.J. Rushmore
Principal Investigator
Jarrett Rushmore
Categories:
“Prior to analysis, MRI volumes typically undergo a series of preprocessing steps, including spatial normalization where brains are placed in a standard space. The Talairach space is a commonly used standard space. Aligning to this space involves adjusting the brain’s position so the anterior and posterior commissure line up on a horizontal line (ACPC line). Automated tools do this by comparing the MRI volume to an existing brain atlas and then adjusting the input brain so it aligns with the atlas’ ACPC line. It is unclear whether demographic variables such as sex and race affect standard MRI preprocessing steps. Accordingly, we examined the impact of race and sex on the accuracy of spatial normalization preprocessing.
High resolution T1-weighted sMRI data was obtained from 177 subjects from the Human Connectome Project-Early Psychosis (HCP-EP) dataset. These scans had been preprocessed and automatically aligned to the Talairach space using the MNI-152 registration model. To test the accuracy of the image registration, the angle of deviation from the ACPC line for each subject was measured using the Markups Module of 3D Slicer v4.11.
A two-way ANOVA revealed a significant main effect of sex on the angle of deviation from the ACPC line, indicating a greater angle of deviation from the ACPC line in males than in females (p<.0001). There was also a significant main effect of race on the angle of deviation from the ACPC line (p=0.03). A significant interaction effect indicated that the main effects were driven by larger ACPC deviations in Black and Asian males (p=0.04).
These effects suggest that automated tools that spatially normalize brains to the ACPC line are not always accurate and that accuracy varies according to sex and race. More needs to be done to account for demographic factors such as sex and race in neuroimaging preprocessing steps.”
