The effects of motion on parametric fMRI analysis techniques

Area of Science:

• Neuroimaging
• Functional Magnetic Resonance Imaging (fMRI)
• Data Analysis

Background:

• Subject motion during fMRI studies generates spurious signals that can mimic true neural activation.
• Motion correction is crucial for reliable fMRI data analysis.
• Existing rigid body registration algorithms, often applied to fMRI, have not been definitively proven to reduce motion effects to an acceptable level.

Purpose of the Study:

• To assess the effectiveness of rigid body co-registration in mitigating motion artifacts in fMRI correlation analysis.
• To evaluate the impact of motion on different correlation approaches used in fMRI.
• To introduce a novel visualization technique for assessing motion effects in fMRI correlation analysis.

Main Methods:

• Utilized volume data to evaluate rigid body co-registration's impact on motion artifacts across various fMRI correlation methods.
• Developed a scatter plot visualization correlating motion effects with local image gradients.
• Tested the technique on fMRI datasets exhibiting motion-correlated artifacts, both with and without rigid body motion correction.

Main Results:

• Demonstrated that rigid body co-registration is necessary for all correlation-based fMRI analysis techniques.
• Investigated and explained specific differences between popular correlation forms.
• Showed that while motion effects are not entirely eliminated, they are statistically minimized for certain correlation analyses.

Conclusions:

• Rigid body co-registration is proven to be effective in reducing motion artifacts in fMRI.
• The developed visualization technique aids in verifying analysis results and identifying unambiguous activation regions.
• Co-registration is essential for accurate correlation-based fMRI analysis, with specific methods offering improved statistical elimination of motion effects.