Subject-Motion Correction in HARDI Acquisitions: Choices and Consequences

Area of Science:

• Neuroimaging
• Diffusion MRI
• Computational Neuroscience

Background:

• Diffusion-weighted imaging (DWI) is susceptible to motion artifacts from physiological and mechanical sources.
• Current motion correction in DWI often relies on heuristics, lacking standardized guidelines and potentially introducing confounding factors in population studies.
• Existing software like DTIprep, FSL, and TORTOISE offer motion correction, but their impact on downstream analyses is not fully understood.

Purpose of the Study:

• To develop and apply a comprehensive framework for systematically evaluating the impact of different motion correction strategies on DWI-derived measures.
• To provide clear guidelines and recommendations for users regarding motion correction choices in diffusion MRI data analysis.
• To investigate the consequences of motion correction choices on the reliability and reproducibility of neuroimaging findings.

Main Methods:

• Utilized human brain high-angular-resolution diffusion imaging (HARDI) data from a controlled motion experiment.
• Simulated various degrees of motion corruption and noise contamination.
• Assessed correction choices including exclusion/scrubbing, registration with different interpolations, and interpolation of all directions.

Main Results:

• Compared the effects of motion correction strategies on fiber orientation distribution functions (fODFs), local fiber orientation deviation, global brain connectivity (graph diffusion distance), and tract reproducibility.
• Systematically explored and illustrated the effects of various motion correction choices on these quantitative metrics.
• Identified significant impacts of correction choices on DWI-derived measures, highlighting the need for careful selection.

Conclusions:

• Discourages the use of arbitrary thresholds on motion parameters for identifying corrupted volumes.
• Emphasizes the importance of a systematic evaluation of motion correction strategies to avoid introducing confounding factors.
• Underscores the need for standardized guidelines in practical diffusion MRI data processing and analysis.