Quality control in functional MRI studies with MRIQC and fMRIPrep

Area of Science:

• Neuroimaging
• Data Science

Background:

• Quality assessment (QA) and quality control (QC) in magnetic resonance imaging (MRI) research are vital but often resource-intensive, leading to inconsistent practices across institutions.
• Substandard MRI data can significantly increase false positive and false negative rates in research findings, compromising results.
• Current QA/QC methods in functional MRI (fMRI) research vary widely, highlighting a need for standardized, efficient protocols.

Purpose of the Study:

• To demonstrate a practical, visual assessment protocol for quality control (QC) of functional (blood-oxygen dependent-level; BOLD) MRI data.
• To define specific exclusion criteria for whole-brain voxel-wise BOLD analyses based on common artifacts.
• To contribute to the standardization of fMRI data quality assessment and encourage community discussion.

Main Methods:

• Developed a protocol involving one-by-one visual assessment of fMRI images.
• Utilized reports generated by MRIQC and fMRIPrep for QC checkpoints.
• Applied the protocol to a composite dataset (n=181) from open fMRI studies, documenting exclusion criteria and common artifacts.

Main Results:

• The protocol led to the exclusion of 97% of the data (176 out of 181 subjects) due to artifacts, as subjects were specifically selected to demonstrate potential issues.
• Common artifacts and defects necessitating data exclusion were identified and described.
• All generated materials and QC decisions were released to support standardization.

Conclusions:

• The proposed visual assessment protocol, leveraging MRIQC and fMRIPrep, offers a structured approach to fMRI data QC.
• Standardizing QA/QC in fMRI research is essential for reducing data errors and improving the reliability of neuroimaging findings.
• The open release of materials and documented decisions aims to foster community engagement and advance QA/QC practices in the field.