Group specific optimisation of fMRI processing steps for child and adult data

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Data Analysis

Background:

• Motion artifacts are a significant challenge in functional magnetic resonance imaging (fMRI) data, particularly in pediatric populations.
• Excessive motion can obscure neural signals of interest, impacting the reliability of fMRI studies.

Purpose of the Study:

• To evaluate the effectiveness of various preprocessing techniques in correcting motion artifacts in fMRI data from children and adults.
• To compare the performance of different analytical methods, including agnostic canonical variates analysis (aCVA) and mixed effects general linear model (mGLM).

Main Methods:

• Analysis of fMRI data from 35 children (4-8 years) and 13 adults (18-30 years) during an emotional face paradigm.
• Children were categorized into low and high motion groups based on voxel and degree movement thresholds.
• Evaluation of preprocessing steps, including motion parameter regression (MPR), within the Nonparametric, Prediction, Activation, Influence, Reproducibility, re-Sampling (NPAIRS) framework using aCVA.

Main Results:

• Motion parameter regression (MPR) demonstrated a substantial beneficial impact across all datasets, especially in motion-prone child datasets.
• While motion correction alone had limited impact, MPR significantly improved data quality and reproducibility in children to levels comparable to adults.
• aCVA showed higher sensitivity to task response patterns in face processing regions compared to mGLM, though mGLM identified responses in other areas.

Conclusions:

• Preprocessing choices in fMRI analysis should be tailored to specific groups, particularly considering motion levels in pediatric data.
• MPR is a highly effective method for mitigating motion artifacts in child fMRI, enhancing data reliability.
• The findings underscore the need for group-specific preprocessing strategies to optimize fMRI results in diverse populations.