Excessive Censoring Degrades Individual-Specific Cortical Parcellations and Personalized TMS Targets

Area of Science:

• Neuroimaging
• Neuroscience
• Brain Mapping

Background:

• Head motion significantly biases functional connectivity (FC) estimates in resting-state functional MRI (rs-fMRI).
• Current mitigation strategies involve censoring high-motion data, risking signal loss.
• The impact of censoring stringency on personalized brain parcellations and transcranial magnetic stimulation (TMS) targets is not fully understood.

Purpose of the Study:

• To evaluate the efficacy of different head motion censoring strategies in rs-fMRI.
• To determine the optimal censoring approach for individual-specific cortical parcellations and personalized TMS target selection.
• To compare lenient versus strict censoring methods under varying motion conditions.

Main Methods:

• Utilized precision-fMRI datasets from 50 individuals to establish individualized "ground-truth" references from low-motion data.
• Simulated rs-fMRI sessions of 10-min and 20-min durations with controlled motion levels.
• Assessed the quality of cortical parcellations and TMS targets derived from data processed with various censoring strategies.

Main Results:

• Higher motion levels led to greater deviations in parcellations and TMS targets from ground-truth references.
• Lenient censoring consistently yielded higher quality parcellations and TMS targets compared to strict censoring at equivalent motion levels.
• Lenient censoring of high-motion-only runs produced TMS targets comparable or superior to strictly censored mixed-motion runs.

Conclusions:

• Lenient censoring strategies in rs-fMRI processing offer superior results for personalized brain mapping and TMS targeting compared to strict censoring.
• Patients with high-motion rs-fMRI runs may not require re-scanning if lenient censoring is applied.
• These findings suggest a more flexible approach to managing motion artifacts in clinical and research rs-fMRI applications.