Sensorimotor network parcellation for pre-surgical patients using low-pass filtered fMRI

Area of Science:

• Neuroimaging
• Neurosurgery
• Epilepsy Research

Background:

• Pre-surgical mapping of sensorimotor and language functions is vital for reducing neurological deficits during epilepsy and tumor resection.
• Functional MRI (fMRI), including resting-state and task-evoked paradigms, is used for non-invasive pre-surgical mapping.
• The reliability of fMRI mapping is a clinical concern due to the lack of standardized task paradigms.

Purpose of the Study:

• To improve the reliability and accuracy of fMRI-based pre-surgical mapping.
• To develop a method for enhancing fMRI mapping by integrating diverse task paradigms with resting-state data.
• To validate the enhanced mapping technique against invasive electrophysiological methods.

Main Methods:

• Task fMRI data from motor, word repetition, and picture naming paradigms were low-pass filtered (0.01-0.08Hz) and concatenated.
• K-means clustering was applied to the processed fMRI data for network parcellation.
• The resulting sensorimotor network parcellations were validated using electrocorticography (ECoG) and electrical cortical stimulation (ECS) mapping.

Main Results:

• The sensorimotor network was reliably parcellated into distinct hand and tongue sub-regions.
• The enhanced fMRI mapping method demonstrated improved accuracy and specificity compared to motor-task fMRI alone.
• This approach successfully provided accurate mapping for patients who previously failed task fMRI.

Conclusions:

• Integrating multiple task paradigms with resting-state fMRI significantly enhances pre-surgical mapping reliability.
• The intrinsic nature of cortical sensorimotor networks across tasks supports a link between spontaneous and task-evoked BOLD signals.
• This validated method offers a more robust tool for pre-surgical functional brain mapping, especially in challenging cases.