Modeling subcutaneous fat improves skull segmentation for individualized volume conductor head models.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Medical Physics

Background:

• Accurate head and brain tissue segmentation is crucial for Transcranial Brain Stimulation (TBS) and magneto-/electroencephalography (M/EEG) source localization.
• Current CHARM pipeline optimizes segmentation using specific T1- and T2-weighted MRI sequences, but common datasets lack this combination, leading to poor scalp and skull segmentation.
• Bone segmentation is inherently difficult in MRI due to low signal and scan sequences optimized for brain contrast.

Purpose of the Study:

• To present an updated CHARM pipeline with an extended tissue probability atlas and optimized parameters.
• To improve the segmentation accuracy of skull and subcutaneous fat across various MRI sequence combinations.

Main Methods:

• Updated CHARM pipeline incorporating an extended tissue probability atlas including subcutaneous fat.
• Optimized segmentation parameters for enhanced accuracy.
• Evaluation across different combinations of T1- and T2-weighted MRI sequences.

Main Results:

• The updated CHARM configuration significantly improves skull segmentation accuracy.
• Enhanced performance is observed across all tested input MRI sequence combinations compared to the standard CHARM configuration.
• Improved segmentation of subcutaneous fat was achieved.

Conclusions:

• The updated CHARM pipeline offers more robust and accurate head model creation, particularly for skull segmentation.
• This advancement benefits TBS and M/EEG applications by improving the reliability of simulations and source localization.
• The enhanced atlas and parameters address limitations of commonly available MRI datasets.