A low-dimensional representation for individual head geometries.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Modeling

Background:

• Individual head models are crucial for accurate source localization and electrical stimulation in neuroelectric studies.
• Current methods often rely on complex geometrical models extracted directly from anatomical images, which can be computationally intensive.

Purpose of the Study:

• To develop a novel, efficient method for representing individual 4-shell Boundary Element Method (BEM) head models.
• To enable the creation of personalized head models that can be fitted to individual head data.

Main Methods:

• Utilized dimensionality reduction techniques to decrease the complexity of head models from thousands to tens of parameters.
• Developed a method to represent individual 4-shell Boundary Element Method (BEM) head models using a reduced parameter set.
• Demonstrated the ability to fit these reduced models to individual head data, such as head shape and impedances.

Main Results:

• Successfully represented complex individual head geometries using a significantly reduced number of parameters.
• Achieved satisfactory results with approximately 15-20 components for individual head model representation.
• Showcased the model's capability for fitting to diverse individual head data.

Conclusions:

• The proposed dimensionality reduction method offers an efficient approach for creating personalized BEM head models.
• This technique facilitates accurate neuroelectric investigations and applications by simplifying complex geometrical representations.
• The reduced-parameter head models are suitable for individual head model creation and fitting.