neurojax-phantom-ctf

name: neurojax_phantom_ctf description: Validate BEM Forward Model Accuracy using the MNE CTF Phantom dataset.

CTF Phantom Validation

This agent validates the accuracy of the neurojax.geometry.bem_jinns (PINN Forward Model) using the CTF Phantom dataset.

Objectives

1. Data Ingestion: Load the CTF Phantom data (known spherical geometry).
2. BEM Forward Solve:
   • Set up the BemSolver (PINN) to model the conducting sphere.
   • Compute the potential field at the sensor locations for the known active dipoles.
3. Analytical Comparison:
   • Compute the Analytical Solution (Sarvas formula for sphere) for the same dipoles.
4. Error Quantification:
   • Relative Difference Measure (RDM): $\sqrt{\sum (y_{est} - y_{ref})^2 / \sum y_{ref}^2}$.
   • Magnitude Error (MAG).

Instructions

1. Create/Run examples/demo_phantom_ctf.py.
2. Define the spherical geometry in the PINN.
3. Compare PINN predictions vs Analytical sphere model for the specific sensor layout of the CTF system.
4. Report accuracy metrics.