neurojax-phantom-elekta

name: neurojax_phantom_elekta description: Validate Inverse Solver Accuracy using the MNE Elekta Phantom dataset.

Elekta Phantom Validation

This agent validates the localization accuracy of neurojax.source.inverse_scico using the physical Elekta Phantom dataset (available via MNE-Python).

Objectives

1. Data Ingestion: Download/Load the phantom_elekta dataset using mne.datasets.phantom_4dbti (or similar standard set).
   • Note: The standard MNE sample dataset includes phantom data. Check mne.datasets.sample or mne.datasets.hf_sef if specific phantom sets are unavailable. Use mne.dipole.get_phantom_dipoles to get ground truth.
2. Preprocessing:
   • MaxFilter (if raw).
   • Epoching (if continuous).
3. Inverse Solve:
   • Run neurojax.source.inverse_scico.solve_inverse_admm (Native ADMM) on the evoked data for each dipole.
4. Error Quantification:
   • Compute Euclidean Distance between the estimated peak source location and the known ground truth dipole location.
   • Check if error is within acceptable bounds (< 5mm for good SNR).

Instructions

1. Create/Run examples/demo_phantom_elekta.py.
2. Use MNE utilities to fetch the phantom data and ground truth coordinates.
3. Convert MNE Forward/Data structures to JAX arrays.
4. Solve and compute error statistics.
5. Generate a report results_phantom_elekta/report.md.