A weighted iterative algorithm for neuromagnetic imaging
- Los Alamos National Lab., NM (United States)
The goal of neuromagnetic source reconstruction is high resolution 3-D mapping of the current distribution within the brain. However, the neuromagnetic inverse problem is ill-posed and typically underdetermined. The Moore-Penrose pseudoinverse provides a linear algebraic inverse calculation that simultaneously minimizes chisquare and the Euclidean norm of the component currents. Such ``minimum norm`` reconstructions tend to produce diffuse and superficial current distributions because voxels nearer the sensor array can account for more power in the data with less current than deeper voxels. We describe an algorithm that overcomes the bias of minimum norm procedures toward superficial solutions by using weights chosen to compensate for the distance dependence of magnetic signal strength. We also apply a Bayesian weighting strategy in an iterative pseudoinverse computation, to address the bias of the linear estimator procedure toward diffuse solutions. This strategy produces a progressively more focal current distribution while accomodating distributed current sources, and appears to effectively reduce the problems associated with the under-determined linear system.
- Research Organization:
- Los Alamos National Lab., NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 10174816
- Report Number(s):
- LA-UR-92-2698; CONF-921197-1; ON: DE92040244
- Resource Relation:
- Conference: 14. annual international IEEE in medicine and biology society,Lyon (France),2-4 Nov 1992; Other Information: PBD: [1992]
- Country of Publication:
- United States
- Language:
- English
Similar Records
An optimal constrained linear inverse method for magnetic source imaging
Algorithms for biomagnetic source imaging with prior anatomical and physiological information