[diss] Perustieteiden korkeakoulu / SCI

Permanent URI for this collectionhttps://aaltodoc.aalto.fi/handle/123456789/52

Browse

Browsing [diss] Perustieteiden korkeakoulu / SCI by Degree programme/Major subject "Biofysiikka ja lääketieteellinen tekniikka"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Neuromagnetic source localization using anatomical information and advanced computational methods
    (2003-06-09) Tissari, Satu
     Doctoral dissertation (article-based)
    Brain function can be studied noninvasively by magnetoencephalography (MEG) which measures weak magnetic fields outside the head. The fields are caused by neural currents, which are often estimated from the measurements. Neuromagnetic source localization results can be combined with individual anatomical information on three-dimensional reconstructions of the subject's brain based on magnetic resonance images. Applying this technique, new significant information was obtained about the continuity of visual perception during eyeblinks in humans. Individual anatomical information can also be used to form realistically shaped volume conductor models in order to improve the neuromagnetic source localization accuracy. It was shown that the whole head MEG with individual brain shaped conductor models can be used to study the cognitive processes associated with deep brain structures like memory functions in the hippocampus. Advanced computational methods were studied to facilitate the use of realistically shaped conductor models in neuromagnetic source localization. As a first step, the methods were tested with unit spheres and applied to brain shaped homogeneous conductor models using a single current dipole. The accuracy of the forward problem solution, especially near the surface of the conductor model, could be significantly improved using the Galerkin method with linear basis functions. Contrary to a common expectation, the Galerkin method was shown to be efficient and even faster than the widely used collocation method with linear basis functions for a given forward solution accuracy. Large problems could be solved quickly without the explicit formation of the dense matrix with small computer memory requirements using the iterative Bi-CGSTAB method with the precorrected-FFT method. The methods studied can be applied for MEG, EEG, MCG or ECG with multicompartment volume conductor models and with focused or distributed current source models. The forward problem solution is needed in all bioelectromagnetic source localization. The methods can be taken into routine use for neuromagnetic source localization to improve the source localization accuracy and to decrease the computational costs of accurate modelling.