METU Student Chapter of SIAM
General Seminars
Speaker: Neslihan Özmen, Fatma Yerlikaya, Doğa Gürsoy
Title: "The Inverse Problem of Magnetoencepkalography: Sourse Localization and The Shape of Ball "
Date: May 25, 15:40
Place:Institute of Applied Mathematics S209
Abstract:
Magnetoencephalography (MEG) is a non-invasive neurophysiological technique that measures the magnetic fields generated by neuronal activity of the brain. The inverse problem of magnetoencephalography (MEG) is to estimate impressed currents from observations of magnetic fields outside the skull. A common way to model impressed currents is to generate a grid that covers the region of interest in the brain and attach mutually orthogonal electric dipoles with unknown amplitudes at each grid point. The MEG inverse problem of reconsructing electrical sources in human brain is an ill-posed problem. Users of MEG are faced with a vast array of inverse methods that can be used to process their data. For solving this problem there are some regularization techniques (i.e., Tikhonov regularization).
Speaker: Nurgül Gökgöz, Serdar Tanıl, Ahmet Onur
Title: "Iterative Methods for Discrete Tomography Implementation & Comparison Kaczmarz’s Method and Conjugate Gradient Least Squares Method "
Date: May 25, 15:40
Place: Institute of Applied Mathematics S209
Abstract:
In this study, the main objective is to reconstruct an image from its projections using some means of tomography. To do this Kaczmarz method and conjugate gradient least squares method were used. In order to illustrate the two methods, a software was written in java language. This software gets an image file as an input and creates projections of it from different directions. Then it tries to reconstruct the image using Kaczmarz and conjugate gradient least squares methods. The result shows that as the number of projection increases the accuracy of reconstruction increases. Also in a brain image, Kaczmarz gives better results than conjugate gradient least squares with the same number of projections.
Full texts of seminars are avaliable.
The Inverse Problem of Magnetoencepkalography: Source Localization and The Shape of Ball
Iterative Methods for Discrete Tomography Implementation & Comparison Kaczmarz’s Method and Conjugate Gradient Least Squares Method
"Lectures on Cooperative Game Theory", by Stef Tijs (Tilburg
University), November 2006
| 
