STUDY OF WAVEFORM ANALYSIS WITH A 3-D VECTOR MEASUREMENT OF THE MEG.

点击翻译按钮获取中文摘要

In the research on the magnetoencephalogaram (MEG) measurement the magnetic field perpendicular to the scalp is widely used. There are problems of separating multiple sources overlapping in time when many distinct areas of the cortex are active. In these cases the magnetic field distribution of normal component is not helpful for estimating the location and number of sources. owing to the lack of a dipole pattern. From viewponit of waveform analysis, it is also desired a certain method to detect the transient weak signal hid by the intense signal. Concerning to overcom the previous ploblems, new methodology is necessary to develop in both of not only signal processing but also SQUID instrumentation. First, we have developed a three dimensional (3-D) second-order gradiometer connected to three superconducting quantum interference devices (SQUIDs) for vector measurement of the MEG that can detect magnetic field components perpendicular and tangential to the scalp simultaneously. Second, we proposed a new method of source estimation algorithm using spatio-temporal data of 3-D MEG and singular value decomposition analysis. In this algorithm the number of sources are decided by number of dominant singular value of spatio-temporal 3-D MEG data. The source localization is done as pattern matching by signal subspace of singular vectors and projection of an estimating source.This proposed method can estimate two and/or three sources overlapping in time, which are 2 cm apart and 15 degree in phase difference. effectively.