Electromagnetic brain imaging

Present imaging methods of cerebral neuro-activity like brain functional MRI and positron emission tomography (PET) secondarily measure only average activities within a time of the second-order (low time-resolution). In contrast, the electromagnetic brain imaging (EMBI) directly measures the faint magnetic field (10-12-10-13 T) yielded by the cerebral activity with use of multiple arrayed sensors equipped on the head surface within a time of sub-millisecond order (high time-resolution). The sensor array technology to find the signal source from the measured data is common in wide areas like signal procession for radar, sonar, and epicenter detection by seismic wave. For estimating and reconstructing the active region in the brain in EMBI, the efficient method must be developed and this paper describes the direct and inverse problems concerned in signal and image processions of EMBI. The direct problem involves the cerebral magnetic field/lead field matrix and inverse problem for reconstruction of signal source, the MUSIC (multiple signal classification) algorithm, GLRT (generalized likelihood ratio test) scan, and adaptive beamformer. As an example, given are results of magnetic intensity changes (unit, fT) in the somatosensory cortex vs time (msec) measured by 160 sensors and of images reconstructed from EMBI and MRI during electric muscle afferent input from the hand. The real-time imaging is thus possible with EMBI and extremely, the EMBI image, the real-time cerebral signals, can inversely operate a machine, of which application directs toward the brain/machine interface development. (R.T.)