The Neuromag system installed at UPMC -- Manufactured by Elekta Instruments in collaboration with Helsinki Technical University -- contains 306 superconducting sensors mounted within a low temperature container (called a dewar) shaped to allow the head to be placed adjacent to the sensors. Thus 306 separate recordings of magnetic activity may be simultaneously made. The spatial distributions of the magnetic fields recorded by these sensors are then computationally analyzed to localize the sources of the activity within the brain.
The most frequently used computational model is that of a current dipole. The computed locations of the sources are than superimposed on anatomical images, such as MRIs or CTs, to provide information about the relationship between structure and function in the brain. Not all studies require the merging of anatomical and MEG data to be useful. For example, predictive information about recovery from brain concussions may be obtained without merging the data. Many other computational models exist that are tuned for specific goals, such as distributed methods that are very good at looking at the patterns of activity across the entire cortical surface. Alternatively, beamformer techniques allow a concentration on a specific brain region, maximizing the accuracy of source analysis in that area. Which source localization method is chosen depends on the goals of the analysis application, the hypothesized depth and location of the targeted brain regions and the hypothesized simultaneity of activity across the brain.
With MEG, signals are extremely small, on the order of 100-1000 femtoTesla, which are several orders of magnitude smaller than other signals in the environment. Stray electromagnetic signals from outside sources, such as the electrically controlled door in the hospital hallway or a helicopter landing on the roof, can obscure the brain signals even with signal processing. Thus, the Neuromag system is installed in a specialized triple-ply magnetically shielded room (MSR) to eliminate as much magnetic interference as possible.
Housed inside the MSR is the MEG sensor, which is comprised of an array of detector coils cooled with liquid helium (-269° C) and contained within an insulating dewar. All devices and materials within or near the dewar must be non-magnetic and must not produce magnetic fields that can interfere with the recordings. This necessarily eliminates patients and participants with metal in their bodies (pacemaker, braces, etc) from being able to engage in MEG data collection.
The interior of the MSR is lit and has a quiet, soothing atmosphere. During a MEG study, the patient is comfortably positioned within the MSR on a patient support system. Research studies typically seat the participant in a specialized chair, while for most clinical studies the support system is configured as a bed, and the patient lies on the bed in a supine position. The dewar is rotated into position such that the detector array surrounds the patient’s head. A head and neck support further ensures patient comfort. While in the MSR, the patient is monitored by closed circuit monitor and intercom at all times.
In addition to collecting MEG measurements, simultaneous EEG data collection is possible. EEG electrodes may be applied individually or in the form of a multi-channel cap (32-, 64-, or 128-channels). Additionally, for almost all patients/participants pairs of EEG electrodes may be attached to monitor eye movements (EOG), cardiac signals (ECG), or muscular movements (EMG).
The electronics system necessary for recording the magnetic activity from a patient is located outside the MSR. These systems include the MEG amplification and data-acquisition system, a 128 channel EEG recording system, and electrical, visual and auditory stimulation system controllers.
Also located outside the MSR is a computer that interfaces with the MEG machine, on which the staff can view the real time MEG data along with the EEG and stimulus marker data. An additional computer console is provided to drive the experimental paradigms (typically using E-Prime or Matlab), and various computer workstations equipped with software for analyzing the data are also available.
A comprehensive array of highly effective stimulus components are available for visual, auditory, and somatosensory stimulation. These stimulus devices are non-magnetic, and thus do not interfere with the extremely sensitive magnetic measurements of MEG. Complex stimulus patterns may be generated to support both clinical and research studies.
The magnetically-silent visual stimulus delivery system consists of a high-performance commercial projector capable of projecting high-resolution images through an opening in the MSR wall, thereby eliminating electromagnetic interference. The image can be viewed from either the seated or supine position. A non-magnetic auditory amplification and stimulus delivery system is provided by way of an amplifier, attenuator and electromechanical transducer located outside the MSR. The non-magnetic somatosensory stimulator applies tactile stimuli through electrical stimulation. The timing of the electrical stimulation can be controlled through either external or internal timing signals.
When responses are required from the patient/participant, there are two available options. Two optically driven button response pads are available (one for each hand with just one button on each pad), as well as magnetically safe response gloves with buttons embedded at the end of each finger, allowing for up to 10 separate responses. A magnetically safe joystick has also been developed and used, and methods for collecting vocal and motor responses are currently in development. Regardless of the response type, signals relating to participant/patient responses travel to the MEG data file where they are recorded and later used in the analysis phase to synchronize brain responses that all belong to a single type of stimulation and condition.
During the MEG examination, Head Position Indicator (HPI) marker coils are placed on the patient’s head, utilized to record the position of the head during the exam by running a small amount of current through each coil. The nasion and the pre-auricular points are marked for identification during the digitization process, wherein a digital representation of the HPI coils, anatomical landmarks, and the shape of the head is generated. If an EEG cap or single electrodes are used, these may also be digitized. This information is later used to coregister the MEG source localization with the patient’s structural MRI.
Many software packages are available to complete the different phases of data analysis. Several workstations with a subset of these packages are available at the UPMC Brain Mapping Center to facilitate clinical and research analysis. Elekta provides software with the capability to perform the following analysis steps and more: temporal filtering, spatial filtering (SSS, tSSS, and SSP), PCA, artifact rejection, averaging, plotting, BEM creation software, single and multiple dipole fitting, minimum norm estimate, and minimum current estimate. Freesurfer and MNE-Suite software, developed at the Martinos Center at Harvard University, is also available and offers temporal filtering, spatial filtering (SSP), PCA, artifact rejection, averaging, plotting, covariance analysis, BEM creation software that can cortically constrain to the gray matter, single dipole fitting, minimum norm estimate, dSPM, and sLORETA. Additional analyses, such as frequency or spectrum analyses, can be performed using some pre-established software packages and with matlab. Matlab is also made available on the UPMC Brain Mapping Center analysis computers.