Registration of magnetic signals from the brain using high temperature superconducting magnetometers (High-Tc SQUID; right image). EEG recordings are performed simultaneously.
This work includes:
- Development of high-Tc detector technology optimized for magnetophysiological recordings of evoked fields from hippocampal neurons in vitro
- Advancement of source-localization algorithms catered to the advanced reduction in brain activity source-detector spacing provided by high-Tc SQUID technology
- Optimization of detector and measurement environment for MEG recordings of the human brain
Project Leader: Prof. Dag Winkler
Why MEG?
MagnetoEncephaloGraphy (MEG) systems record the weak magnetic fields generated by neural currents in the brain. ElectroEncephaloGraphy (EEG) is a similar technique that measures electric fields. However, EEG yields only modest spatial accuracy because the signals are distorted by the skull and scalp. These negative effects can be reduced with surgically implanted EEG electrodes placed on the surface of the brain. MEG does not require such invasive measures because magnetic fields are not distorted by the physiology of the head. Furthermore, applying hundreds of EEG electrodes on e.g. non-cooperating infants is extremely difficult. The non-contact nature of MEG provides 300+ channel recordings easily and instantaneously.
MEG is more accurate, safer, and faster than alternative methods for studying the brain. For example, functional magnetic resonance imaging (fMRI) is one of the most commonly used neural imaging techniques. Because fMRI is an indirect measure of brain activity—it highlights parts of the brain that are consuming more oxygen than others—fMRI images are notoriously difficult to interpret, and worse: the temporal resolution of the technique is limited to seconds. MEG, on the other hand, has exceptional temporal resolution that can, in principle, be reduced below the millisecond range. Such speed is paramount to enabling the study of networks and rapid communication in the active brain. Finally, positron emission tomography (PET) has gained traction as another advanced brain-imaging modality, but this method requires expensive, difficult to handle, and potentially dangerous radioactive tracers.
OUR TECHNICAL ADVANTAGE
The first direct magnetic recordings of brain activity were performed in the 1960s with inductive coils and led to a significant leap in neuroscience research. The development of ultrasensitive sensors based on superconducting technology revolutionized the field by improving the sensitivity of MEG systems by orders of magnitude in the 1970s. MEG recordings have not changed significantly since then and the most challenging aspects for contemporary systems remain: ultra-low signal levels and solving the “inverse problem” in order to localize brain activity from the measured magnetic field values around the head.
State-of-the-art MEG systems employ superconducting quantum interference devices (SQUIDs) that are among the most sensitive magnetic field detectors. The exquisite sensitivity of these devices, however, comes at a cost: they must be cooled to 4 Kelvin, or -269 °C, because they are made of low-temperature superconducting (LTS) materials. In the 1980s, high-temperature superconductors (HTS) were discovered and soon afterward SQUIDs that operate at 77 Kelvin (-196 °C) were made with this new class of materials. Because they operate at much higher temperatures, these SQUIDs are ideal for biomedical applications. The thermal insulation that LTS detectors require keeps them from coming closer than a few centimeters of e.g. the human head. Our HTS sensors, on the other hand, allow for a reduction of this distance to less than a millimeter. The improved coupling to the neural activity responsible for the magnetic signals recorded in MEG results in a significant improvement in signal-to-noise ratios for our HTS sensors. Better signal-to-noise leads to dramatic impacts on the field e.g. enabling the study of brain signals too “quiet” to be observed before, improving accuracy in locating brain activity, enhancing the capability to discriminate multiple sources of such activity, and reducing up-front and running costs.
However, bringing the sensors closer to the head requires more sophisticated source-localization algorithms that account for varying sensor positions in accordance with varying head sizes and shapes. Furthermore, sensitivity to multiple sources of neural activity demands advanced algorithms and signal processing techniques capable of transforming the complicated data streams into easy-to-interpret visual representations of the brain as it works.
WHERE DOES THE EXTRA SIGNAL-TO-NOISE COME FROM?
The simplest sources of magnetic field are dipoles, often represented by small loops of circulating charge (while theoretically possible, magnetic monopoles have yet to be found in nature). Dipole fields decay as 1/r^3, resulting in a rapidly decaying field strength when one moves away from the source. Quadrupole and higher-order sources have even stronger spatial dependence (1/r^5 and higher). A real-world example illustrates the importance of standoff distance: when recording somatosensory activity on the surface of the human brain, our HTS SQUIDs can be placed less than 1 mm from the surface of the head. The layers of the scalp, skull, cerebrospinal fluid, &etc. add up to ~1 cm, so our HTS technology reaches a source-sensor spacing of 11 mm at most. State-of-the-art MEG, on the other hand, achieves a scalp-to-sensor spacing of 18 mm at best, resulting in a source-sensor spacing of 28 mm.
An aside: in reality, it is rare for a subject’s head to fit exactly inside the rigid helmet-shaped dewar of a state-of-the-art MEG system. This means that most of the SQUID sensors in such a system are more than 20 mm from the subject’s scalp. But we’ll give them the benefit of the doubt and stick to 18 mm.
We can calculate the signal available to HTS and LTS SQUIDs by numerically integrating the dipole and quadrupole fields over the SQUID pickup loops at their respective standoff distances. HTS SQUIDs, due to the reduction in source-sensor spacing, gain in signal strength by a factor of more than 10 for dipole and more than 50 for quadrupole sources. Physically speaking, these large factors come from the fact that the magnetic field is not only decaying but the field lines are also spreading as a function of distance.
Dr. Anders Hedström, Dept. of Clinical Neurophysiology, Sahlgrenska University Hospital, and Dr. Justin Schneiderman, MedTech West
