MEG is routinely used in clinical practice to prepare for certain neurosurgery interventions targeting issues such as tumors and drug-resistant epilepsy. Initial tests suggest that it could also be useful for the early diagnosis of various neurological disease, including Alzheimer's disease. But the need to cool the SQUID sensors to 4 K and the system's imposing magnetic shielding make it very costly, and not user friendly. Consequently, there are only around a hundred such systems in the world, including five in France.

• No heating, no cryogenics and lightweight magnetic shielding

CEA-Leti has chosen a technological direction that considerably reduces these constraints while retaining the investigative power of MEG. Its sensors use helium-4 excited to a metastable state as their sensitive element. They operate at room temperature and are placed in contact with the skull, as close as possible to the signal being detected. This is not possible with SQUID sensors, which are too cold, or alkali-based sensors at 150°C, another solution examined by several teams.

These new sensors have no need for cryogenic cooling and require five times less magnetic shielding than SQUIDs, making it possible to expand access to magnetoencephalography and extend its use. They have 20 times more bandwidth than heated alkaline atoms (2 kHz compared with 100 Hz), which is vital for imaging rapid phenomena such as epilepsy.

Mag4Health has been recently launched with the aim of developing a 64-sensor MEG helmet covering the whole head by 2023, and then a MEG system to be marketed for clinical use by 2026.

Two five-sensor prototypes have already been built. They are currently being tested at six university hospitals.