Master thesis presentation by Fredrik Boldizar and Una Slipac on 30 May
On Tuesday 30 May, Fredrik Boldizar and Una Slipac will be presenting their master thesis with the title “Optimizing design of on-scalp MEG systems” here at MedTech West. Supervisors: Justin Schneiderman and Bushra Riaz.
Where: Conference room, MedTech West, Sahlgrenska University Hospital, Röda stråket 10B
When: Tuesday May 30, 10:00
Magnetoencephalography (MEG) is a non-invasive functional neuroimaging method. The advantages of MEG include high spatial and temporal resolutions which give precise, detailed information about the neurological activity in the brain. State-of-the-art MEG systems use low critical temperature superconducting quantum interference device (low-Tc SQUID) sensors.
A major limitation of commercial MEG related to the low-Tc sensors on which it relies is the need for an insulation space of ∼20mm between the sensors and the scalp of the subject,which in turn increases the distance to the magnetic sources under study, i.e. neural currents in the brain. The insulation also decreases the flexibility of the system, so much so that anything other than a rigid helmet design is impractical. New advances in sensor technology such as the more precise high-Tc SQUIDs have made it possible to significantly reduce this insulation between sensor and scalp, from a few cms down ∼1 mm. This allows for new MEG system designs that are flexible and more suitable for different head sizes and shapes. The aim of this project is to evaluate different configurations of on-scalp MEG sensor layouts that may be viable alternatives to a rigid helmet system which are more adaptable for different subjects.
The layouts were simulated on four different subjects, two primary and two secondary subjects. Simulations were done using matlab, python and C, with software toolkits MNE-C and MNE-Python. Within this project, 8 different on-scalp layouts were simulated and compared to one another, the commercial Elekta Neuromag system, and the four on-scalp layouts previously done by Riaz et. al. The metrics used to evaluate the new MEG arrays were total information capacity (Itot) and Spatial Information Density (SID) maps. Two different sensor noise levels were chosen, an optimistic one (10 fT/Hz) and a conservative one (50fT/Hz). There were two approaches used when designing the new layouts: sets of similar layouts based on a simple geometrical shape and then distributed around the head and segmentation of a helmet design into parts. These shapes were then covered with sensors using a Chebyshev net algorithm.
Itot results indicate the Elekta system does not perform as well as most of the high-Tc layouts. This held true for both the optimistic and conservative noise levels. However, SID-maps provide evidence for the high-Tcs only being viable when using the optimistic sensor noise level of 10 fT/Hz. Out of these results, the square-shaped cryostats and the segmented helmet were noteworthy, both of which ranked at the top according to all metrics.
Designing a fully adaptable MEG sensor layout that is still viable performance-wise is a difficult task, and can be done in a number of ways. If using a large number of commonly shaped cryostats, then a square shape is preferred. When using the optimistic sensor noise level, a segmented helmet design was found to be most efficient regarding number of sensors and total information.