Advertisement

Blog

imec MEMS-based Neural Probe: Striving for less invasive brain monitoring

Recently I became aware that imec, the research and innovation hub in nanoelectronics and digital technologies, has designed and fabricated a demonstrator of a neural probe with industry-leading electrode density and a pressure sensor based on micro-optomechanical systems (MOMS) technology. The design uses the best of MEMS and Photonics technology.

The neural probe demonstrator is designed and fabricated in silicon chip technology and is suitable for beyond the state-of-the-art bio-interfaces and implants out there now. This development looks like it will greatly change the neuroscience world. The probe incorporates innovations leading to a better understanding of the brain and can help tackle human brain diseases.

The pressure sensor exhibits high measurement precision over a large pressure range while being compact, resistant to electromagnetic interference (EMI) and having multiplexing capabilities. This sensor can be used in applications that require high-quality sensing, particularly in medical and life sciences.

Existing pressure sensors that measure parameters such as altitude and depth, or to engage in flow sensing are currently based on either MEMS or optical fiber technologies. MEMS-based pressure sensors are popular since they have good performance and small size. Optical fiber sensors are suitable for use in harsh environments which can be characterized by EMI or high temperatures; their weakness is that they make for less integrated and more complex/high cost systems.

With the sensor, imec researchers have demonstrated a root mean square (rms) precision lower than one Pa across a large range that could easily reach 100 kPa.

Here is the problem with existing neuroprobe designs:

Neural probe design needs to have high density and high number of electrodes because of large-scale recording from individual neurons in multiple brain areas. Here, you want to minimize damage to brain tissue, the “shank”, which is the part that gets implanted in the brain, needs to be very narrow and as thin as possible.

A far more dense, simultaneous readout, coupled with a smaller compact design architecture needs to be developed. See Figure 1 for the difficulties with existing probes.

Figure 1

On the left side of this image is shown the existing neural probe architecture that has a shank wiring 'bottleneck'; On the right side of the image is a probe cross-section containing six metal layers. (Image courtesy of Reference 1)

On the left side of this image is shown the existing neural probe architecture that has a shank wiring ‘bottleneck’; On the right side of the image is a probe cross-section containing six metal layers. (Image courtesy of Reference 1)

Existing probe architectures have a large number of tiny active electrodes that locally amplify/buffer the neural signals. One problem here is that with such limited space, the CMOS pixel amplifiers (PA) underneath the electrodes cannot have any additional circuitry that is needed on the front end—most of the signal processing is done in the ‘base’ of the probe instead of being close to the PA. These existing active and passive neural probes use a dedicated single metal line per electrode whose function is to send the signal to the base circuitry. This architecture greatly limits the number of simultaneous recording electrodes to the number of metal lines fitted in the cross section of the shank as can be seen in Figure 1.

A better design would include the use of time division multiplexing and techniques that reduce the associated noise right at the PA. A radical new architecture is needed to overcome this fundamental bottleneck and achieve a denser simultaneous readout.

In my upcoming EDN article, I will show you how the team in Reference 1 achieved this seemingly impossible task with some very creative design architectures as well as using a creative CMOS process to achieve twice the increase in number of simultaneous recording channels that state-of-the-art existing probes have with comparable power and noise performance. I will also go into depth regarding the MOMS pressure sensor technology.

References

1 Time multiplexed active neural probe with 678 parallel recording sites, Bogdan C. Raducanu, Refet F. Yazicioglu, Carolina M. Lopez, Marco Ballini, Jan Putzeys, Shiwei Wang, Alexandru Andrei, Marleen Welkenhuysen, Nick van Helleputte, Silke Musa, Robert Puers, Fabian Kloosterman, Chris van Hoof, Srinjoy Mitra, IEEE

2 comments on “imec MEMS-based Neural Probe: Striving for less invasive brain monitoring

  1. Jimelectr
    November 23, 2017

    Are you talking about pressure sensors or voltage sensors?

    Say, I understand the Analog Afficionados dinner will be on a Saturday instead of Sunday in 2018.  Yea!  Now I have a better chance of coming up from So Cal to attend!

  2. Steve Taranovich
    November 25, 2017

    @Jimelectr—This is a pressure sensor

     

    The Analog Aficionados dinner will be on Sunday, February 11, 2018

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.