IEEE Solid-States Circuits Magazine - Fall 2021 - 89

activity and treat the symptoms of
advanced neurological disorders.
For example, deep brain stimulation
devices [Figure 1(a)] provide electrical
stimulation to regulate dysfunctional
neural circuits, much like a pacemaker
for the brain, and are used to treat
Parkinson's disease, dystonia, and
more [2]. The devices consist of very
long electrodes that are implanted
in deep brain structures and a pulse
generator that is implanted in the
chest cavity and transmits waves of
voltage or current to targeted brain
regions. Similarly, peripheral nerve
stimulators are used to regulate physiological
functions, from modulating
blood pressure [3] to easing inflammation
in rheumatoid arthritis [4].
Neural stimulation can also be used
to provide sensorimotor feedback to
BMIs, enabling tactile and proprioceptive
reactions [5]. These technologies
significantly impact the quality of life
for many patients, but there remain
numerous opportunities for improvement
in miniaturization, intelligence,
and the invasiveness of implantation.
Implantable neural interfaces are
medical devices that require U.S. Food
Peripheral nerve stimulators are used
to regulate physiological functions, from
modulating blood pressure to easing
inflammation in rheumatoid arthritis.
and Drug Administration approval, and
they are designed by interdisciplinary
teams that, importantly, include biologists
and clinicians. Patient safety and
the invasiveness of implantation are of
paramount concern, making implant
volume and power dissipation key
to designs. Wireless ICs have paved
a path toward fully implantable and
minimally invasive neural interfacing.
The black elements in Figure 1(b) are
the components required for a bidirectional
(recording and stimulating),
wireless neural interface. A power
source, typically in the form of a battery,
antenna, or energy harvesting
element, is used to power the implant.
Wireless data transmission eliminates
cables, enabling surgeons to close
implant incisions, greatly reducing
infection risk. Mixed-signal recording
circuits amplify and digitize neural
signals from the electrodes, while stimulator
circuits and a controller provide
the needed voltage or current to
excite the neural tissue. In a closedloop
device, integrated edge compute
and machine learning [Figure 1(b), in
gray] can reduce the wireless burden
by enabling devices to make decisions
in situ and with low latencies. These
building blocks are common across
many types of BMIs and neurostimulation
implants, and the past decade
has included significant advances in
the design of all these components.
There are three major trends driving
current technological progress
in neural interfaces, including the
following:
1) sensor-IC integration for scaling
the number of electrodes
2) closing the loop
3) miniaturizing wireless implants.
Cortical Recording
Implanted System
Recording Path
Wireless Implant
LNA
ADCTx
Stimulating Path
DAC
Reference
Deep Brain
Stimulation
Electrode Interface
[6]
2 mm
(a)
(c)
FIGURE 1: (a) Two clinical neural interfaces: a percutaneous neural recording implant similar to [1], and a deep brain stimulator. (b) Components
for an implantable bidirectional (black) or closed-loop (black and gray) wireless neural interface. (c) Example neural interface electrodes.
[From left to right (adapted from Utah microelectrode array [6], SENSIGHT by Medtronic [7], ultrathin polymer threads by Neuralink
[8], high-density neural matrix [9], Stentrode by Synchron [10] and ultraflexible syringe-injectable mesh [11]).] LNA: low-noise amplifier; ADC:
analog-to-digital converter; Tx: transmitter; ML: machine learning; DAC: digital-to-analog converter; Rx: receiver.
IEEE SOLID-STATE CIRCUITS MAGAZINE
FALL 2021
89
[7]
[8]
[9]
Controller
ML
(b)
[10]
[11]
Power Supply
Rx
IEEE Solid-States Circuits Magazine - Fall 2021

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Fall 2021
