Caltech’s Leading Role in Ultrasound Brain–Machine Interface Advancement

Functional ultrasound (fUS) represents a significant advancement in Brain-Machine Interface technology, providing a less invasive way to control electronic equipment precisely via brain activity interpretation.

BMIs are devices that can scan brain activity and interpret it to operate an electronic device such as a prosthetic arm or computer cursor, allowing persons with paralysis to move prosthetic devices with their minds.

Many BMIs necessitate invasive operations to implant electrodes inside the brain to interpret neural activity; however, Caltech researchers devised a method to monitor brain activity using functional ultrasound (fUS), a far less invasive technology, in 2021.

Functional Ultrasound: A Game Changer for BMIs

A new study proves that fUS technology can be the foundation for an “online” BMI—one that interprets brain activity, uncovers its meaning with machine learning-programmed decoders, and governs a computer that can accurately predict motion with very little delay time.

The research was carried out in the laboratories of Caltech’s Richard Andersen, James G. Boswell, Professor of Neuroscience and director and leadership chair of the T&C Chen Brain-Machine Interface Center, and Mikhail Shapiro, Max Delbrück Professor of Chemical Engineering and Medical Engineering and Howard Hughes Medical Institute Investigator, as well as Mickael Tanter’s laboratory at INSERM in Paris, France.

Advantages of Functional Ultrasound

Andersen states that functional ultrasound represents a novel addition to the array of tools for brain–machine interfaces designed to aid individuals with paralysis. He highlights its appeal in terms of being less invasive than brain implants and not necessitating continuous recalibration. Moreover, he emphasizes that this technology’s development resulted from a collaborative effort, underscoring the need for multiple laboratories working together.

Sumner Norman, a former senior postdoctoral scholar research associate at Caltech and a co-first author on the study, notes that tools for measuring brain activity have advantages and disadvantages. While electrodes offer precise measurements of single-neuron activity, their requirement for brain implantation limits scalability to a few small brain regions. On the other hand, non-invasive techniques, such as functional magnetic resonance imaging (fMRI), provide whole-brain access but face limitations in sensitivity and resolution. Poor signal quality and an inability to pinpoint deep brain function hinder portable methods like electroencephalography (EEG).

Ultrasound Imaging Explained

Ultrasound imaging operates by sending out high-frequency sound pulses and then gauging how these sound vibrations rebound within a substance, like different tissues in the human body. The varying speeds of sound waves through these tissues and their reflection at boundaries enable the creation of images. This method is frequently employed for capturing images of a developing fetus in the womb and for various diagnostic imaging purposes.

Due to the impermeability of the skull to sound waves, incorporating ultrasound for brain imaging necessitates the installation of a transparent “window” in the skull. Whitney Griggs (PhD ’23), a co-first author of the study, emphasizes that ultrasound technology doesn’t require implantation into the brain directly. This substantially lowers the risk of infection and keeps the brain tissue and its protective dura completely intact.

Norman explains that as neurons alter their activity, their utilization of metabolic resources like oxygen also changes. Functional ultrasound relies on the bloodstream to replenish essential resources. This study utilized ultrasound to monitor variations in blood flow to specific brain areas. Similar to how the pitch of an ambulance siren alters with proximity, red blood cells modify the pitch of reflected ultrasound waves as they approach and recede from the source. Observing this Doppler-effect phenomenon enabled the researchers to capture minute changes in the brain’s blood flow within spatial regions as narrow as 100 micrometers, approximately the width of a human hair. This capability allowed simultaneous monitoring of the activity of small neural populations, some consisting of just 60 neurons, distributed widely throughout the brain.

Innovative Application in Non-Human Primates

The researchers employed functional ultrasound to gauge brain activity in non-human primates’ posterior parietal cortex (PPC). This region oversees the planning and execution of movements and has been a subject of study in the Andersen lab for many years, employing various techniques.

The animals underwent training in two tasks: one task involved planning hand movements to guide a cursor on a screen, while the other task involved planning eye movements to focus on a specific part of the screen. In executing these tasks, the animals only needed to contemplate the actions without physically moving their eyes or hands. The brain-machine interface (BMI) interpreted the planning activity in the posterior parietal cortex (PPC).

Shapiro recalls expressing admiration for the success of predictive decoding with electrodes two decades ago and finds it remarkable to witness its effectiveness with a considerably less invasive method like ultrasound.

Promising Results and Future Plans

The real-time ultrasound information was transmitted to a decoder, which had been trained through machine learning to interpret the data’s meaning. The decoder then produced control signals to guide a cursor to the intended destination specified by the animal. The brain-machine interface (BMI) achieved this successfully for eight radial targets, with an average error of less than 40 degrees.

Griggs points out the significance of the technique, which does not necessitate daily recalibration for the brain-machine interface (BMI), unlike other BMIs. To illustrate, she draws an analogy by likening it to the inconvenience of recalibrating a computer mouse for up to 15 minutes each day before using it.

The team’s next step involves examining the performance of brain-machine interfaces (BMIs) using ultrasound technology in humans. Additionally, they aim to enhance the functional ultrasound (fUS) technology to facilitate three-dimensional imaging, thereby improving accuracy.

The paper, “Decoding motor plans using a closed-loop ultrasonic brain–machine interface,” was published in Nature Neuroscience on November 30.

Read the original article on: SciTechDaily

Also read: Does the Brain Learn in the Same Manner that Machines Learn?\