Thread by @SumnerLN

Thrilled to share our latest work out today in @NeuroCellPress using functional #ultrasound (fUS) neuroimaging to decode movement from single trials in non-human primates 🐒 - a new benchmark for fUS & a precursor to closed-loop #BCI.

authors.elsevier.com/c/1cnFe3BtfGzj…

#tweetprint 🧵 1/n

Congrats to all co-authors @DMarescaLab, @VasileiosChris2, @MisterGriggz, @CharlieDemene,

dream-team collaborators @TanterM, @CharlieDemene (@INSERM @PhysMedParis, @ESPCI_Paris),

& PIs/ #neuroscience luminaries Richard Andersen & @mikhailshapiro (@Caltech, @CaltechN) 2/n

Background: Brain-machine interfaces (BMI) are powerful devices for restoring function to people living with paralysis. Already, our patients control external devices like computers and robotic limbs with incredible fidelity. 3/n

gph.is/g/ZkxRqpA

These interfaces are based on intracortical electrophysiology, which provides direct access to neuron spiking and local field potentials. However, these techniques cause acute and chronic damage to brain tissue, limiting their use to niche populations. 4/n

Here, we introduce single-trial decoding (a precursor to a brain/computer interface #BCI) based on functional ultrasound (fUS) neuroimaging - a recently developed minimally invasive & portable technique. 5/n

For an intro to fUS imaging, check the newest review from @DeffieuxThomas @CharlieDemene, & @TanterM sciencedirect.com/science/articl…

...or just gawk at this incredible video from @AlexandreDizeux & @ESPCI_Paris 🤯
6/n

First, we trained two non-human primates to perform memory-delayed eye and/or hand movements while we recorded from the posterior parietal cortex (PPC).

PPC is a visuomotor association area involved in movement planning. 7/n

We found LIP activity during movement planning that was direction-specific. This finding agrees with previous fMRI findings (Kagan PNAS 2010, Wilke PNAS 2012) - but fUS was nearly 10x as sensitive! 8/n

First, we showed the direction the animal *intended* to move (before executing the movement) using only the changes in fUS signal. Single-trial decoding! 9/n

Now, we show that the fUS is capable of much more! We simultaneously decoded
1) when the movement was about to occur,
2) which direction the movement will be, and
3) which effector the animal will use! 10/n

High decoding rates persisted across the movement preparation and execution phases. I.e. we could decode movement phase data using decoders trained on memory-delay data and vice-versa. 11/n

That means the information encoded in PPC was consistent across phases. This is evidence that we are decoding goal information (not position or trajectory).

This is important because it allows us to circumvent delays introduced by neurovascular coupling 12/n

This paper makes some important advances for large-scale, high-resolution recording and opens the door for single-trial decoding and #BCI using ultrasound.

... but what excites me most is that fUS is a *very* young technique; we're just getting started. 13/13

The link above is open access. Cell.com link here: cell.com/neuron/fulltex… Thanks
@CellPressNews!