,
11 tweets,
4 min read
Read on Twitter
Our new preprint is now up! Online decoding of arm reach direction using 2P calcium imaging in NHP with @djoshea, Xulu Sun, @shenoystanford @KarlDeisseroth, @JimMarshel. @samvesuna, @weallen1, Maneesh Sahani, Eyal Seidemann, and many others. biorxiv.org/cgi/content/sh… 1/10
High level goal: Develop 2P calcium imaging for macaque motor cortex. This approach complements electrical recording by enabling 1) dense sampling, 2) genetic targeting, 3) sampling different neurons by moving the lens, 4) access to many neurons simultaneously, and others. 2/10
In addition to showing that we could image neurons during a motor behavior, we also found that we could record from neurons in deep layers by imaging signals from dendrites that project to upper layers. 3/10 
These dendritic signals were modulated by reaching movements and directionally tuned with high SNR. They also show stable directional tuning over weeks. 4/10
Here are signals from three dendrites. The different colors represent the fluorescence signals for reaches to one of four targets, averaged across trials. 5/10
We used this system to implement a proof of concept optical Brain-Computer Interface (oBCI), where we used low-latency image processing to decode a monkey’s reach within about 10ms of acquiring our last frame of imaging. 6/10
We think of optical imaging as an important tool to address questions that are challenging to answer with electrical recording but are critical to the design of next-generation neural prosthetics to help people. 7/10
There are a lot more interesting details packed in the paper - check it out here: biorxiv.org/content/10.110… 8/10
We’ve also uploaded the C++ and Cython/10umpy code we used for the low-latency decoder to github.com/djoshea/obci, along with a technical walkthrough of how the low-latency pipeline works. 9/10
And of course, none of this would have been possible without major contributions from a huge list of collaborators at Stanford, UT Austin, and University College London. 10/10
while you're here - follow @SamVesuna's new account
