Speaker
Description
Neural decoding is a critical task for understanding the function of the brain and providing solutions for neurological injury and disease. Two-photon calcium imaging has been a promising recording technique to observe a large population of neurons; however, decoding from two-photon calcium images is challenging because of the indirect and nonlinear representation of neural activity, low sampling rates, and slow kinematics. Here, we present the approach of using a recurrent neural network encoder-decoder to decode the limb positions of a running mouse from two-photon calcium images. The neural network could decode limb coordinates sampled at 30 Hz from two-photon calcium images sampled below 8 Hz with the root mean squared errors of 25.35 pixels (3.80 mm). Information about all four limbs (contralateral and ipsilateral front and hind limbs) could be decoded from a single cortical hemisphere. A fraction of the most informative neurons yielded higher decoding accuracy than randomly-sampled neurons. Nevertheless, overall accuracy was directly proportional to the number of neurons used to decode. This study validates the feasibility of using calcium imaging to decode continuous behavior variables with a higher sampling rate for understanding brain function and brain-machine interfaces.
