Neural Spike Identification From Extracellular Electrophysiological Measurements in vivo

Neurons can produce various spike waveforms depending on the neuron type, distance and orientation to the electrode. Neural spike identification consists of detecting and sorting spikes in neural activity recordings. This is done to analyze brain activity and understand how neurons communicate.

In this project, we are developing a self-supervised neuron tracking method for chronically implanted floating sparse arrays across days. The overall pipeline includes, spike detection, pre-processing, temporal alignment, electrode drift learning and neuron ID query stages. The outcomes of this project will enable:

• Neural activity tracking across days/months
• Learning of a distance metric tailored to neural activity
• Minimizing manual neuron annotation efforts
• More explainable and accurate brain-machine interfaces (BMIs)
• Exploration of multi-tasking capabilities of neurons