Spike Recording Quality Assessment

During an chronic extracellular recording, microelectrodes usually record action potentials from more than one neuron. It is usually necessary to sort the recorded potentials by shape to extract the firing times of the contributing single units. This is a challenging problem for more details, see this seminar.

We are developing a novel algorithm which solves a related, but simpler problem. Given a very brief (~1s) section of data, we determine how many neurons contribute spikes to the recording. In the case where more than one cell is detected, we also estimate how well separated the potentials are. Taken together, these measures provide an assessment of the quality of a recording.

The algorithm is very fast, and can be run in real time. We have embedded it within a control algorithm that will automatically move an array of 16 electrodes (Eckhorn matrix, Thomas Recording to find and then keep clean single unit activity.

People involved in this project


Sponsors of this study

Back to the Projects