Mission:
To bring together designers and power users of spike sorting algorithms for high-count electrodes. We aim to leave the workshop with a plan for how we can as a community develop fully-automatic spike sorting for Neuropixels electrodes.
In particular, the workshop will address the following:
-
Transitioning from manual curation to full automation
-
Technical challenges, especially electrode drift in both chronic and acute recordings
-
Developing metrics and benchmarks for quality control
-
Discussion of ground truth methods
Goals:
Participants will be asked to come out of the workshop with a plan for how to collaborate on making fully automatic sorting a reality, with a focus on next steps that are not restricted to describing existing algorithms and their results, but on how to improve them.
In particular, we aim to:
-
Reach consensus on metrics to judge sorting quality, both for simulated ground truth data and for general recordings without ground truth
-
Identify datasets for algorithm evaluation OR a process of how to incorporate additional datasets, especially in different brain regions
-
Identify candidate algorithms for drift correction and identify people to implement and test them
