| Autor: |
Victor Llobet, Aurélien Wyngaard, Boris Barbour |
| Přispěvatelé: |
Barbour, Boris, Institut de biologie de l'ENS Paris (IBENS), Département de Biologie - ENS Paris, École normale supérieure - Paris (ENS-PSL), Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)-Institut National de la Santé et de la Recherche Médicale (INSERM)-Centre National de la Recherche Scientifique (CNRS)-École normale supérieure - Paris (ENS-PSL), Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)-Institut National de la Santé et de la Recherche Médicale (INSERM)-Centre National de la Recherche Scientifique (CNRS)-Institut National de la Santé et de la Recherche Médicale (INSERM)-Centre National de la Recherche Scientifique (CNRS) |
| Rok vydání: |
2022 |
| Předmět: |
|
| DOI: |
10.1101/2022.02.08.479192 |
| Popis: |
The increasing site counts of multielectrodes used in extracellular recording preclude traditional manual spike-sorting and have led to the development of several new software suites with increased automation. It remains difficult, however, to find parameters enabling optimal processing of signals from all neuronal types simultaneously. Here, we describe a procedure for automatically post-processing and merging the outputs of multiple spike-sorting analyses, enabling the accumulation of results optimised for different activity features. This approach was motivated by and first applied to the analysis of cerebellar activity, in which the complex spikes challenge standard analyses. The core of our modular procedure involves the identification of cluster-fusion candidates and assurance that only fusions that increase a clearly defined cluster-quality metric are performed; important subsidiary steps that have also been automated include spike alignment and duplicate removal. The procedure was validated by recovery of injected spike waveforms. In tests combining analyses from Kilosort and MountainSort, our procedure yielded a 50 % increase in the detected number of Purkinje cell clusters with both simple and complex spikes. A further validation used the synthetic SpikeForest dataset approximating neocortical activity, in which our package also improved performance. Combining multiple analyses therefore offers a general method for improving spike-sorting, and furthermore Lussac reduces manual input and increases the objectivity of analyses. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
