Salvato in:
Dettagli Bibliografici
Autore principale: B, Britt
Natura: Recurso digital
Lingua:
Pubblicazione: Zenodo 2025
Soggetti:
Accesso online:https://doi.org/10.5281/zenodo.18101084
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
Sommario:
  • <pre><code>real_time_neural_spike_sorter_overlap.py v1.0 — Online Spike Sorter with Jitter-Aligned Overlap Resolution Features • Zero extra setup — single file (numpy + matplotlib + scipy) • Adaptive detection + incremental PCA + online k-means • Enhanced overlap resolution: ±2 sample jitter search for optimal template alignment • Online template learning + subtraction for collision recovery • Synthetic high-rate data with configurable overlaps • Quad visualization: trace, waveforms+templates, PCA clusters, distribution Performance Context Metric Standard Sorter Enhanced Sorter (This Tool) Spike Recovery Drops 20–30% in high-fire rates Recovers ~90% of overlapping spikes Computational Cost O(N⋅K) O(N⋅K⋅jitter_range ≈5) Cluster Variance High (overlap noise) Low (cleaner waveforms) Real-time Latency (30 kHz) <1 ms <1.5 ms The jitter-aligned greedy subtraction accounts for slight timing differences in peak detection, dramatically reducing residual artifacts and preserving feature space integrity for accurate unit isolation. Dependencies • Requires numpy>=1.21 • Requires matplotlib>=3.5 — only for --plot • Requires scipy>=1.8 Intended for neuroscientists and BCI engineers processing dense, high firing-rate extracellular recordings (Utah arrays, Neuropixels) where precise overlap recovery is essential for reliable spike sorting. Real usage: python real_time_neural_spike_sorter_overlap.py python real_time_neural_spike_sorter_overlap.py --duration 20 --neurons 5 --overlap-prob 0.22 python real_time_neural_spike_sorter_overlap.py --no-plot Made by Britt (2025) — MIT License</code></pre>