Emulated-Inc/calcium-spike-inference-gcamp6f
Calcium spike inference, GCaMP6f Two-photon and one-photon calcium-imaging recordings with ground-truth spikes, GCaMP6f only, prepared for a spike-inference task. Eighty-one sessions across six behavioural paradigms, four quality levels (typical, low SNR, high SNR, high motion) and two modalities, with per-session frame rate (6.1 to 30.9 Hz), duration (50 to 403 s) and cell count (41 to 140) all varying. heldout/heldout-000.npz ... heldout-026.npz 27 sessions: traces only, no… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/calcium-spike-inference-gcamp6f.
Calcium spike inference, GCaMP6f
Two-photon and one-photon calcium-imaging recordings with ground-truth spikes, GCaMP6f only, prepared for a spike-inference task. Eighty-one sessions across six behavioural paradigms, four quality levels (typical, low SNR, high SNR, high motion) and two modalities, with per-session frame rate (6.1 to 30.9 Hz), duration (50 to 403 s) and cell count (41 to 140) all varying.
heldout/heldout-000.npz ... heldout-026.npz 27 sessions: traces only, no labels
heldout/manifest.csv id, modality, n_cells, n_frames, fps, duration_s
train/train-000.npz ... train-053.npz 54 sessions: traces plus spike_counts
train/manifest.csv id, condition, quality, modality, n_cells, n_frames, fps, duration_sEach npz holds time (frames,) in seconds, dff_corrected (cells, frames) the neuropil-subtracted percentile-baselined dF/F that a pipeline would hand to a spike-inference method, dff_raw the same without neuropil subtraction, F_raw raw ROI fluorescence in photon units, F_neuropil the annulus fluorescence, and the cell positions cells_y_um, cells_x_um, cells_depth_um, cells_roi_radius_um. Training files add spike_counts (cells, frames) int16, the number of spikes in each frame. Never assume a fixed frame rate: use time.
The held-out labels and grid positions are not published.
Licence
CC BY 4.0. Attribute as "Calcium spike inference, GCaMP6f, Emulated".
