detect_()

Spike detection step is performed using detect_() function ("detect" command) and it performs filtering and spike detection. Multiple copies of spikes from nearby sites are grouped together. Spike waveforms (filtered and raw) and low-dimentional features are extracted during the detection step. To deal with large files, detection step is performed in multiple memory loading cycles.

Syntax

S0 = detect_(P)

Input

• Parameters struct (P) sections below
  • Recording file format
  • Pre-processing
  • Spike detection and grouping

Output

• S0: Master struct (S0)
• Global variables: tnWav_raw (raw spike waveforms), tnWav_spk (filtered spike waveforms), trFet_spk (spike features)
• Output files: _jrc.mat (contains S0), _spkfet.jrc (contains trFet_spk), _spkwav.jrc (contains tnWav_spk), _spkraw.jrc (contains tnWav_raw). See Input and output files.

Operations

• file2spk_(): load recordings from file(s) and divide the problem to the available memory size and merge the solutions
  • wav2spk_(): Filter, detect and group spikes, and extract features for each memory loading block
    • fft_clean(): Adaptive notch filter (skip if P.fft_thresh==0)
    • filt_car_(): Filter (P.vcFilter) and apply common average referencing (P.vcCommonRef)
    • detect_spikes_(): Detect spikes using a threshold setting (P.qqFactor).
      • spikeDetectSingle_fast_(): Perform spike detection for each channel using GPU
      • spikeMerge_(): Merge spikes detected at multiple neighboring sites using multiple CPU cores
    • mn2tn_wav_(): Extract spike waveforms from a list of specified site (viSite_spk) and time (viTime_spk)
    • trWav2fet_(): Extract spike features from the supplied spike waveforms (see P.vcFet)
• spk_pos_(): compute the center-of-mass position for each spiking event.
• save0_(): save the spike detection result