It can happen that the bbad channel detection is failing, especially in the case of noisy channels (the threshold about the PSD above 0.8 Nyquist).
For this we have implemented a function in ibldsp: channel_labels, _ = ibldsp.voltage.detect_bad_channels(raw, fs=sr_ap.fs, display=True)
That displays a snippet of raw data along with the features used for bad channel detection.
This widget is quite useful in determining the thresholds manually in case of a collapse due to HF noise, and should be ported to spike interface.
Cf:https://github.com/int-brain-lab/iblcourse/blob/main/2025-cajal-neural-circuits/01_exploring_raw_data.ipynb
It can happen that the bbad channel detection is failing, especially in the case of noisy channels (the threshold about the PSD above 0.8 Nyquist).
For this we have implemented a function in ibldsp:
channel_labels, _ = ibldsp.voltage.detect_bad_channels(raw, fs=sr_ap.fs, display=True)That displays a snippet of raw data along with the features used for bad channel detection.
This widget is quite useful in determining the thresholds manually in case of a collapse due to HF noise, and should be ported to spike interface.
Cf:https://github.com/int-brain-lab/iblcourse/blob/main/2025-cajal-neural-circuits/01_exploring_raw_data.ipynb