dslaborg / sumo Goto Github PK

Hello, I'm wondering how to handle artifacts with the use of SUMO.

I understand that SUMO was trained only on clean data.

Would it be valid to first run the SUMO detector on all N2 stages of a recording and then exclude all the spindles occurring during an artifact? Can a huge artifact corrupt how SUMO detects the spindle on clean data near the artifact?

If I need to provide only clean data to SUMO: Does SUMO need 115s of clean data to work properly? On the GitHub you provided as demo data a single 30s epoch, so it looks like we can use SUMO properly with a shorter amount of data (I mean shorter than 115s). What is the minimum signal duration we can use to run SUMO on? This is important if we need to extract only clean segments.

Thank you !

Dear Dr.Dslaborg,
I tried to download your project, and during the debugging process, you used a training tool containing eeg_ The MAT format dataset for vector is missing. We would like to know what the form of internal data looks like. If possible, please provide it to EEG_ Mat file for vector.
Sincere thanks!

Hello, could you please tell me the specific format of MASS data used in training SUMO model? (I have applied for MASS data set, and I am replicating some code based on MASS data set)