To extract conti-time-frequency transformations from preprocessed data

💬 Support & Discussions
1
  • MNE version: 1.0.1
  • operating system: Windows 10

Is there a guideline for how data pre-processed in mne is time-frequency transformation and how it is extracted?

image

2

Hello @PhD-GOAT, could you please elaborate what exactly you want to do? I’m not sure I fully understand from your posting. Thanks!

3

hi!! @richard

I want to preprocess raw data and extract preprocessing data for machine learning. It was confirmed that only the morlet calculation method exists in mne. But here, I wonder how to convert the time frequency and how to extract it. I am thinking about putting this into a train and test for machine learning.

4

I have to admit I’m still unsure where your problem / question arises. Could you please share some code that demonstrates what you’ve done so far and where you’re now stuck?

Best wishes,
Richard

5

Now I’m at the stage of thinking about how to code. First, the PSG EDF is converted into a wav file for each signal for convenience of use. The converted wav file is then divided into 30-second intervals, which is the sleep stage reading cycle, and classified by reading result file provided with the data. Finally, the divided files are made into spectrogram images of wav files, and the brain waves are used as band-pass filters to extract characteristics by band.

6

You can connsider mne-feature to design an efficient ML. Have a look at this accompanying tutorial

7

good, but cloning that git and run examples

—> 65 paths = download_bonn_ieeg(’./bonn_data’)
print error: BadZipFile: File is not a zip file

at plot_mne_sample_features.py

8

Yay, yet another approach to sleep stage classification :sweat_smile::sweat_smile::sweat_smile:

You may also want to consider @cbrnr’s SleepECG, or YASA, or visbrain’s Sleep.

1 Like
9

Thanks for the shoutout, I just wanted to clarify that SleepECG is intended for ECG data only (no EEG).

1 Like