【脑机接口】利用MNE进行EEG数据预处理（SEED数据集）_mne eeg数据处理平台

简单写下毕设里用到的预处理步骤
数据集详细信息可查看：SEED数据集
预处理分为以下6步：

1. 数据读取
2. 坏导插值
3. 滤波+重参考
4. 去伪迹
5. 分段

---

需要用到的python库

import numpy as np
import matplotlib.pyplot as plt
import mne
import os, sys

from scipy.io import loadmat
from pathlib import Path
from PyEMD import EMD
1
2
3
4
5
6
7
8

1.数据读取

SEED数据集中的Preprocessed_EEG文件夹包含了原始的EEG数据，本文将针对这些原始EEG数据进行预处理。
由于SEED数据集中的EEG数据是.mat格式储存的，因此需要转换成MNE的Raw格式

file_path = '...' #数据集地址
raw_file = loadmat(file_path) #scipy.loadmat()
std1020_eeglab = mne.channels.read_custom_montage("Standard-10-20-Cap81.locs",coord_frame='head')
ch_names = ['Fp1','Fpz','Fp2','AF3','AF4','F7','F5','F3','F1','Fz','F2','F4','F6'
            ,'F8','FT7','FC5','FC3','FC1','FCz','FC2','FC4','FC6','FT8','T7','C5'
            ,'C3','C1','Cz','C2','C4','C6','T8','TP7','CP5','CP3','CP1','CPz','CP2'
            ,'CP4','CP6','TP8','P7','P5','P3','P1','Pz','P2','P4','P6','P8','PO7'
            ,'PO5','PO3','POz','PO4','PO6','PO8','CB1','O1','Oz','O2','CB2']
for key in raw_file.keys():
	if key == '__header__' or key == '__version__' or key == '__globals__':
        print(key,' ',raw_file[key])
        pass
    else:
        print(key, raw_file[key].shape, raw_file[key].ndim)
        print(raw_file[key].shape)
        info = mne.create_info(
            ch_names = ch_names,
            ch_types = ['eeg' for _ in range(62)], 
            sfreq = 200 #采样频率
        ) 
        raw = mne.io.RawArray(raw_file[key], info) #生成raw
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

输出：

dict_keys(['__header__', '__version__', '__globals__', 'ww_eeg1', 'ww_eeg2', 'ww_eeg3', 'ww_eeg4', 'ww_eeg5', 'ww_eeg6', 'ww_eeg7', 'ww_eeg8', 'ww_eeg9', 'ww_eeg10', 'ww_eeg11', 'ww_eeg12', 'ww_eeg13', 'ww_eeg14', 'ww_eeg15'])
__header__   b'MATLAB 5.0 MAT-file, Platform: PCWIN64, Created on: Sun Dec 01 11:40:07 2013'
__version__   1.0
__globals__   []
ww_eeg1 (62, 47001) 2
(62, 47001)
Creating RawArray with float64 data, n_channels=62, n_times=47001
    Range : 0 ... 47000 =      0.000 ...   235.000 secs
Ready.
1
2
3
4
5
6
7
8
9

2.坏导插值

首先要找出坏导（电极损坏或者其他原因），然后利用临近的电极数据对坏导进行插值处理
需要对上一步中生成的Raw数据进行可视化

scalings = {'eeg': 50}
raw.load_data()
raw.plot(scalings=scalings)
plt.show(block=True)
1
2
3
4

从下图中可与看出P1明显是坏导，直接点选P1通道进行坏导标记

选中之后关闭视窗
终端输出：

`Channels marked as bad:
['P1']`
1
2

之后进行插值处理

raw_pass = raw.copy()
raw_pass.load_data()
raw_pass.set_montage(std1020_eeglab,on_missing='warn')
raw_pass.interpolate_bads()
1
2
3
4

插值完成后效果如图所示

3.滤波

包括0.3 ~ 50Hz带通滤波与50Hz陷波滤波（0.3 ~ 49Hz带通滤波效果应当一致）

raw_pass = raw_pass.filter(l_freq=0.3, h_freq=50,method='fir')
raw_pass = raw_pass.notch_filter(freqs=50)
raw_pass.plot_psd(fmin=0, fmax=52,sphere=(0,0,0,0.11))
raw_pass.plot(scalings=scalings)
plt.show(block=True)
1
2
3
4
5

4.重参考

采用62导全脑平均参考

raw_ref = raw_pass.copy()
raw_ref.load_data()
raw_ref.set_eeg_reference(ref_channels='average')
1
2
3

5.去伪迹

利用ICA独立成分分析去除眼电和肌电伪迹

ica = mne.preprocessing.ICA(n_components=48, 
                                noise_cov=None, 
                                random_state=None, 
                                method='infomax', 
                                fit_params=None, 
                                max_iter='auto', 
                                allow_ref_meg=False, 
                                verbose=True)
ica.fit(raws)
eog_idx_auto, eog_scores = ica.find_bads_eog(raws,ch_name=['Fp1','Fpz','Fp2','AF3','AF4','F7','F5','F3','F1','Fz','F2','F4','F6','F8','FT7','FC5','FC3','FC1','FCz','FC2','FC4','FC6','FT8'])
muscle_idx_auto, scores = ica.find_bads_muscle(raws)
print(f"Automatically found eog artifact ICA components: {eog_idx_auto}")
print(f"Automatically found muscle artifact ICA components: {muscle_idx_auto},{scores}")
ica.plot_sources(raws, show_scrollbars=True)
plt.show(block=True)
#time.sleep(199)
raw_recons = raws.copy()
raw_recons = ica.apply(raw_recons)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

检测出的眼电成分与肌电成分

Automatically found eog artifact ICA components: [0, 1]
Automatically found muscle artifact ICA components: [1, 2, 4, 6, 7, 14, 15, 16, 17, 18, 19, 21, 23, 24, 25, 26, 28, 29, 30, 32, 33, 34, 35, 36, 37, 39, 40, 41, 43, 44, 45]
1
2

选中所检测出的伪迹成分

关闭视窗后完成伪迹去除

…明天再继续写