Dreamer脑电情感识别数据集数据提取

最新推荐文章于 2024-09-19 11:57:26 发布

乔小乔丫

最新推荐文章于 2024-09-19 11:57:26 发布

阅读量2.1k

点赞数 15

分类专栏： EEG学习文章标签： python pandas 开发语言

本文链接：https://blog.csdn.net/m0_71596228/article/details/136460784

版权

EEG学习专栏收录该内容

6 篇文章

订阅专栏

该文章介绍了如何从DREAMER数据集中提取和处理EEG和ECG信号，以及被试对电影片段情绪的评分，使用Python脚本将数据转换为numpy数组并分别保存为npy文件。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

DREAMER数据集包含了一个实验的被试的评分和生理记录，23个志愿者看了由Gabert-Quillen等人挑选和评价的18个电影片段。EEG和ECG信号被记录，并且每个被试通过报告5个点量表上感觉到的唤醒，效价和支配度来对他们的情绪打分。

下面是Dreamer数据提取程序，并把每个人的数据保分别存到npy文件中

import scipy.io as sio
import seaborn as sns
import pandas as pd
import numpy as np
import time
import mne
path = "E:/DREAMER.mat"
raw = sio.loadmat(path)
participant =0
video = 0
electrode = 0
B, S = [], []  
basl = (raw["DREAMER"][0, 0]["Data"][0, participant]["EEG"][0, 0]["baseline"][0, 0][video, 0][:, electrode])  
stim = (raw["DREAMER"][0, 0]["Data"][0, participant]["EEG"][0, 0]["stimuli"][0, 0][0, 0][:, ]) 
print(basl.shape)
print(stim.shape)

(7808,)
(25472, 14)

def feat_extract_EEG(raw):
    
    EEG_tmp = np.zeros((23, 18, 107520))
    for participant in range(0, 23):
        for video in range(0, 18):
          column=0
          for electrode in range(0,14):
            stim = (raw["DREAMER"][0, 0]["Data"][0, participant]["EEG"][0, 0]["stimuli"][0, 0][video, 0][:, electrode])
            for i in range(0,7680):
              EEG_tmp[participant,video,column]= stim[i]
              column+=1                            
    col = []
    for i in range(0, 7680):
      col.append("Electrode1_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode2_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode3_"+str(i + 1))
    for i in range(0, 7680):
     col.append("Electrode4_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode5_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode6_"+str(i + 1))
    for i in range(0, 7680):
        col.append("Electrode7_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode8_"+str(i + 1))
    for i in range(0, 7680):
     col.append("Electrode9_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode10_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode11_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode12_"+str(i + 1))
    for i in range(0, 7680):
      col.append("Electrode13_"+str(i + 1))
    for i in range(0, 7680):
     col.append("Electrode14_"+str(i + 1))

    EEG_tmp = EEG_tmp.reshape(-1,EEG_tmp.shape[2])
    data_EEG = pd.DataFrame(EEG_tmp, columns=col)
    # print(data_EEG)
    # from sklearn import preprocessing

    # x = data_EEG.values
    # min_max_scaler = preprocessing.MinMaxScaler()
    # x_scaled = min_max_scaler.fit_transform(x)
    # data_EEG = pd.DataFrame(x_scaled)
    # data_EEG.columns = col
    # print(data_EEG)
    return data_EEG

def participant_affective(raw):
    a = np.zeros((23, 18, 4), dtype=object)
    for participant in range(0, 23):
        for video in range(0, 18):
            # a[participant, video, 0] = (raw["DREAMER"][0, 0]["Data"]
            #                             [0, participant]["Age"][0][0][0])
            
            # a[participant, video, 1] = (raw["DREAMER"][0, 0]["Data"]
            #                             [0, participant]["Gender"][0][0][0])
            
            # a[participant, video, 2] = int(participant+1)
            
            # a[participant, video, 3] = int(video+1)
            
            # a[participant, video, 4] = ["Searching for Bobby Fischer",
            #                             "D.O.A.", "The Hangover", "The Ring",
            #                             "300", "National Lampoon\'s VanWilder",
            #                             "Wall-E", "Crash", "My Girl",
            #                             "The Fly", "Pride and Prejudice",
            #                             "Modern Times", "Remember the Titans",
            #                             "Gentlemans Agreement", "Psycho",
            #                             "The Bourne Identitiy",
            #                             "The Shawshank Redemption",
            #                             "The Departed"][video]
            
            a[participant, video, 0] = ["calmness", "surprise", "amusement",
                                        "fear", "excitement", "disgust",
                                        "happiness", "anger", "sadness",
                                        "disgust", "calmness", "amusement",
                                        "happiness", "anger", "fear",
                                        "excitement", "sadness",
                                        "surprise"][video]
            a[participant, video, 1] = int(raw["DREAMER"][0, 0]["Data"]
                                           [0, participant]["ScoreValence"]
                                           [0, 0][video, 0])
            a[participant, video, 2] = int(raw["DREAMER"][0, 0]["Data"]
                                           [0, participant]["ScoreArousal"]
                                           [0, 0][video, 0])
            a[participant, video, 3] = int(raw["DREAMER"][0, 0]["Data"]
                                           [0, participant]["ScoreDominance"]
                                           [0, 0][video, 0])
    b = pd.DataFrame(a.reshape((23*18, a.shape[2])),
        columns=[ "target_emotion", "valence", "arousal", "dominance"])
    return b
def_EEG3 = feat_extract_EEG(raw)

df_participant_affective = participant_affective(raw)
df_participant_affective

df = pd.concat([def_EEG3, df_participant_affective], axis=1)
df

num_rows = 18
total_iterations = 23

# 保存文件的初始编号
file_index = 1
b= []
for i in range(1):
    # 计算起始行和结束行的索引
    start_row = i * num_rows
    end_row = (i + 1) * num_rows
    a = []
    for j in range(1,15):
        selected_columns = [col for col in df.columns if col.startswith('Electrode'+str(j)+'_')]
        result_df = df[selected_columns].iloc[start_row:end_row]
        a.append(result_df)
    b.append(a)
    b = np.array(b)
    # 保存结果到 Excel 文件
    excel_filename = f'C:/Users/86157/Desktop/dreamer/output_electrode1_{start_row + 1}_to_{end_row}.npy'
    np.save(excel_filename,b)

    print(f'提取的数据已保存到 {excel_filename}')

    # 更新文件编号
    file_index += 1

提取的数据已保存到 C:/Users/86157/Desktop/dreamer/output_electrode1_1_to_18.npy

num_rows = 18
total_iterations = 23



for i in range(total_iterations):
    b= []
    start_row = i * num_rows
    end_row = (i + 1) * num_rows
    a = []
    for j in range(1,15):
        selected_columns = [col for col in df.columns if col.startswith('Electrode'+str(j)+'_')]
        result_df = df[selected_columns].iloc[start_row:end_row]
        a.append(result_df)
    b.append(a)
    b = np.array(b)
    c = b.reshape(14,18,7680)
    c= np.transpose(c,(1,0,2))
    # 保存结果到 Excel 文件
    excel_filename = f'C:/Users/86157/Desktop/dreamer/dreamer{str(i)}.npy'
    np.save(excel_filename,c)

    print(f'提取的数据已保存到 {excel_filename}')