144 lines
6.1 KiB
Python
144 lines
6.1 KiB
Python
import os
|
||
import pyedflib
|
||
import numpy as np
|
||
import pandas as pd
|
||
import shutil
|
||
from scipy.signal import resample
|
||
|
||
def writefile(filename,data,style='w'):
|
||
"""
|
||
函数说明：
|
||
在指定地点创建一个文件，写入data数据
|
||
:param filename: 需写入的文件地址和名称
|
||
:param data: 需写入的数据
|
||
:return: 无返回
|
||
"""
|
||
data = data.astype('str') #将data数组的数据转换成字符串
|
||
with open(filename,style) as f :
|
||
for str in data :
|
||
f.write(str+'\n')
|
||
return 0
|
||
|
||
def upsample_resample(data,data_fs,target_fs):
|
||
up_ratio = target_fs / data_fs
|
||
target_num = int(len(data) * up_ratio)
|
||
|
||
# 执行重采样
|
||
if up_ratio > 1:
|
||
data_up = resample(data, target_num)
|
||
target_fs = int(target_fs)
|
||
else:
|
||
print("上采样倍数不大于1，不进行上采样操作")
|
||
data_up = data
|
||
target_fs = int(data_fs)
|
||
|
||
return data_up,target_fs
|
||
|
||
def event_handle(df):
|
||
df['类型'] = df['类型'].replace('腿部运动', 'Leg Movement')
|
||
df['类型'] = df['类型'].replace('低通气', 'Hypopnea')
|
||
df['类型'] = df['类型'].replace('混合型呼吸暂停', 'Mixed apnea')
|
||
df['类型'] = df['类型'].replace('中枢性呼吸暂停', 'Central apnea')
|
||
df['类型'] = df['类型'].replace('阻塞性呼吸暂停', 'Obstructive apnea')
|
||
df['类型'] = df['类型'].replace('相对氧减', 'Desaturation')
|
||
df['类型'] = df['类型'].replace('长 RR', 'Long RR')
|
||
df['类型'] = df['类型'].replace('心率上升', 'Heart Rate Rise')
|
||
df['类型'] = df['类型'].replace('心率下降', 'Heart Rate Drop')
|
||
df['类型'] = df['类型'].replace('PTT 下降', 'PTT drop')
|
||
df['类型'] = df['类型'].replace('打鼾', 'Snore')
|
||
df['类型'] = df['类型'].replace('觉醒', 'u-arousal')
|
||
df['睡眠期'] = df['睡眠期'].replace('WK', 'W')
|
||
df['睡眠期'] = df['睡眠期'].replace('REM', 'R')
|
||
|
||
column_mapping = {
|
||
"类型": "Event type",
|
||
"睡眠期": "Stage",
|
||
"时间": "Time",
|
||
"时期": "Epoch",
|
||
"持续时间": "Duration",
|
||
"体位": "Body Position",
|
||
"验证": "Validation"
|
||
} # 列名映射
|
||
df = df.rename(columns=column_mapping) # 重命名列
|
||
target_columns = [
|
||
"Event type", "Stage", "Time", "Epoch", "Date", "Duration",
|
||
"HR bef.", "HR extr.", "HR delta", "O2 bef.", "O2 min.",
|
||
"O2 delta", "Body Position", "Validation", "Unnamed: 14"
|
||
] # 添加目标文件中要求的其他列，并赋值为None
|
||
for col in target_columns: # 添加缺失的列并用None填充
|
||
if col not in df.columns:
|
||
df[col] = None
|
||
df = df[target_columns] # 按目标列顺序排序
|
||
|
||
return df
|
||
|
||
def read_edf(edfpath,stagepath,eventpath,outpath):
|
||
list = np.array([f for f in os.listdir(edfpath) if f.endswith('.edf')])
|
||
for sample_num in list:
|
||
print("样本编号：", sample_num)
|
||
|
||
output_sample = outpath + sample_num.split('.')[0] + '/'
|
||
if not os.path.exists(output_sample):
|
||
os.makedirs(output_sample)
|
||
|
||
##读取edf文件数据
|
||
fpath = edfpath + sample_num
|
||
shutil.copy(fpath, output_sample)
|
||
with pyedflib.EdfReader(fpath) as file:
|
||
##读取导出的睡眠分期标签csv文件
|
||
df = pd.read_csv(stagepath + sample_num.split('.')[0] + '_Stage.csv')
|
||
stage_values = df.values[:,2].reshape(-1)
|
||
|
||
##读edf文件信息
|
||
start_time = file.getStartdatetime()
|
||
sec = file.getFileDuration()
|
||
sec_start = start_time.strftime("%H") + ":" + start_time.strftime("%M") + ":" + start_time.strftime("%S")
|
||
assert (df.values[0,1] == sec_start)
|
||
print("edf文件与睡眠分期标签开始时间一致！！！！！！！")
|
||
|
||
##读取Event列表
|
||
df1 = pd.read_csv(eventpath + sample_num.split('.')[0] + '_Event.csv', header=0, encoding='utf-8')
|
||
df1 = event_handle(df1)
|
||
df1.to_csv(output_sample + 'SA Label_Raw.csv', index=False, encoding='GBK')
|
||
|
||
signal_num = file.signals_in_file
|
||
signal_label = file.getSignalLabels()
|
||
min_length = int(np.min([sec,df.shape[0]*30]))
|
||
print("裁剪为最小长度:",min_length)
|
||
flag = 0
|
||
for i, index in enumerate(signal_label):
|
||
signal = file.readSignal(i)
|
||
sample_frequency = int(file.getSampleFrequency(i))
|
||
data_org = signal[0:min_length * sample_frequency]
|
||
|
||
if (index== 'Flow Patient') :
|
||
if flag == 0:
|
||
writefile(output_sample + 'Flow T_Raw_' + str(sample_frequency) + '.txt', data_org)
|
||
flag = 1
|
||
else:
|
||
writefile(output_sample + 'Flow P_Raw_' + str(sample_frequency) + '.txt', data_org)
|
||
else:
|
||
writefile(output_sample + index + '_Raw_' + str(sample_frequency) + '.txt', data_org)
|
||
with open(os.path.join(output_sample, "StartTime_Raw.txt"), "a") as text_file:
|
||
text_file.write(start_time.strftime("%Y-%m-%d %H:%M:%S\n"))
|
||
text_file.write("{:g}\n".format(int(sec)))
|
||
|
||
##裁剪对齐睡眠分期标签
|
||
stage_values = np.where(stage_values == 'WK', 'W', stage_values)
|
||
stage_values = np.where(stage_values == 'REM', 'R', stage_values)
|
||
stage_values = np.where(stage_values == 'N1', 'N1', stage_values)
|
||
stage_values = np.where(stage_values == 'N2', 'N2', stage_values)
|
||
stage = np.where(stage_values == 'N3', 'N3', stage_values)
|
||
stage = stage.repeat(30)
|
||
writefile(output_sample + '5_class_Raw_1.txt',stage[0:min_length])
|
||
|
||
if __name__ == '__main__':
|
||
edf_dir = 'E:\DB\\testdata\ZD5Y\PSG_Origin/Edf/'
|
||
stage_dir = 'E:\DB\\testdata\ZD5Y\PSG_Origin/Stage/'
|
||
event_dir = 'E:\DB\\testdata\ZD5Y\PSG_Origin/Event/'
|
||
output_dir = 'E:\DB\\testdata\ZD5Y\PSG_Text/'
|
||
|
||
read_edf(edf_dir,stage_dir,event_dir,output_dir)
|
||
|
||
print("finished")
|