def importBDF(filename, useCpp=False):
"""returns dataMatrix, eventTable, chanLabels"""
if len(filename) > 0:
rec1 = pybdf.bdfRecording(filename)
data1 = rec1.getData(trigChan=True, useCpp=useCpp)
#retrieve sampling rates (list of sampling rate of each channel)
sampRate1 = rec1.sampRate
print("**********************")
print("The sampling rate of, ", filename, "is", sampRate1[0], "Hz")
print()
dur1 = rec1.duration
#retrieve recording durations (list of recording durations of each channel)
print("**********************")
print("The duration of, ", filename, "is", dur1, "seconds")
print()
dataMatrix = data1['data']
#time_array1 = np.arange(dataMatrix1.shape[1]) / rec1.sampRate[0]
chanLabels = rec1.chanLabels
eventTable = data1['eventTable']
return dataMatrix, eventTable, chanLabels
else:
print("--Please enter a file--")
def readSigfile(self, filename):
""" Load ECG signal.
Return a 2-dim signal (t, bmp(t))
"""
bdfRec = pybdf.bdfRecording(filename)
rec = bdfRec.getData(channels=[33])
self.SIG_SampleRate = bdfRec.sampRate[33]
data = np.array(rec['data'][0])[rec['eventTable']['idx'][2]:]
return ECGsignal(data, self.SIG_SampleRate)
def load_data_bdf(data_bdf_filename, decimation_factor, num_cores=1):
logging.debug('%s data_bdf_filename: %s', TAG, data_bdf_filename)
recording_obj = pybdf.bdfRecording(data_bdf_filename)
# Get recording information
logging.debug('sampling_rate [Hz]: %f', recording_obj.sampRate[0])
logging.debug('duration [s]: %f', recording_obj.duration)
logging.debug('#channels: %d', recording_obj.nChannels)
logging.debug('unit of measure: %s', recording_obj.physDim[0])
print 'channel labels:', recording_obj.chanLabels
print 'dataChanLabels:', recording_obj.dataChanLabels
# Convert the data channel labels to integers, otherwise pybdf runs into a bug
for i_label in range(len(recording_obj.dataChanLabels)):
recording_obj.dataChanLabels[i_label] = i_label
print 'dataChanLabels:', recording_obj.dataChanLabels
# Get the data object
data_obj = recording_obj.getData()
# Get the signal data from the data object
logging.debug('Getting the signal data from the data object...')
X_raw = data_obj['data'].T
logging.debug('Getting the signal data from the data object finished.')
logging.debug('X_raw shape: %d, %d', X_raw.shape[0], X_raw.shape[1])
#time_axis = np.arange(X_raw.shape[1]) / recording_obj.sampRate[0]
# Get the event information from the data object
# Event codes: 247 is button pressed (1), 255 is button released (0)
logging.debug('Getting the event data from the data object...')
event_table = data_obj['eventTable']
print 'event_table[\'code\']:\n', event_table['code']
print 'event_table[\'idx\']:\n', event_table['idx']
print 'event_table[\'dur\']:\n', event_table['dur']
logging.debug('Getting the event data from the data object finished.')
# Build the label feed from the event information
logging.debug('Building the label feed from the event information...')
labels = np.zeros((X_raw.shape[0], 1), np.float32)
n_events = len(event_table['code'])
EVENT_CODE_ACTION = 247
#EVENT_CODE_IDLE = 255
for i_event in range(1, n_events):
if event_table['code'][i_event] == EVENT_CODE_ACTION:
labels[(event_table['idx'][i_event] -
params.EVENT_LENGTH_SAMPLES / 2):(
event_table['idx'][i_event] +
params.EVENT_LENGTH_SAMPLES / 2), :] = 1.0
#if event_table['code'][i_event] == EVENT_CODE_ACTION:
# labels[event_table['idx'][i_event]:event_table['idx'][i_event+1], :] = 1.0
logging.debug('np.sum(labels): %f', np.sum(labels))
logging.debug(
'Building the label feed from the event information finished.')
# Low-pass-filter the data before downsampling
freq_sampling = recording_obj.sampRate[0]
logging.debug('freq_sampling: %f', freq_sampling)
freq_cut_downsampling = freq_sampling / (2.0 * decimation_factor)
logging.debug('freq_cut_downsampling: %f', freq_cut_downsampling)
M_fir = freq_sampling
tdfilt_numer, tdfilt_denom = create_lowpass_filter(
freq_cut=freq_cut_downsampling,
freq_s=freq_sampling,
M_fir=M_fir,
plot=True)
logging.debug('Low-pass filtering the data before downsampling...')
logging.debug(
'Timestamp: %s',
datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
if num_cores > 1:
#X_lpfiltered = np.array(joblib.Parallel(n_jobs=params.NUM_PARALLEL_JOBS)
# (joblib.delayed(scipy.signal.lfilter)
# (tdfilt_numer, tdfilt_denom,
# X_raw[:, i_ch]) for i_ch in range(X_raw.shape[1]))).T
X_lpfiltered = np.array(
joblib.Parallel(n_jobs=params.NUM_PARALLEL_JOBS)(
joblib.delayed(scipy.signal.filtfilt)(
tdfilt_numer, tdfilt_denom, X_raw[:, i_ch])
for i_ch in range(X_raw.shape[1]))).T
else:
#X_lpfiltered = scipy.signal.lfilter(tdfilt_numer, tdfilt_denom, X_raw.T).T
X_lpfiltered = scipy.signal.filtfilt(tdfilt_numer, tdfilt_denom,
X_raw.T).T
logging.debug(
'Timestamp: %s',
datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
logging.debug('Low-pass filtering the data before downsampling finished.')
logging.debug('X_lpfiltered shape: %d, %d', X_lpfiltered.shape[0],
X_lpfiltered.shape[1])
# Plot the data
if False:
time_to = 2048 * 5
time_axis = np.arange(time_to)
channels_to_plot = (54, 111)
plt.plot(time_axis, X_raw[0:time_to, channels_to_plot], label='raw')
plt.plot(time_axis,
X_lpfiltered[0:time_to, channels_to_plot],
label='filt')
#plt.plot(time_axis, labels[0:time_to], label='event')
plt.legend(loc='lower right')
plt.show()
# Downsample the data
downsample_indices = np.arange(
0, X_lpfiltered.shape[0],
int(decimation_factor)) + (int(decimation_factor) - 1)
print 'downsample_indices:\n', downsample_indices
logging.debug('Downsampling the data...')
logging.debug(
'Timestamp: %s',
datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
X_downsampled = X_lpfiltered[downsample_indices, :]
#X_downsampled = X_lpfiltered[downsample_indices, 0:params.NUM_CHANNELS_BDF]
labels = labels[downsample_indices, :]
#X_raw = X_raw[::decimation_factor]
#labels = labels[::decimation_factor]
logging.debug(
'Timestamp: %s',
datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
logging.debug('Downsampling the data finished.')
logging.debug('X_downsampled shape: %d, %d', X_downsampled.shape[0],
X_downsampled.shape[1])
return X_downsampled, labels
def load_data_bdf(data_bdf_filename, decimation_factor, num_cores=1):
logging.debug('%s data_bdf_filename: %s', TAG, data_bdf_filename)
recording_obj = pybdf.bdfRecording(data_bdf_filename)
# Get recording information
logging.debug('sampling_rate [Hz]: %f', recording_obj.sampRate[0])
logging.debug('duration [s]: %f', recording_obj.duration)
logging.debug('#channels: %d', recording_obj.nChannels)
logging.debug('unit of measure: %s', recording_obj.physDim[0])
print 'channel labels:', recording_obj.chanLabels
print 'dataChanLabels:', recording_obj.dataChanLabels
# Convert the data channel labels to integers, otherwise pybdf runs into a bug
for i_label in range(len(recording_obj.dataChanLabels)):
recording_obj.dataChanLabels[i_label] = i_label
print 'dataChanLabels:', recording_obj.dataChanLabels
# Get the data object
data_obj = recording_obj.getData()
# Get the signal data from the data object
logging.debug('Getting the signal data from the data object...')
X_raw = data_obj['data'].T
logging.debug('Getting the signal data from the data object finished.')
logging.debug('X_raw shape: %d, %d', X_raw.shape[0], X_raw.shape[1])
#time_axis = np.arange(X_raw.shape[1]) / recording_obj.sampRate[0]
# Get the event information from the data object
# Event codes: 247 is button pressed (1), 255 is button released (0)
logging.debug('Getting the event data from the data object...')
event_table = data_obj['eventTable']
print 'event_table[\'code\']:\n', event_table['code']
print 'event_table[\'idx\']:\n', event_table['idx']
print 'event_table[\'dur\']:\n', event_table['dur']
logging.debug('Getting the event data from the data object finished.')
# Build the label feed from the event information
logging.debug('Building the label feed from the event information...')
labels = np.zeros((X_raw.shape[0], 1), np.float32)
n_events = len(event_table['code'])
EVENT_CODE_ACTION = 247
#EVENT_CODE_IDLE = 255
for i_event in range(1, n_events):
if event_table['code'][i_event] == EVENT_CODE_ACTION:
labels[(event_table['idx'][i_event] - params.EVENT_LENGTH_SAMPLES/2)
:(event_table['idx'][i_event] + params.EVENT_LENGTH_SAMPLES/2), :] = 1.0
#if event_table['code'][i_event] == EVENT_CODE_ACTION:
# labels[event_table['idx'][i_event]:event_table['idx'][i_event+1], :] = 1.0
logging.debug('np.sum(labels): %f', np.sum(labels))
logging.debug('Building the label feed from the event information finished.')
# Low-pass-filter the data before downsampling
freq_sampling = recording_obj.sampRate[0]
logging.debug('freq_sampling: %f', freq_sampling)
freq_cut_downsampling = freq_sampling / (2.0 * decimation_factor)
logging.debug('freq_cut_downsampling: %f', freq_cut_downsampling)
M_fir = freq_sampling
tdfilt_numer, tdfilt_denom = create_lowpass_filter(
freq_cut=freq_cut_downsampling,
freq_s=freq_sampling,
M_fir=M_fir,
plot=True)
logging.debug('Low-pass filtering the data before downsampling...')
logging.debug('Timestamp: %s', datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
if num_cores > 1:
#X_lpfiltered = np.array(joblib.Parallel(n_jobs=params.NUM_PARALLEL_JOBS)
# (joblib.delayed(scipy.signal.lfilter)
# (tdfilt_numer, tdfilt_denom,
# X_raw[:, i_ch]) for i_ch in range(X_raw.shape[1]))).T
X_lpfiltered = np.array(joblib.Parallel(n_jobs=params.NUM_PARALLEL_JOBS)
(joblib.delayed(scipy.signal.filtfilt)
(tdfilt_numer, tdfilt_denom,
X_raw[:, i_ch]) for i_ch in range(X_raw.shape[1]))).T
else:
#X_lpfiltered = scipy.signal.lfilter(tdfilt_numer, tdfilt_denom, X_raw.T).T
X_lpfiltered = scipy.signal.filtfilt(tdfilt_numer, tdfilt_denom, X_raw.T).T
logging.debug('Timestamp: %s', datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
logging.debug('Low-pass filtering the data before downsampling finished.')
logging.debug('X_lpfiltered shape: %d, %d', X_lpfiltered.shape[0], X_lpfiltered.shape[1])
# Plot the data
if False:
time_to = 2048*5
time_axis = np.arange(time_to)
channels_to_plot = (54, 111)
plt.plot(time_axis, X_raw[0:time_to, channels_to_plot], label='raw')
plt.plot(time_axis, X_lpfiltered[0:time_to, channels_to_plot], label='filt')
#plt.plot(time_axis, labels[0:time_to], label='event')
plt.legend(loc='lower right')
plt.show()
# Downsample the data
downsample_indices = np.arange(0, X_lpfiltered.shape[0], int(decimation_factor)) + (int(decimation_factor)-1)
print 'downsample_indices:\n', downsample_indices
logging.debug('Downsampling the data...')
logging.debug('Timestamp: %s', datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
X_downsampled = X_lpfiltered[downsample_indices, :]
#X_downsampled = X_lpfiltered[downsample_indices, 0:params.NUM_CHANNELS_BDF]
labels = labels[downsample_indices, :]
#X_raw = X_raw[::decimation_factor]
#labels = labels[::decimation_factor]
logging.debug('Timestamp: %s', datetime.datetime.now().strftime(params.TIMESTAMP_FORMAT_STR))
logging.debug('Downsampling the data finished.')
logging.debug('X_downsampled shape: %d, %d', X_downsampled.shape[0], X_downsampled.shape[1])
return X_downsampled, labels
import matplotlib.pyplot as plt
matplotlib_available = True
except:
matplotlib_available = False
#on linux the following automatically
#downloads example datasets from BIOSEMI
#if you're on windows or mac, download manually
# http://www.biosemi.com/download/BDFtestfiles.zip
#subprocess.call("wget http://www.biosemi.com/download/BDFtestfiles.zip", shell=True)
#subprocess.call("unzip BDFtestfiles.zip", shell=True)
fName1 = "Newtest17-256.bdf"
fName2 = "Newtest17-2048.bdf"
#let's see how long it takes
rec1 = pybdf.bdfRecording(fName1)
data1 = rec1.getData()
rec2 = pybdf.bdfRecording(fName2)
data2 = rec1.getData()
#retrieve sampling rates (list of sampling rate of each channel)
sampRate1 = rec1.sampRate
sampRate2 = rec2.sampRate
print("**********************")
print("The sampling rate of, ", fName1, "is", sampRate1[0], "Hz")
print("The sampling rate of, ", fName2, "is", sampRate2[0], "Hz")
print("--------------\n")
dur1 = rec1.duration
dur2 = rec2.duration
else:
print("--Please enter a file--")
files = sorted(glob('./*2048*'))
if len(files) > 1:
os.remove(files.pop(1))
files.append(files[0][0:-4] + '_out.bdf')
print(files)
copyfile(files[0], files[1])
source_file = pybdf.bdfRecording(files[0])
data = source_file.getData()
cpp_event = data['eventTable']
print(cpp_event)
rec_out = pybdf.bdfRecording(files[1])
out_cpp_codes = np.arange(1, cpp_event['code'].shape[0] + 1, 1, dtype=np.uint8)
out_cpp_times = (cpp_event['idx'] + 1)
# print(len(out_cpp_codes), len(out_cpp_times))
# print(out_cpp_codes, '\n', out_cpp_times)
from pybdf import bdfRecording
import pandas as pd
import numpy as np
from matplotlib import pyplot
from biosppy import ecg
import matplotlib as ml
import matplotlib.pyplot as plt
#bdfRec = bdfRecording("../Dataset/Sessions/789/Part_7_N_Trial5_emotion.bdf")
bdfRec = bdfRecording("789/Part_7_N_Trial5_emotion.bdf")
ECG_idx = bdfRec.dataChanLabels.index('EXG2')
rec = bdfRec.getData(channels=[ECG_idx])
sampling_rate = bdfRec.sampRate[ECG_idx]
signal = np.array(rec['data'][0])
rec_s = pd.Series(rec['data'][0])
rec_s.plot()
pyplot.pause(0.1)
hr = ecg.ecg(signal=signal, sampling_rate=sampling_rate, show=False)[6]
#Must figure out a better way to find the time table.
#Video is 22 seconds and there is 85 samples
#video_duration = 22
#step = video_duration/float(len(hr))
step = 1/(float(sampling_rate)/60)
length = len(hr)
video_duration = step * float(len(hr))
time_array = np.arange(0, video_duration, step)
import pybdf
import numpy as np
# def readfile():
# pass
# def main(person):
# for i in range(1,3):
# path = 'Part'+person+'_IAPS_SES'+'_EEG_fNIRS'
# readfile()
# if __name__ == '__main__':
# #选择实验者序号
# person = '1';
# main(person)
# 读取EEG数据
rec = pybdf.bdfRecording('sample.bdf')
data = rec.getData()
sampRate = rec.sampRate
durtime = rec.duration
chanlabels = rec.chanLabels
labels = rec.getData(channels=[-1])
print('EEG通道个数为：%d' % len(sampRate))
print('EEG信号采样频率是%d' % sampRate[0])
print('经过时间是：%.2f秒，%.2f分钟' % (durtime, durtime / 60.0))
print('EEG信号采样频率是%d' % sampRate[0])
print('EEG采样通道:', chanlabels)
print('其中脑电信号通道为%d个通道，眼电信号为%d个通道' % ((len(chanlabels) - 9), 8))
print('EEG数据结构：')
for key in labels.keys():
print(key)
print('EEG-Data：', labels['data'].shape)
print('EEG-TrigChan：', labels['trigChan'])
matplotlib_available = True
except:
matplotlib_available = False
#on linux the following automatically
#downloads example datasets from BIOSEMI
#if you're on windows or mac, download manually
# http://www.biosemi.com/download/BDFtestfiles.zip
#subprocess.call("wget http://www.biosemi.com/download/BDFtestfiles.zip", shell=True)
#subprocess.call("unzip BDFtestfiles.zip", shell=True)
fName1 = "Newtest17-256.bdf"
fName2 = "Newtest17-2048.bdf"
#let's see how long it takes
rec1 = pybdf.bdfRecording(fName1)
data1 = rec1.getData()
rec2 = pybdf.bdfRecording(fName2)
data2 = rec1.getData()
#retrieve sampling rates (list of sampling rate of each channel)
sampRate1 = rec1.sampRate
sampRate2 = rec2.sampRate
print("**********************")
print("The sampling rate of, ", fName1, "is", sampRate1[0], "Hz")
print("The sampling rate of, ", fName2, "is", sampRate2[0], "Hz")
print("--------------\n")
dur1 = rec1.duration
dur2 = rec2.duration
