/
nedf.py
219 lines (183 loc) · 7.46 KB
/
nedf.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
# -*- coding: utf-8 -*-
"""Import NeuroElectrics DataFormat (NEDF) files."""
from copy import deepcopy
from datetime import datetime, timezone
from xml.etree import ElementTree
import numpy as np
from ..base import BaseRaw
from ..meas_info import create_info
from ..utils import _mult_cal_one
from ...utils import warn, verbose, _check_fname
def _getsubnodetext(node, name):
"""Get an element from an XML node, raise an error otherwise.
Parameters
----------
node: Element
XML Element
name: str
Child element name
Returns
-------
test: str
Text contents of the child nodes
"""
subnode = node.findtext(name)
if not subnode:
raise RuntimeError('NEDF header ' + name + ' not found')
return subnode
def _parse_nedf_header(header):
"""Read header information from the first 10kB of an .nedf file.
Parameters
----------
header : bytes
Null-terminated header data, mostly the file's first 10240 bytes.
Returns
-------
info : dict
A dictionary with header information.
dt : numpy.dtype
Structure of the binary EEG/accelerometer/trigger data in the file.
n_samples : int
The number of data samples.
"""
info = {}
# nedf files have three accelerometer channels sampled at 100Hz followed
# by five EEG samples + TTL trigger sampled at 500Hz
# For 32 EEG channels and no stim channels, the data layout may look like
# [ ('acc', '>u2', (3,)),
# ('data', dtype([
# ('eeg', 'u1', (32, 3)),
# ('trig', '>i4', (1,))
# ]), (5,))
# ]
dt = [] # dtype for the binary data block
datadt = [] # dtype for a single EEG sample
headerend = header.find(b'\0')
if headerend == -1:
raise RuntimeError('End of header null not found')
headerxml = ElementTree.fromstring(header[:headerend])
nedfversion = headerxml.findtext('NEDFversion', '')
if nedfversion not in ['1.3', '1.4']:
warn('NEDFversion unsupported, use with caution')
if headerxml.findtext('stepDetails/DeviceClass', '') == 'STARSTIM':
warn('Found Starstim, this hasn\'t been tested extensively!')
if headerxml.findtext('AdditionalChannelStatus', 'OFF') != 'OFF':
raise RuntimeError('Unknown additional channel, aborting.')
n_acc = int(headerxml.findtext('NumberOfChannelsOfAccelerometer', 0))
if n_acc:
# expect one sample of u16 accelerometer data per block
dt.append(('acc', '>u2', (n_acc,)))
eegset = headerxml.find('EEGSettings')
if eegset is None:
raise RuntimeError('No EEG channels found')
nchantotal = int(_getsubnodetext(eegset, 'TotalNumberOfChannels'))
info['nchan'] = nchantotal
info['sfreq'] = int(_getsubnodetext(eegset, 'EEGSamplingRate'))
info['ch_names'] = [e.text for e in eegset.find('EEGMontage')]
if nchantotal != len(info['ch_names']):
raise RuntimeError(
f"TotalNumberOfChannels ({nchantotal}) != "
f"channel count ({len(info['ch_names'])})")
# expect nchantotal uint24s
datadt.append(('eeg', 'B', (nchantotal, 3)))
if headerxml.find('STIMSettings') is not None:
# 2* -> two stim samples per eeg sample
datadt.append(('stim', 'B', (2, nchantotal, 3)))
warn('stim channels are currently ignored')
# Trigger data: 4 bytes in newer versions, 1 byte in older versions
trigger_type = '>i4' if headerxml.findtext('NEDFversion') else 'B'
datadt.append(('trig', trigger_type))
# 5 data samples per block
dt.append(('data', np.dtype(datadt), (5,)))
date = headerxml.findtext('StepDetails/StartDate_firstEEGTimestamp', 0)
info['meas_date'] = datetime.fromtimestamp(int(date) / 1000, timezone.utc)
n_samples = int(_getsubnodetext(eegset, 'NumberOfRecordsOfEEG'))
n_full, n_last = divmod(n_samples, 5)
dt_last = deepcopy(dt)
assert dt_last[-1][-1] == (5,)
dt_last[-1] = list(dt_last[-1])
dt_last[-1][-1] = (n_last,)
dt_last[-1] = tuple(dt_last[-1])
return info, np.dtype(dt), np.dtype(dt_last), n_samples, n_full
# the first 10240 bytes are header in XML format, padded with NULL bytes
_HDRLEN = 10240
class RawNedf(BaseRaw):
"""Raw object from NeuroElectrics nedf file."""
def __init__(self, filename, preload=False, verbose=None):
filename = _check_fname(filename, 'read', True, 'filename')
with open(filename, mode='rb') as fid:
header = fid.read(_HDRLEN)
header, dt, dt_last, n_samp, n_full = _parse_nedf_header(header)
ch_names = header['ch_names'] + ['STI 014']
ch_types = ['eeg'] * len(ch_names)
ch_types[-1] = 'stim'
info = create_info(ch_names, header['sfreq'], ch_types)
# scaling factor ADC-values -> volts
# taken from the NEDF EEGLAB plugin
# (https://www.neuroelectrics.com/resources/software/):
for ch in info['chs'][:-1]:
ch['cal'] = 2.4 / (6.0 * 8388607)
with info._unlock():
info['meas_date'] = header['meas_date']
raw_extra = dict(dt=dt, dt_last=dt_last, n_full=n_full)
super().__init__(
info, preload=preload, filenames=[filename], verbose=verbose,
raw_extras=[raw_extra], last_samps=[n_samp - 1])
def _read_segment_file(self, data, idx, fi, start, stop, cals, mult):
dt = self._raw_extras[fi]['dt']
dt_last = self._raw_extras[fi]['dt_last']
n_full = self._raw_extras[fi]['n_full']
n_eeg = dt[1].subdtype[0][0].shape[0]
# data is stored in 5-sample chunks (except maybe the last one!)
# so we have to do some gymnastics to pick the correct parts to
# read
offset = start // 5 * dt.itemsize + _HDRLEN
start_sl = start % 5
n_samples = stop - start
n_samples_full = min(stop, n_full * 5) - start
last = None
n_chunks = (n_samples_full - 1) // 5 + 1
n_tot = n_chunks * 5
with open(self._filenames[fi], 'rb') as fid:
fid.seek(offset, 0)
chunks = np.fromfile(fid, dtype=dt, count=n_chunks)
assert len(chunks) == n_chunks
if n_samples != n_samples_full:
last = np.fromfile(fid, dtype=dt_last, count=1)
eeg = _convert_eeg(chunks, n_eeg, n_tot)
trig = chunks['data']['trig'].reshape(1, n_tot)
if last is not None:
n_last = dt_last['data'].shape[0]
eeg = np.concatenate(
(eeg, _convert_eeg(last, n_eeg, n_last)), axis=-1)
trig = np.concatenate(
(trig, last['data']['trig'].reshape(1, n_last)), axis=-1)
one_ = np.concatenate((eeg, trig))
one = one_[:, start_sl:n_samples + start_sl]
_mult_cal_one(data, one, idx, cals, mult)
def _convert_eeg(chunks, n_eeg, n_tot):
# convert uint8-triplet -> int32
eeg = chunks['data']['eeg'] @ np.array([1 << 16, 1 << 8, 1])
# convert sign if necessary
eeg[eeg > (1 << 23)] -= 1 << 24
eeg = eeg.reshape((n_tot, n_eeg)).T
return eeg
@verbose
def read_raw_nedf(filename, preload=False, verbose=None):
"""Read NeuroElectrics .nedf files.
NEDF file versions starting from 1.3 are supported.
Parameters
----------
filename : str
Path to the .nedf file.
%(preload)s
%(verbose)s
Returns
-------
raw : instance of RawNedf
A Raw object containing NEDF data.
See Also
--------
mne.io.Raw : Documentation of attribute and methods.
"""
return RawNedf(filename, preload, verbose)