def plvAnalysis(epochs, Type):
# phase locking analysis
params = dict(Fs=16384, tapers=[1, 1], fpass=[400, 600], itc=0)
# plv to TFS
epochs = np.transpose(epochs,
(1, 0, 2)) # switching the first and second columns
plv, f = mtplv(epochs, params)
index = np.where(plv[topchans[0], ] == max(plv[topchans[0], ]))
plv_max = max(plv[topchans[0], ])
f_max = f[index]
dimen = epochs.shape
numtrials = dimen[1]
return plv_max, f_max, numtrials
def plvAnalysis(epochs, Type):
# phase locking analysis
params = dict(Fs=16384, tapers=[1, 1], fpass=[400, 600], itc=0)
# plv to TFS
epochs = np.transpose(epochs,
(1, 0, 2)) # switching the first and second columns
plv, f = mtplv(epochs, params)
index = np.where(plv[topchans[0], ] == max(plv[topchans[0], ]))
plv_max = max(plv[topchans[0], ])
f_max = f[index]
dimen = epochs.shape
numtrials = dimen[1]
# saving plvs
np.savez(froot + '/plvs/' + subj + Type,
plv=plv,
plv_max=plv_max,
f=f_max,
numtrials=numtrials)
# collecting individual plvs into array for all subjects
return plv_max, f_max, numtrials
def plvAnalysis(epochs, Type):
# phase locking analysis
params = dict(Fs=16384, tapers=[1, 1], fpass=[400, 600], itc=0)
# plv to TFS
epochs = np.transpose(epochs,
(1, 0, 2)) # switching the first and second columns
plv, f = mtplv(epochs, params)
index = np.where(plv[31, ] == max(
plv[31, ])) # CHANGE AS NEEDED: 32st channel (Cz)
plv_32 = plv[31, ]
plv_32_max = plv_32[index]
f_max = f[index]
dimen = epochs.shape
numtrials = dimen[1]
# saving plvs
dictMat = {
"plv_32": plv_32_max,
"f": f_max,
"Fs": fs,
"trialNum": numtrials
}
savemat('/media/agudemu/Storage/Data/EEG/FFR/plvs/' + subj + Type, dictMat)
# collecting individual plvs into array for all subjects
return plv_32_max, f_max, numtrials
def bdf2mat(froot, fs, Fs_new, hpf, t_stim, topchans, trial_name):
full_raw, full_eves = EEGconcatenateFolder(froot,
nchans=34,
refchans=['EXG1', 'EXG2'],
exclude=[],
fs_new=Fs_new)
event_id = {'Positive': 1}
epochs = mne.Epochs(full_raw,
full_eves,
event_id,
tmin=t_stim[0],
tmax=t_stim[1],
reject_tmax=1.3,
picks=topchans,
reject=dict(eeg=100e-6))
epochs.load_data()
epochs_filtered = epochs.filter(hpf, None)
pos_data = epochs_filtered.get_data()
event_id = {'Negative': 2}
epochs = mne.Epochs(full_raw,
full_eves,
event_id,
tmin=t_stim[0],
tmax=t_stim[1],
reject_tmax=1.3,
picks=topchans,
reject=dict(eeg=100e-6))
epochs.load_data()
epochs_filtered = epochs.filter(hpf, None)
neg_data = epochs_filtered.get_data()
neg_l = len(neg_data)
pos_l = len(pos_data)
length = (neg_l >= pos_l) * pos_l + (neg_l < pos_l) * neg_l
tot_arr = np.zeros(2 * length, dtype=np.object)
ind = 0
for i in range(0, length):
tot_arr[ind] = pos_data[i]
tot_arr[ind + 1] = neg_data[i]
ind = ind + 2
os.chdir(froot)
savemat(trial_name + '_Data_full.mat',
{(trial_name + '_tot_arr'): tot_arr})
###########################PLOT SPECTROGRAM/PLV####################################
x = np.add(neg_data[:length, :, :], pos_data[:length, :, :])
params = dict(Fs=4e3, tapers=[1, 1], fpass=[0, 1000], itc=0)
S_psd, N_psd, f_psd = mtspec(x, params)
mt_plv, f_plv = mtplv(x, params)
fig, ax = plt.subplots(num=2, figsize=(12, 8))
ax.plot(f_psd, np.subtract(S_psd, N_psd))
#ax.plot(f_psd, N_psd)
ax.grid(color='k', linestyle='-', linewidth=1)
ax.set_xlabel('Freq (Hz)', fontsize=18)
ax.set_ylabel('PSD', fontsize=18)
ax.set_xlim([70., 500.])
ax.set_title('psd all')
fig, ax = plt.subplots(num=1, figsize=(12, 8))
ax.plot(f_plv, mt_plv)
#ax.plot(f_psd, N_psd)
ax.grid(color='k', linestyle='-', linewidth=1)
ax.set_xlabel('Freq (Hz)', fontsize=18)
ax.set_ylabel('PLV', fontsize=18)
ax.set_xlim([70., 500.])
ax.set_title('PLV all')
# Calculate power, plv
if not preEpoched:
Fs = raw.info['sfreq']
times = epochs.times
#########################################################################
# Fourier domain stuff
pl.figure()
params = dict(Fs=Fs, fpass=[5, 140], tapers=[1, 1], itc=0)
for badname in epochs.info['bads']:
bad_ind = epochs.info['ch_names'].index(badname)
if bad_ind in mags:
mags.remove(bad_ind)
y = x[:, mags, :].transpose((1, 0, 2))
plv, f = spectral.mtplv(y, params, verbose='DEBUG')
pl.plot(f, plv.T, linewidth=2)
pl.xlabel('Frequency (Hz)', fontsize=16)
pl.ylabel('Intertrial PLV', fontsize=16)
pl.title('MEG Magnetometers', fontsize=16)
if SSSR:
pl.xlim([70, 140])
else:
pl.xlim([15, 90])
pl.show()
figname_mag = subj + '_' + condstem + ssstag + '_mag-results.pdf'
matname_mag = subj + '_' + condstem + ssstag + '_mag_results.mat'
io.savemat(fpath + matname_mag, dict(f=f, plv=plv, chans=mags))
pl.savefig(fpath + figname_mag)
lout = mne.find_layout(epochs.info)
pos = lout.pos[mags]
dat_epochs_5 = dat_epochs_5[:, 0:32, t1:t2].transpose(1, 0, 2) dat_epochs_6 = dat_epochs_6[:, 0:32, t1:t2].transpose(1, 0, 2) dat_epochs_7 = dat_epochs_7[:, 0:32, t1:t2].transpose(1, 0, 2) dat_epochs_8 = dat_epochs_8[:, 0:32, t1:t2].transpose(1, 0, 2) dat_epochs_9 = dat_epochs_9[:, 0:32, t1:t2].transpose(1, 0, 2) params = dict() params['Fs'] = fs params['tapers'] = [2, 2 * 2 - 1] params['fpass'] = [1, 300] params['itc'] = 1 plt.figure() plt.plot(dat_epochs_7[31, :, :].T) plvtap_1, f = mtplv(dat_epochs_1, params) plvtap_2, f = mtplv(dat_epochs_2, params) plvtap_3, f = mtplv(dat_epochs_3, params) plvtap_4, f = mtplv(dat_epochs_4, params) plvtap_5, f = mtplv(dat_epochs_5, params) plvtap_6, f = mtplv(dat_epochs_6, params) plvtap_7, f = mtplv(dat_epochs_7, params) plvtap_8, f = mtplv(dat_epochs_8, params) plvtap_9, f = mtplv(dat_epochs_9, params) np.max(plvtap_8[:, 75:85], axis=1) plt.figure() plt.plot(f, plvtap_1.T) plt.title(labels[0])
fs = 16384
chop = 0.0e-3
chop2 = 22.e-3
plv_31s = np.zeros(len(subjectList))
plv_auds = np.zeros(len(subjectList))
trialNums_ref = np.zeros(len(subjectList))
for k, subj in enumerate(subjectList):
data_masker_sum, numtrials_masker = subjectProcessing(froot, subj, [1, 2], fs)
# phase locking analysis
params = dict(Fs = 16384, tapers = [1, 1], fpass = [100, 2000], itc = 0)
# plv to ABR (onset response)
data_masker_sum = np.transpose(data_masker_sum, (1, 0, 2)) # switching the first and second columns
plv, f = mtplv(data_masker_sum, params)
plv_31 = plv[30,]
plv_aud = plv[[4, 26, 25, 30, 31], :].mean(axis = 0)
plv_31_avg = plv_31.mean()
plv_aud_avg = plv_aud.mean()
# saving plvs
dictMat = {"plv_31": plv_31_avg, "plv_aud": plv_aud_avg, "Fs": fs, "trialNum": numtrials_masker}
savemat(subj + '_ABR', dictMat)
# collecting individual plvs into array for all subjects
#plv_31s[k] = plv_31_avg
#plv_auds[k] = plv_aud_avg
#trialNums_ref[k] = numtrials_masker
#dictDataArray = {"subjects": subjectList, "plv_31s": plv_31s, "plv_auds": plv_auds, "trialNums_ref": trialNums_ref}
#savemat('plvs_ABR', dictDataArray)
#plv = spectral_connectivity(data_adpt_diff, method = 'plv')
exclude=read_data_params['exclude'])
# raw= mne.io.read_raw_edf(fileVar)
# raw.load_data()
# eves= mne.find_events(raw, shortest_event=1, mask=255)
raw.filter(70, 1e3, phase='zero')
epochs = mne.Epochs(raw,
eves,
1,
tmin=-0.1,
proj=False,
tmax=1.2,
baseline=(-0.1, 0.0),
reject=dict(eeg=200e-6))
evoked = epochs.average()
params = dict(Fs=raw.info['sfreq'], tapers=[1, 1], fpass=[70, 1000], itc=0)
x = epochs.get_data()
x = x[:, :32, :]
x = x.transpose((1, 0, 2))
plv, f = mtplv(x[30, :, :], params)
if plotYes:
filename = op.basename(fileVar)
filename, _ = op.splitext(filename)
fig, ax = plt.subplots(num=1, figsize=(12, 8))
ax.plot(f, plv)
ax.grid(color='k', linestyle='-', linewidth=1)
ax.set_xlabel('Freq (Hz)', fontsize=18)
ax.set_ylabel('PLV', fontsize=18)
ax.set_xlim([70., 500.])
ax.set_title(filename)
plt.savefig(dir_dict['figOut'] + filename + '.png')
# raw.filter(70, 1e3, phase='zero') # Not needed here as mtspec/mtplv have filter params
epochs = mne.Epochs(raw,
eves,
1,
tmin=-0.1,
proj=False,
tmax=1.2,
baseline=(-0.1, 0.0),
picks=31,
reject=dict(eeg=200e-6))
evoked = epochs.average()
params = dict(Fs=raw.info['sfreq'], tapers=[1, 1], fpass=[70, 1000], itc=0)
x = epochs.get_data()
x = x[:, :32, :]
x = x.transpose((1, 0, 2))
plv, f_plv = mtplv(x[30, :, :], params)
S_psd, N_psd, f_psd = mtspec(x[30, :, :], params)
if plotYes:
fig, ax = plt.subplots(num=1, figsize=(12, 8))
ax.plot(f_plv, plv)
ax.grid(color='k', linestyle='-', linewidth=1)
ax.set_xlabel('Freq (Hz)', fontsize=18)
ax.set_ylabel('PLV', fontsize=18)
ax.set_xlim([70., 500.])
ax.set_title('plv all')
if saveYes:
plt.savefig(dir_dict['figOut'] + 'plv_all.png')
if plotYes:
fig, ax = plt.subplots(num=2, figsize=(12, 8))
# target_mods[0,:] = np.sin(2*np.pi*6*t)
# target_mods[1,:] = np.sin(2*np.pi*40*t)
dat_epochs_1 = epochs_1.get_data()
dat_epochs_2 = epochs_2.get_data()
dat_epochs_1 = dat_epochs_1[0:200, 0:32, t1:t2].transpose(1, 0, 2)
dat_epochs_2 = dat_epochs_2[0:200, 0:32, t1:t2].transpose(1, 0, 2)
params = dict()
params['Fs'] = fs
params['tapers'] = [2, 2 * 2 - 1]
params['fpass'] = [1, 300]
params['itc'] = 0
plvtap_1, f = mtplv(dat_epochs_1, params)
plvtap_2, f = mtplv(dat_epochs_2, params)
np.max(plvtap_2[:, 870:900], axis=1)
fig, ax = plt.subplots(figsize=(5.5, 5))
fontsize = 15
ax.plot(f, plvtap_2[31, :], label='CORR', linewidth=2)
ax.plot(f, plvtap_1[31, :], label='ACORR', linewidth=2)
#plt.title('223 Hz',fontsize=fontsize)
ax.legend(fontsize=fontsize)
plt.xlabel('Frequency (Hz)', fontsize=fontsize, fontweight='bold')
#plt.ylabel('PLV',fontsize=fontsize)
plt.xlim((218, 228))
plt.xticks([218, 223, 228], fontsize=fontsize)
plt.yticks([0, 0.04, 0.08], fontsize=fontsize)
evoked_AM40.plot(titles='AM 40', picks=Aud_picks, xlim=[-0.1, 2.2])
t = epochs_AM40.times
t1 = np.where(t >= 0)[0][0]
t2 = np.where(t >= 2.0)[0][0]
t = t[t1:t2]
epoch40_dat = epochs_AM40.get_data()
epoch40_dat = epoch40_dat[:, 0:27, t1:t2].transpose(1, 0, 2)
params = dict()
params['Fs'] = epochs_AM40.info['sfreq']
params['tapers'] = [2, 2 * 2 - 1]
params['fpass'] = [1, 100]
params['itc'] = 0
plvtap_1, f = spectral.mtplv(epoch40_dat, params)
Aud_inds = np.flip(np.argsort(np.max(plvtap_1[:, 75:85], axis=1)))
Aud_chans = []
for ind in Aud_inds[0:9]:
Aud_chans.append(epochs_AM40.ch_names[ind])
fig, ax = plt.subplots(figsize=(5, 3.3))
fontsize = 10
ax.plot(f, plvtap_1.T)
plt.ylabel('PLV', fontsize=fontsize)
plt.xlabel('Frequency (Hz)', fontsize=fontsize)
plt.xticks([0, 20, 40, 60, 80, 100], fontsize=fontsize)
plt.yticks([0.1, 0.2, 0.3], fontsize=fontsize)
fig, ax = plt.subplots(figsize=(5, 3.3))
epoch_dat.append(epoch_dat_m)
AMf = AMf[2:] #first 2 triggers mest up
epoch_dat = epoch_dat[2:]
params = dict()
params['Fs'] = epochs_all[0].info['sfreq']
params['tapers'] = [1, 2 * 1 - 1]
params['fpass'] = [1, 4000]
params['itc'] = 0
All_plvs = []
MaxAud_plvs = np.zeros((epoch_dat[0].shape[0], len(AMf)))
tmtf_plv = np.zeros((epoch_dat[0].shape[0], len(AMf)))
for m in np.arange(len(AMf)):
plvtap, f = spectral.mtplv(epoch_dat[m], params)
floc1 = np.where(f >= AMf[m] - 10)[0][0]
floc2 = np.where(f >= AMf[m] + 10)[0][0]
MaxAud_plvs[:,
m] = np.flip(np.argsort(np.max(plvtap[:, floc1:floc2],
axis=1)))
tmtf_plv[:, m] = np.max(plvtap[:, floc1:floc2], axis=1)
plt.figure()
plt.plot(f, plvtap.T)
plt.title(str(AMf[m]))
#plt.xlim((0,AMf[m]*3))
All_plvs.append(plvtap)
AMlabels = []
for AM in AMf:
AMlabels.append(str(AM))
params = dict(Fs=fs,
fpass=[5, 1000],
tapers=[1, 1],
Npairs=2000,
itc=1)
Ntrials = x.shape[1]
print 'Running Mean Spectrum Estimation'
(S, N, f) = spectral.mtspec(x, params, verbose=True)
print 'Running CPCA PLV Estimation'
(cplv, f) = spectral.mtcpca(x, params, verbose=True)
print 'Running channel by channel PLV Estimation'
(plv, f) = spectral.mtplv(x, params, verbose=True)
print 'Running CPCA Power Estimation'
(cpow, f) = spectral.mtcspec(x, params, verbose=True)
print 'Running raw spectrum estimation'
(Sraw, f) = spectral.mtspecraw(x, params, verbose=True)
# Saving Results
res = dict(cpow=cpow,
plv=plv,
cplv=cplv,
Sraw=Sraw,
f=f,
S=S,
N=N,
else:
x = np.concatenate((x, xtemp), axis=1)
else:
continue
else:
RuntimeError('No BDF files found!!')
# Average data
goods = [1, 5, 6, 27, 28, 30, 31]
if ear == 'L':
refchan = 34
else:
refchan = 35
y = x[goods, :, :].mean(axis=0) - x[refchan, :, :]
params = dict(Fs=raw.info['sfreq'], tapers=[4, 7],
fpass=[1, 4000], itc=0)
plv, f = mtplv(y, params)
ph, f_ph = mtphase(y, params)
tdresp = np.median(y, axis=0) * 1.0e6 # Just time domain average
# Make dictionary and save
mdict = dict(t=t, x=tdresp, f=f, f_ph=f_ph, ph=ph, plv=plv)
savepath = froot + '/CentralGainResults/'
savename = subj + name + '_CentralGain.mat'
savemat(savepath + savename, mdict)
pl.plot(f, plv)
pl.subplot(413)
pl.plot(t, diffAdpt, 'b')
pl.xlim([0.01, 0.04])
pl.ylim([-4e-7, 4e-7])
pl.subplot(414)
pl.plot(t, diffAdptHp, 'b')
pl.xlim([0.01, 0.04])
pl.ylim([-4e-7, 4e-7])
# frequency component of polarity dependent adapted component
diffAdpt_fft = np.fft.fft(diffAdptHp)
magAdpt = np.abs(diffAdpt_fft)
phase = np.angle(diffAdpt_fft)
freq = np.linspace(0, fs, len(diffAdpt_fft))
pl.figure(3)
pl.plot(freq, magAdpt)
# PLV calculation: between stimulus and evoked response
fstim = 500
stim = np.sin(2*np.pi*fstim*t)
anl_stim = hilbert(stim)
anl_resp = hilbert(diffAdptHp)
pAgl_stim = np.angle(anl_stim)
pAgl_resp = np.angle(anl_resp)
plv = np.abs(np.sum(np.exp(1j*(pAgl_stim - pAgl_resp))))/len(pAgl_stim) # plv across time works for narrow-band
# PLV calculations across trials
mtplv()
else:
RuntimeError("No bdf files found!")
