def plot_erp_specgrams(
d,
samplerate=None,
NFFT=256,
freq_range=[0.1, 50],
fig=None):
assert d.data.ndim == 3
assert d.feat_lab is not None
if fig is None:
fig = plot.figure()
tf = trials.trial_specgram(d, samplerate, NFFT)
tf_erp = np.mean(tf.d, axis=3)
ch_labs = tf.feat_lab[0]
freqs = np.array([float(x) for x in tf.feat_lab[1]])
times = np.array([float(x) for x in tf.feat_lab[2]])
selection = np.logical_and(freqs >= freq_range[0], freqs <= freq_range[1])
freqs = freqs[selection]
tf_erp = tf_erp[:,selection,:]
clim = (-np.max(np.abs(tf_erp)), np.max(np.abs(tf_erp)))
num_channels = tf_erp.shape[0]
num_cols = max(1, num_channels/8)
num_rows = min(num_channels, 8)
fig.subplots_adjust(hspace=0)
for channel in range(num_channels):
s = tf_erp[channel,:,:]
col = channel / num_rows
row = channel % num_rows
ax = plot.subplot(num_rows, num_cols, num_cols*row+col+1)
im = plot.imshow(
np.flipud(s), aspect='auto',
extent=[np.min(times), np.max(times), np.min(freqs), np.max(freqs)],
)
plot.ylim(freq_range[0], freq_range[1])
plot.clim(clim)
plot.ylabel(ch_labs[channel])
if row == num_rows-1 or channel == num_channels-1:
plot.xlabel('Time (s)')
else:
ax.get_xaxis().set_visible(False)
cax = fig.add_axes([0.91, 0.1, 0.01, 0.8])
fig.colorbar(im, cax=cax)
return fig
def plot_erp_specdiffs(
d,
samplerate=None,
NFFT=256,
freq_range=[0.1, 50],
classes=[0,1],
significant_only=False,
pval=0.05,
fig=None):
assert d.data.ndim == 3
assert len(classes) == 2
assert d.feat_lab is not None
if fig is None:
fig = plot.figure()
tf = trials.trial_specgram(d, samplerate, NFFT)
tf_erp = trials.erp(tf)
diff = np.log(tf_erp.data[...,classes[0]]) - np.log(tf_erp.data[...,classes[1]])
if significant_only:
_,ps = scipy.stats.ttest_ind(tf.get_class(classes[0]).data,
tf.get_class(classes[1]).data, axis=3)
diff[ps > pval] = 0
ch_labs = tf_erp.feat_lab[0]
freqs = np.array([float(x) for x in tf_erp.feat_lab[1]])
times = np.array([float(x) for x in tf_erp.feat_lab[2]])
selection = np.logical_and(freqs >= freq_range[0], freqs <= freq_range[1])
freqs = freqs[selection]
diff = diff[:,selection]
clim = (-np.max(np.abs(diff)), np.max(np.abs(diff)))
num_channels = d.data.shape[0]
num_cols = max(1, num_channels/8)
num_rows = min(num_channels, 8)
fig.subplots_adjust(hspace=0)
cdict = {'red': ((0.0, 1.0, 1.0),
(0.5, 1.0, 1.0),
(1.0, 0.0, 0.0)),
'green': ((0.0, 0.0, 0.0),
(0.5, 1.0, 1.0),
(1.0, 0.0, 0.0)),
'blue': ((0.0, 0.0, 0.0),
(0.5, 1.0, 1.0),
(1.0, 1.0, 1.0))}
cmap = matplotlib.colors.LinearSegmentedColormap('polarity',cdict,256)
for channel in range(num_channels):
s = diff[channel,:,:]
col = channel / num_rows
row = channel % num_rows
ax = plot.subplot(num_rows, num_cols, num_cols*row+col+1)
im = plot.imshow(
s, aspect='auto',
extent=[np.min(times), np.max(times), np.min(freqs), np.max(freqs)],
cmap=cmap
)
plot.ylim(freq_range[0], freq_range[1])
plot.clim(clim)
plot.ylabel(ch_labs[channel])
ax.xaxis.grid(True, color='w', which='major')
ax.yaxis.grid(False)
if row == num_rows-1 or channel == num_channels-1:
plot.xlabel('Time (s)')
else:
[label.set_visible(False) for label in ax.get_xticklabels()]
[tick.set_visible(False) for tick in ax.get_xticklines()]
cax = fig.add_axes([0.91, 0.1, 0.01, 0.8])
fig.colorbar(im, cax=cax)
return fig