def _update_target_position(self, xyz=None):
# Check XYZ :
sh = xyz.shape
assert sh[1] in [2, 3]
self._n_sources = sh[0]
pos = xyz if sh[1] == 3 else np.c_[xyz, np.full((len(self), ), _z)]
self._target_xyz = vispy_array(pos)
def on_timer(*args, **kwargs):
if hasattr(brain_obj, 'camera'):
brain_obj.camera.azimuth += 1
t = app_timer.elapsed
frame = int(np.floor(t * sampling_frequency))
new_data = data[:, frame % data.shape[1]].ravel()
source_object._data = vispy_array(new_data)
source_object.update()
source_object.project_sources(brain_obj)
source_object.color_sources()
def _update_position(self, xyz=None):
# Check XYZ :
sh = xyz.shape
assert sh[1] in [2, 3]
self._n_sources = sh[0]
pos = xyz if sh[1] == 3 else np.c_[xyz, np.full((len(self), ), _z)]
self._xyz = vispy_array(pos)
# Assign XYZ
self._sources._data['a_position'] = self._xyz
self.update()
def set_data(self, data, time=None):
"""Set data to the grid of signals.
Parameters
----------
data : array_like
Array of data of shape (n_pts,).
time : array_like | None
Array of time points of shape (n_pts,)
"""
data = np.asarray(data)
assert data.ndim == 1
self._n = len(data)
time = np.arange(len(data)) if time is None else np.asarray(time)
assert len(time) == len(self)
# Transient detection :
self.transient = data
#
self._pos = vispy_array(np.c_[time, data])
self._position_vbo.set_data(self._pos)
def set_data(self, data, time=None):
"""Set data to the grid of signals.
Parameters
----------
data : array_like
Array of data of shape (n_pts,).
time : array_like | None
Array of time points of shape (n_pts,)
"""
data = np.asarray(data)
assert data.ndim == 1
self._n = len(data)
time = np.arange(len(data)) if time is None else np.asarray(time)
assert len(time) == len(self)
# Transient detection :
self.transient = data
#
self._pos = vispy_array(np.c_[time, data])
self._position_vbo.set_data(self._pos)
def set_data(self,
data,
index,
color='black',
lw=2.,
nbins=10,
symbol='disc',
size=10.,
form='line',
th=None,
norm=None,
window=None,
overlap=0.,
baseline=None,
clim=None,
cmap='viridis',
interpolation='gaussian',
nperseg=256,
noverlap=128):
"""Set data to the plot.
Parameters
----------
data : array_like
Raw data vector of shape (N,)
index : int | 0
Index of the 3-d array.
color : array_like/string/tuple | None
Color of the plot.
lw : float | None
Line width (form='line').
symbol : string | None
Marker symbol (form='marker').
size : float | None
Marker size (form='marker').
nbins : int | None
Number of bins for the histogram (form='histogram')
form : {'line', 'marker', 'histogram', 'tf'}
Plotting type.
th : tuple | None
Tuple of floats for line thresholding.
norm : int | None
Normalization method for (form='tf').
window : tuple | None
Averaging window (form='tf').
overlap : float | 0.
Overlap between successive windows (form='tf').
baseline : tuple | None
Baseline period for the normalization (form='tf').
"""
# Update variable :
self.form = form
self._index = index
color = color2vb(color)
# Get data index :
if data.ndim == 1:
idx = slice(None)
elif data.ndim in [2, 3]:
idx = list(self._navidx[index])
idx.insert(self._axis, slice(None))
idx = tuple(idx)
# Convert data to be compatible with VisPy and prepare data :
data_c = vispy_array(data[idx]).copy()
_data = self._prep._prepare_data(self._sf, data_c, self._time)
# Set data :
if form in ['line', 'marker', 'psd', 'butterfly']: # line and marker
# Get position array :
pos = np.c_[self._time, _data]
# Send position :
if form in ['line', 'psd']:
if form == 'psd':
fmax = self._sf / 4.
f, pxx = welch(_data,
self._sf,
nperseg=nperseg,
noverlap=noverlap)
f_sf4 = abs(f - fmax)
f_1 = abs(f - 1.)
fidx_sf4 = np.where(f_sf4 == f_sf4.min())[0][0]
fidx_1 = np.where(f_1 == f_1.min())[0][0]
pos = np.c_[f[fidx_1:-fidx_sf4], pxx[fidx_1:-fidx_sf4]]
# Threshold :
is_th = isinstance(th, (tuple, list, np.ndarray))
col = color2vb(color, length=pos.shape[0])
if is_th:
# Build threshold segments :
t_min, t_max = self._time.min(), self._time.max()
pos_th = np.vstack(
([t_min, th[0]], [t_max,
th[0]], [t_min,
th[1]], [t_max, th[1]]))
self._th.set_data(pos_th,
connect='segments',
color=color2vb('#ab4642'))
# Build line color :
col = color2vb(color, length=len(_data))
cond = np.logical_or(_data < th[0], _data > th[1])
col[cond, :] = color2vb('#ab4642')
self._th.visible = is_th
self._line.set_data(pos, width=lw, color=col)
self._line.update()
elif form == 'marker':
self._mark.set_data(pos,
face_color=color,
symbol=symbol,
size=size,
edge_width=0.)
self._mark.update()
elif form == 'butterfly':
# Get soe shape related variables :
n, m = len(self._time), int(np.prod(data.shape))
n_rep = int(m / n)
data = vispy_array(data)
# Build position :
pos = np.c_[np.tile(self._time.ravel(), n_rep), data.ravel()]
# Disconnect some points :
connect = np.c_[np.arange(m - 1), np.arange(1, m)]
to_delete = np.linspace(n - 1, m - 1, n_rep)
connect = np.delete(connect, to_delete, axis=0)
# Build color :
col = color2vb(color, length=m)
# singcol = np.random.uniform(size=(n_rep, 3), low=.2,
# high=.8).astype(np.float32)
# col = np.repeat(singcol, n, 0)
# Send data :
self._line.set_data(pos, width=lw, color=col, connect=connect)
self._line.update()
# Get camera rectangle :
t_min, t_max = pos[:, 0].min(), pos[:, 0].max()
d_min, d_max = pos[:, 1].min(), pos[:, 1].max()
off = .05 * (d_max - d_min)
self.rect = (t_min, d_min - off, t_max - t_min,
d_max - d_min + 2 * off)
elif form == 'histogram': # histogram
# Compute the mesh :
mesh = scene.visuals.Histogram(_data, nbins)
# Get the vertices and faces of the mesh :
vert = mesh.mesh_data.get_vertices()
faces = mesh.mesh_data.get_faces()
# Pass vertices and faces to the histogram :
self._hist.set_data(vert, faces, color=color)
# Compute the histogram :
raw, xvec = np.histogram(_data, nbins)
# Get camera rectangle :
t_min, t_max = xvec.min(), xvec.max()
d_min, d_max = 0.9 * raw[np.nonzero(raw)].min(), 1.01 * raw.max()
self.rect = (t_min, d_min, t_max - t_min, d_max - d_min)
# Update object :
self._hist.update()
elif form == 'tf': # time-frequency map
self._tf.set_data(_data,
self._sf,
cmap=cmap,
contrast=.5,
norm=norm,
baseline=baseline,
n_window=window,
overlap=overlap,
window='hanning',
clim=clim)
self._tf.interpolation = interpolation
self.rect = self._tf.rect
# Hide non form elements :
self._visibility()
# Update annotations :
self.update_annotations(str(self))
def __init__(self,
name,
xyz,
data=None,
color='red',
alpha=1.,
symbol='disc',
radius_min=5.,
radius_max=10.,
edge_width=0.,
edge_color='black',
system='mni',
mask=None,
mask_color='gray',
mask_radius=5.,
text=None,
text_size=2.,
text_color='white',
text_bold=False,
text_translate=(0., 2., 0.),
visible=True,
transform=None,
parent=None,
verbose=None,
_z=-10.,
**kw):
"""Init."""
VisbrainObject.__init__(self, name, parent, transform, verbose, **kw)
# _______________________ CHECKING _______________________
# XYZ :
sh = xyz.shape
assert sh[1] in [2, 3]
self._n_sources = sh[0]
pos = xyz if sh[1] == 3 else np.c_[xyz, np.full((len(self), ), _z)]
logger.info(' %i sources detected' % self._n_sources)
# Radius min and max :
assert all(
[isinstance(k, (int, float)) for k in (radius_min, radius_max)])
radius_max = max(radius_min, radius_max)
self._radius_min, self._radius_max = radius_min, radius_max
self._mask_radius = mask_radius
# Data :
if data is None:
data = np.ones((len(self), ))
else:
assert np.shape(data)[0] == len(self)
self._data = vispy_array(data)
# System :
pos = pos if system == 'mni' else tal2mni(pos)
self._xyz = vispy_array(pos)
# Color :
self._color = color
# Edges :
self._edge_color, self._edge_width = edge_color, edge_width
# Mask :
if mask is None:
mask = [False] * len(self)
self._mask = np.asarray(mask).ravel().astype(bool)
assert len(self._mask) == len(self)
self._mask_color = color2vb(mask_color)
# Text :
self._text_size = text_size
self._text_color = text_color
self._text_translate = text_translate
# _______________________ MARKERS _______________________
self._sources = visuals.Markers(pos=self._xyz,
name='Markers',
edge_color=edge_color,
edge_width=edge_width,
symbol=symbol,
parent=self._node)
self._sources.set_gl_state('translucent',
depth_test=True,
cull_face=False)
# _______________________ TEXT _______________________
tvisible = text is None
self._text = [''] * len(self) if tvisible else text
self._text = np.array(self._text)
assert len(self._text) == len(self)
self._sources_text = visuals.Text(self._text,
pos=self._xyz,
bold=text_bold,
name='Text',
color=color2vb(text_color),
font_size=text_size,
parent=self._node)
self._sources_text.visible = not tvisible
tr = vist.STTransform(translate=text_translate)
self._sources_text.transform = tr
# _______________________ UPDATE _______________________
# Radius / color :
self.visible = visible
self._update_radius()
self.alpha = alpha
self._update_color()
def set_data(self, data=None, axis=None, color=None, title=None,
force_shape=None, plt_as='grid'):
"""Set data to the grid of signals.
Parameters
----------
data : None
Array of data. Could be 1-D, 2-D or 3-D.
axis : int | None
Time axis location.
random : array_like/string/tuple | 'random'
Use 'random' for random colors or a color name for uniform color.
"""
# ====================== CHECKING ======================
# Axis :
axis = axis if isinstance(axis, int) else self._axis
axis = len(self._sh) - 1 if axis == -1 else axis
# Data :
if isinstance(data, np.ndarray):
# -------------- (n_rows, n_cols, n_time) --------------
if data.ndim == 1: # 1-D array
data = data.reshape(1, 1, -1)
g_size = (1, 1)
elif data.ndim == 2: # 2-D array
if axis == 0: # data need to be transposed
data = np.swapaxes(data, 0, 1)
axis = 1
g_size = (data.shape[0], 1) # (n_row, 1)
data = data[np.newaxis, ...]
elif data.ndim == 3: # 3-D array
if axis != data.ndim - 1: # data need to be transposed
data = np.swapaxes(data, axis, -1)
axis = data.ndim - 1
g_size = (data.shape[0], data.shape[1])
# -------------- Signals index --------------
m = np.prod(list(data.shape)[0:-1])
sig_index = np.arange(m).reshape(*g_size)
# -------------- Plot type --------------
if plt_as == 'row':
force_shape = (1, g_size[0] * g_size[1])
elif plt_as == 'col':
force_shape = (g_size[0] * g_size[1], 1)
# -------------- Optimal 2-D --------------
self._data = data
self._ori_shape = list(data.shape)[0:-1]
if force_shape is None:
n_rows, n_cols = ndsubplot(m)
elif len(g_size) == 2:
n_rows, n_cols = force_shape
data = data.reshape(n_rows, n_cols, len(self))
sig_index = sig_index.reshape(n_rows, n_cols)
g_size = (n_rows, n_cols)
self._opt_shape = list(data.shape)[0:-1]
self._sig_index = sig_index
# -------------- (n_rows * n_cols, n_time) --------------
data = np.reshape(data, (m, len(self)), order='F')
# -------------- Prepare --------------
# Force demean / detrend of _prep :
self._prep.demean, self._prep.detrend = False, False
data = self._prep._prepare_data(self._sf, data, 0)
# Demean and normalize :
kw = {'axis': -1, 'keepdims': True}
dmax = np.abs(data).max(**kw)
dmax[dmax == 0.] = 1.
data -= data.mean(**kw)
data /= dmax
# data /= data.max()
self._dbuffer.set_data(vispy_array(data))
self.g_size = g_size
# ====================== INDEX ======================
n, m = len(self), np.prod(g_size)
self._sig_index = self._sig_index.reshape(n_rows, n_cols)
idg = np.c_[np.repeat(np.repeat(np.arange(n_cols), n_rows), n),
np.repeat(np.tile(np.arange(n_rows), n_cols), n)[::-1],
np.tile(np.arange(n), m)].astype(np.float32)
self._ibuffer.set_data(vispy_array(idg))
# ====================== COLOR ======================
if color is not None:
color_1d = color2vb(color)
self.shared_program.frag['u_color'] = color_1d.ravel()
# ====================== TITLES ======================
# Titles checking :
if title is None or (len(title) != m):
st, it = '({}, {})', product(range(n_rows), range(n_cols))
title = [st.format(i, k) for i, k in it]
# Set text and font size :
if not self._txt.text:
self._txt.text = title
# Get titles position :
x_factor, y_factor = 1. / n_cols, 1. / n_rows
r_x = np.linspace(-1. + x_factor, 1. - x_factor, n_cols)
r_x = np.tile(r_x, n_rows)
r_y = np.linspace(-1. + y_factor, 1. - y_factor, n_rows)[::-1]
r_y += y_factor
r_y = np.repeat(r_y, n_cols)
pos = np.c_[r_x, r_y, np.full_like(r_x, -10.)]
self._txt.pos = pos.astype(np.float32)
def set_data(self, data=None, axis=None, color=None, title=None,
force_shape=None, plt_as='grid'):
"""Set data to the grid of signals.
Parameters
----------
data : None
Array of data. Could be 1-D, 2-D or 3-D.
axis : int | None
Time axis location.
random : array_like/string/tuple | 'random'
Use 'random' for random colors or a color name for uniform color.
"""
# ====================== CHECKING ======================
# Axis :
axis = axis if isinstance(axis, int) else self._axis
axis = len(self._sh) - 1 if axis == -1 else axis
# Data :
if isinstance(data, np.ndarray):
# -------------- (n_rows, n_cols, n_time) --------------
if data.ndim == 1: # 1-D array
data = data.reshape(1, 1, -1)
g_size = (1, 1)
elif data.ndim == 2: # 2-D array
if axis == 0: # data need to be transposed
data = np.swapaxes(data, 0, 1)
axis = 1
g_size = (data.shape[0], 1) # (n_row, 1)
data = data[np.newaxis, ...]
elif data.ndim == 3: # 3-D array
if axis != data.ndim - 1: # data need to be transposed
data = np.swapaxes(data, axis, -1)
axis = data.ndim - 1
g_size = (data.shape[0], data.shape[1])
# -------------- Signals index --------------
m = np.prod(list(data.shape)[0:-1])
sig_index = np.arange(m).reshape(*g_size)
# -------------- Plot type --------------
if plt_as == 'row':
force_shape = (1, g_size[0] * g_size[1])
elif plt_as == 'col':
force_shape = (g_size[0] * g_size[1], 1)
# -------------- Optimal 2-D --------------
self._data = data
self._ori_shape = list(data.shape)[0:-1]
if force_shape is None:
n_rows, n_cols = ndsubplot(m)
elif len(g_size) == 2:
n_rows, n_cols = force_shape
data = data.reshape(n_rows, n_cols, len(self))
sig_index = sig_index.reshape(n_rows, n_cols)
g_size = (n_rows, n_cols)
self._opt_shape = list(data.shape)[0:-1]
self._sig_index = sig_index
# -------------- (n_rows * n_cols, n_time) --------------
data = np.reshape(data, (m, len(self)), order='F')
# -------------- Prepare --------------
# Force demean / detrend of _prep :
self._prep.demean, self._prep.detrend = False, False
data = self._prep._prepare_data(self._sf, data, 0)
# Demean and normalize :
kw = {'axis': -1, 'keepdims': True}
dmax = np.abs(data).max(**kw)
dmax[dmax == 0.] = 1.
data -= data.mean(**kw)
data /= dmax
# data /= data.max()
self._dbuffer.set_data(vispy_array(data))
self.g_size = g_size
# ====================== INDEX ======================
n, m = len(self), np.prod(g_size)
self._sig_index = self._sig_index.reshape(n_rows, n_cols)
idg = np.c_[np.repeat(np.repeat(np.arange(n_cols), n_rows), n),
np.repeat(np.tile(np.arange(n_rows), n_cols), n)[::-1],
np.tile(np.arange(n), m)].astype(np.float32)
self._ibuffer.set_data(vispy_array(idg))
# ====================== COLOR ======================
if color is not None:
color_1d = color2vb(color)
self.shared_program.frag['u_color'] = color_1d.ravel()
# ====================== TITLES ======================
# Titles checking :
if title is None or (len(title) != m):
st, it = '({}, {})', product(range(n_rows), range(n_cols))
title = [st.format(i, k) for i, k in it]
# Set text and font size :
if not self._txt.text:
self._txt.text = title
# Get titles position :
x_factor, y_factor = 1. / n_cols, 1. / n_rows
r_x = np.linspace(-1. + x_factor, 1. - x_factor, n_cols)
r_x = np.tile(r_x, n_rows)
r_y = np.linspace(-1. + y_factor, 1. - y_factor, n_rows)[::-1]
r_y += y_factor
r_y = np.repeat(r_y, n_cols)
pos = np.c_[r_x, r_y, np.full_like(r_x, -10.)]
self._txt.pos = pos.astype(np.float32)