moabb/scripts/data_visualization_p300.py at develop · goodcpp/moabb · GitHub

Repository navigation

History

233 lines (195 loc) · 7.21 KB

Raw

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

220

221

222

223

224

225

226

227

228

229

230

231

232

233

"""

===========================

Script for visualization of ALL P300 datasets

===========================

This script will download ALL P300 datasets and create

descriptive plots for every single session.

Total downloaded size will be (as of now) 120GB.

.. versionadded:: 0.4.5

"""

import warnings

# Authors: Jan Sosulski <mail@jan-sosulski.de>

# License: BSD (3-clause)

from pathlib import Path

import matplotlib

import mne

import numpy as np

import seaborn as sns

from matplotlib import pyplot as plt

from moabb.paradigms import P300

matplotlib.use("agg")

sns.set_style("whitegrid")

mne.set_log_level("WARNING")

def create_plot_overview(epo, plot_opts=None, path=None, description=""):

# Butterflyplot

suptitle = f"{description} ({epo_summary(epo)[1]})"

epo_t = epo["Target"]

epo_nt = epo["NonTarget"]

evkd_t = epo_t.average()

evkd_nt = epo_nt.average()

ix_t = epo.events[:, 2] == epo.event_id["Target"]

ix_nt = epo.events[:, 2] == epo.event_id["NonTarget"]

fig0, ax = plt.subplots(1, 1, figsize=(10, 3), sharey="all", sharex="all")

ax.scatter(

epo.events[ix_t, 0],

np.ones((np.sum(ix_t),)),

color="r",

marker="|",

label="Target",

)

ax.scatter(

epo.events[ix_nt, 0],

np.zeros((np.sum(ix_nt),)),

color="b",

marker="|",

label="NonTarget",

)

ax.legend()

ax.set_title("Event timeline")

fig0.suptitle(suptitle)

fig0.tight_layout()

fig0.savefig(path / f"event_timeline.{plot_opts['format']}", dpi=plot_opts["dpi"])

fig1, axes = plt.subplots(2, 1, figsize=(6, 6), sharey="all", sharex="all")

evkd_t.plot(spatial_colors=True, show=False, axes=axes[0])

axes[0].set_title("Target response")

evkd_nt.plot(spatial_colors=True, show=False, axes=axes[1])

axes[1].set_title("NonTarget response")

fig1.suptitle(suptitle)

with warnings.catch_warnings():

warnings.simplefilter("ignore")

fig1.tight_layout()

fig1.savefig(

path / f"target_nontarget_erps.{plot_opts['format']}", dpi=plot_opts["dpi"]

)

# topomap

tp = plot_opts["topo"]["timepoints"]

tmin, tmax = plot_opts["topo"]["tmin"], plot_opts["topo"]["tmax"]

times = np.linspace(tmin, tmax, tp)

fig2 = evkd_t.plot_topomap(times=times, colorbar=True, show=False)

fig2.suptitle(suptitle)

fig2.savefig(

path / f"target_topomap_{tp}_timepoints.{plot_opts['format']}",

dpi=plot_opts["dpi"],

)

# jointmap

fig3 = evkd_t.plot_joint(show=False)

fig3.suptitle(suptitle)

fig3.savefig(path / f"target_erp_topo.{plot_opts['format']}", dpi=plot_opts["dpi"])

# sensorplot

fig4 = mne.viz.plot_compare_evokeds(

[evkd_t.crop(0, 0.6), evkd_nt.crop(0, 0.6)], axes="topo", show=False

)

fig4[0].suptitle(suptitle)

fig4[0].savefig(path / f"sensorplot.{plot_opts['format']}", dpi=plot_opts["dpi"])

fig5, ax = plt.subplots(2, 1, figsize=(8, 6), sharex="all", sharey="all")

t_data = epo_t.get_data() * 1e6

nt_data = epo_nt.get_data() * 1e6

data = epo.get_data() * 1e6

minmax = np.max(data, axis=2) - np.min(data, axis=2)

per_channel = np.mean(minmax, axis=0)

worst_ch = np.argsort(per_channel)

worst_ch = worst_ch[max(-8, -len(epo.ch_names)) :]

minmax_t = np.max(t_data, axis=2) - np.min(t_data, axis=2)

minmax_nt = np.max(nt_data, axis=2) - np.min(nt_data, axis=2)

ch = epo_t.ch_names

for i in range(minmax_nt.shape[1]):

lab = ch[i] if i in worst_ch else None

sns.kdeplot(minmax_t[:, i], ax=ax[0], label=lab, clip=(0, 300))

sns.kdeplot(minmax_nt[:, i], ax=ax[1], label=lab, clip=(0, 300))

ax[0].set_xlim(0, 200)

ax[0].set_title("Target minmax")

ax[1].set_title("NonTarget minmax")

ax[1].set_xlabel("Minmax in $\\mu$V")

ax[1].legend(title="Worst channels")

fig5.suptitle(suptitle)

fig5.tight_layout()

fig5.savefig(path / f"minmax.{plot_opts['format']}", dpi=plot_opts["dpi"])

with warnings.catch_warnings():

warnings.simplefilter("ignore")

fig6 = epo.plot_psd(0, 20, bandwidth=1)

fig6.suptitle(suptitle)

fig6.tight_layout()

fig6.savefig(path / f"spectrum.{plot_opts['format']}", dpi=plot_opts["dpi"])

plt.close("all")

def epo_summary(epos):

summary = dict()

summary["mne_string"] = repr(epos)

summary["n_channels"] = len(epos.ch_names)

summary["n_target"] = len(epos["Target"])

summary["n_nontarget"] = len(epos["NonTarget"])

info_str = (

f"Ch:{len(epos.ch_names)},T:{len(epos['Target'])},NT:{len(epos['NonTarget'])}"

)

return summary, info_str

if __name__ == "__main__":

FIGURES_PATH = Path.home() / "moabb_figures" / "erps"

# Changing this to False re-generates all plots even if they exist. Use with caution.

cache_plots = True

baseline = None

highpass = 0.5

lowpass = 16

sampling_rate = 100

paradigm = P300(

resample=sampling_rate,

fmin=highpass,

fmax=lowpass,

baseline=baseline,

)

ival = [-0.3, 1]

plot_opts = {

"dpi": 120,

"topo": {

"timepoints": 10,

"tmin": 0,

"tmax": 0.6,

"format": "png",

}

plt.ioff()

# dsets = P300_DSETS

dsets = paradigm.datasets

for dset in dsets:

dset.interval = ival

dset_name = dset.__class__.__name__

print(f"Processing dataset: {dset_name}")

data_path = FIGURES_PATH / dset_name # path of the dataset folder

data_path.mkdir(exist_ok=True)

all_subjects_cached = True

for subject in dset.subject_list:

subject_path = data_path / f"subject_{subject}"

if cache_plots and subject_path.exists():

continue

all_subjects_cached = False

print(f" Processing subject: {subject}")

subject_path.mkdir(parents=True, exist_ok=True)

try:

epos, labels, meta = paradigm.get_data(

dset, [subject], return_epochs=True

)

except Exception: # catch all, dont stop processing pls

print(f"Failed to get data for {dset_name}-{subject}")

(subject_path / "processing_error").touch()

continue

description = f"Dset: {dset_name}, Sub: {subject}, Ses: all"

create_plot_overview(

epos,

plot_opts=plot_opts,

path=subject_path,

description=description,

)

if len(meta["session"].unique()) > 1:

for session in meta["session"].unique():

session_path = subject_path / f"session_{session}"

session_path.mkdir(parents=True, exist_ok=True)

ix = meta.session == session

description = f"Dset: {dset_name}, Sub: {subject}, Ses: {session}"

create_plot_overview(

epos[ix],

plot_opts=plot_opts,

path=session_path,

description=description,

)

if all_subjects_cached:

print(" No plots necessary, every subject has output folder.")

print("All datasets processed.")

moabb/scripts/data_visualization_p300.py at develop · goodcpp/moabb · GitHub

Repository navigation

Expand file tree

Breadcrumbs

data_visualization_p300.py

Latest commit

History

Breadcrumbs

data_visualization_p300.py

File metadata and controls