wonambi-python.github.io/gui/methods.html at master · wonambi-python/wonambi-python.github.io

History

336 lines (321 loc) · 37 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

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

<!DOCTYPE html>

<head>

<title>Detection methods — wonambi 7.11 documentation</title>

<!--[if lt IE 9]>

<![endif]-->

</head>

<a href="../index.html" class="icon icon-home"> wonambi

</a>

7.11

</div>

</form>

</div>

</div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">

<li class="toctree-l1"><a class="reference internal" href="../introduction.html">WONAMBI</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#features">Features</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#installation">Installation</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#documentation">Documentation</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#change-log">Change Log</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#requirements">Requirements</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#optional-requirements">Optional Requirements</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#status">Status</a></li>

<li class="toctree-l2"><a class="reference internal" href="../introduction.html#license">License</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../installation.html">Installation</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../installation.html#linux-mac-os-x">Linux / Mac OS X</a></li>

<li class="toctree-l2"><a class="reference internal" href="../installation.html#windows">Windows</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../analysis/index.html">Tutorial / Analysis</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../analysis/tutorial.html">Get Started</a></li>

<li class="toctree-l2"><a class="reference internal" href="../analysis/spectrum.html">Frequency Domain</a></li>

<li class="toctree-l2"><a class="reference internal" href="../analysis/plot3d.html">Plot 3D images</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../cmd.html">Command Line</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../cmd.html#wonambi">wonambi</a></li>

<li class="toctree-l2"><a class="reference internal" href="../cmd.html#won-convert">won_convert</a></li>

</ul>

</li>

<li class="toctree-l1 current"><a class="reference internal" href="index.html">Graphical User Interface</a><ul class="current">

<li class="toctree-l2"><a class="reference internal" href="open.html">Open Dataset</a></li>

<li class="toctree-l2"><a class="reference internal" href="labels.html">Edit Labels</a></li>

<li class="toctree-l2"><a class="reference internal" href="channels.html">Plot Channels</a></li>

<li class="toctree-l2"><a class="reference internal" href="traces.html">Navigation</a></li>

<li class="toctree-l2"><a class="reference internal" href="notes.html">Annotations</a></li>

<li class="toctree-l2"><a class="reference internal" href="detect.html">Event detection</a></li>

<li class="toctree-l2 current"><a class="current reference internal" href="#">Detection methods</a></li>

<li class="toctree-l2"><a class="reference internal" href="analysis.html">Analysis console</a></li>

<li class="toctree-l2"><a class="reference internal" href="screenshot.html">Screenshot</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../api/wonambi.html">Modules, classes, functions</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../api/wonambi.html#subpackages">Subpackages</a></li>

<li class="toctree-l2"><a class="reference internal" href="../api/wonambi.html#submodules">Submodules</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../testing.html">Development</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#prepare-test-environment">Prepare Test Environment</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#get-files">1. Get Files</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#run-the-tests">2. Run the Tests</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#coverage">3. Coverage</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#documentation">4. Documentation</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#release">5. Release</a></li>

<li class="toctree-l2"><a class="reference internal" href="../testing.html#tips-and-tricks">Tips and Tricks</a></li>

</ul>

</li>

<li class="toctree-l1"><a class="reference internal" href="../changelog.html">Change Log</a><ul>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-7">Version 7</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-6">Version 6</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-5">Version 5</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-4">Version 4</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-3">Version 3</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-2">Version 2</a></li>

<li class="toctree-l2"><a class="reference internal" href="../changelog.html#version-1">Version 1</a></li>

</ul>

</li>

</ul>

</div>

</nav>

<a href="../index.html">wonambi</a>

</nav>

<li class="breadcrumb-item"><a href="index.html">Graphical User Interface</a></li>

<li class="breadcrumb-item active">Detection methods</li>

</li>

</ul>

<hr/>

</div>

<h1>Detection methods<a class="headerlink" href="#detection-methods" title="Permalink to this heading"></a></h1>

The method options in the spindle and slow wave detection dialogs are direct implementations of detection methods reported in the scholarly articles cited below.

These implementations are based on original author code when available.

Original values for the customizable parameters are provided in square brackets [].

Further details on the original methods are provided in italics.

Note

Italicized steps are not automatically implemented in Wonambi’s algorithms.

They are included only for your information.

</div>

<h2>Spindle detection methods<a class="headerlink" href="#spindle-detection-methods" title="Permalink to this heading"></a></h2>

Ferrarelli2007 - Ferrarelli, F. et al. (2007) Am. J. Psychiatry 164, 483-92

<li>Detection is limited to all NREM sleep signal.</li>

<li>Signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> with a zero-phase equiripple Chebyshev FIR filter. Authors used a slightly different and less stable Chebyshev Type II IIR filter. The FIR filter is a more stable approximation. With <code class="docutils literal notranslate">Lowcut</code> at 11 Hz, <code class="docutils literal notranslate">Highcut</code> at 15 Hz and <code class="docutils literal notranslate">Roll-off</code> at 0.9 Hz, the attenuation is -3 dB at 10.7 Hz and 15.3 Hz.</li>

<li>The filtered signal is rectified.</li>

<li>A signal envelope is created from the oscillatory peaks in the rectified signal.</li>

<li>The detection threshold is set to the mean of the signal envelope x <code class="docutils literal notranslate">Detection threshold</code> [8].</li>

<li>For the selection threshold, the signal envelope amplitude values are distributed in a 120-bin histogram, and the amplitude of the highest-count bin x <code class="docutils literal notranslate">Selection threshold</code> [2] yields the selection threshold.</li>

<li>Spindles are detected where the signal envelope exceeds the detection threshold, with start and end times where the envelope dips below the selection threshold, before and after the detected peak.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other (or overlapping).</li>

<li>Spindles within <code class="docutils literal notranslate">Min. duration</code> and <code class="docutils literal notranslate">Max. duration</code> are retained.</li>

</ol>

Nir2011 - Nir, Y. et al. (2011) Neuron 70, 153-69

<li>The channels with spindle activity in NREM sleep are chosen for further analysis (significant spectral power increases in spindle range as compared with a 1/f model, p ‹ 0.001, paired t-test across 10 s segments.)</li>

<li>The EEG signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> Hz and <code class="docutils literal notranslate">Highcut</code> Hz with a zero-phase 4th (2x2) order Butterworth filter. Authors specify -3 dB attenuation at 9.2 Hz and 16.8 Hz. To achieve this with a 4th order filter, <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> must be set to 9.2 Hz and 16.8 Hz, respectively.</li>

<li>Instantaneous amplitude in the sigma frequency is extracted via the Hilbert transform.</li>

<li>To avoid excessive multiple crossings of thresholds within the same spindle event, instantaneous amplitude is temporally smoothed using a Gaussian kernel of σ = <code class="docutils literal notranslate">Gaussian smoothing sigma</code> [0.4] s.</li>

<li>Events with amplitude greater than mean + <code class="docutils literal notranslate">Detection threshold</code> [3] SD (computed across all artifact-free NREM sleep epochs) are considered putative spindles and detections within <code class="docutils literal notranslate">Min. interval</code> [1] s of each other are merged.</li>

<li>A threshold of mean + <code class="docutils literal notranslate">Selection threshold</code> [1] SD defines start and end times, and events with duration between <code class="docutils literal notranslate">Min. duration</code> [0.5] s and <code class="docutils literal notranslate">Max. duration</code> [2] s are selected for further analysis.</li>

<li>Those channels, in which an increase in spectral power within the detected events was restricted to the spindle-frequency range (10-16 Hz) rather than broadband (unpaired t-test (α=0.001) between maximal spectral power in detected vs. random events), and with at least 1 spindle per min of NREM sleep were chosen for further analysis. This highly conservative procedure of including in the analysis only the channels with high spindle SNR, ensured that local occurrence of spindle events does not arise merely as a result of the lack of spindles or poor spindle SNR in some channels.</li>

</ol>

Mölle2011 - Mölle, M. et al. (2011) Sleep 34(10), 1411-21

<li>Detection is limited to NREM signal.</li>

<li>Signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code>, using a zero-phase equiripple FIR filter. Authors specify -3 dB attenuation at 11.3 and 15.7 Hz (Mölle et al., J Neurosci, 2002). To achieve this, <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> must be set to 12 Hz and 15 Hz, and <code class="docutils literal notranslate">Roll-off</code> to 1.7 Hz.</li>

<li>The root-mean-square of the signal is taken, with a moving window of size = <code class="docutils literal notranslate">RMS window length</code> [0.2] s.</li>

<li>The resulting RMS signal is smoothed with a moving average of window size = <code class="docutils literal notranslate">Smoothing window length</code> [0.2] s.</li>

<li>The detection threshold is set to the mean of the RMS signal + <code class="docutils literal notranslate">Detection threshold</code> [1.5] x RMS signal SD.</li>

<li>Spindles are detected as a continuous rise in the smoothed RMS signal above the detection threshold lasting between <code class="docutils literal notranslate">Min. duration</code> [0.5] s and <code class="docutils literal notranslate">Max. duration</code> [3] s. Spindle start and end times are the threshold crossings.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

</ol>

Wamsley2012 - Wamsley, E. J. et al. (2012) Biol. Psychiatry 71, 154-61

<li>Detection is limited to NREM2 signal, filtered between 0.5-35 Hz.</li>

<li>The artefact-free EEG signal is subjected to a time-frequency transformation using an 8-parameter complex Morlet wavelet with the average of <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> as the frequency, with σ = <code class="docutils literal notranslate">Wavelet sigma</code> [0.5] s and with window size = <code class="docutils literal notranslate">Wavelet window length</code> [1] s.</li>

<li>The resulting complex-valued time series is squared.</li>

<li>The imaginary component of the time-series is discarded, and the remaining real-valued time series is squared again.</li>

<li>The moving average of the real signal is calculated, using a sliding window of size = <code class="docutils literal notranslate">Smoothing window length</code> [0.1] s.</li>

<li>A spindle event is identified whenever this wavelet signal exceeds a threshold, defined as <code class="docutils literal notranslate">Detection threshold</code> [4.5] times the mean signal amplitude of all artefact-free epochs, between <code class="docutils literal notranslate">Min. duration</code> [0.4] s and <code class="docutils literal notranslate">Max. duration</code> s [no maximum]. In this implementation, threshold crossings define the spindle start and end times, but see next point for the original method.</li>

<li>The duration of each spindle was calculated as the half-height width of wavelet energy within the spindle frequency range.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

</ol>

Martin2013 - Martin, N. et al. (2013) Neurobio. Aging 34(2) 468-76

<li>Artefact-free NREM sleep epochs are retained for analysis.</li>

<li>Signal is bandpass-filtered between <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> with a zero-phase equiripple Chebyshev FIR filter. The authors used a slightly different filter, but this implementation is more stable, and with the default lowcut and highcut and a <code class="docutils literal notranslate">Roll-off</code> of 0.4 Hz, -3 dB attenuation is achieved at 11.1 Hz and 14.9 Hz.</li>

<li>The root-mean-square of the filtered signal is taken for every consecutive window of duration <code class="docutils literal notranslate">RMS window length</code> [0.25] s and of step <code class="docutils literal notranslate">RMS window step</code> [None] s.</li>

<li>The detection threshold is set at percentile <code class="docutils literal notranslate">Detection threshold</code> [95] of the RMS signal.</li>

<li>Spindles are detected when the RMS signal exceeds the detection threshold for longer than <code class="docutils literal notranslate">Min. duration</code> [0.5] s and shorter than <code class="docutils literal notranslate">Max. duration</code> [3] s.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

</ol>

Ray2015 - Ray, L. B. et al. (2015) Front. Hum. Neurosci. 9-16

<li>The artefact-free signal is bandpass filtered between 0.3 Hz and 35 Hz, using a zero-phase 8th (4x2) order Butterworth filter.</li>

<li>Complex demodulation is deployed on the data about the frequency of interest, set by the mean of <code class="docutils literal notranslate">Lowcut</code> [11] and <code class="docutils literal notranslate">Highcut</code> [16] (Hz).</li>

<li>The complex demodulated signal is lowpass filtered below 5 Hz, using a zero-phase 8th (4x2) order Butterworth filter.</li>

<li>The lowpass filtered signal is smoothed using a triangle convolution, with window size set by <code class="docutils literal notranslate">Smoothing window length</code> [2 / frequency of interest] (sec).</li>

<li>The square of the absolute value of the complex, smoothed signal is taken.</li>

<li>The resulting signal is converted into a z-score signal, with a sliding window set by <code class="docutils literal notranslate">zscore window length</code> [60] (sec).</li>

<li>Spindles are detected where the z-score signal exceeds <code class="docutils literal notranslate">Detection threshold</code> [2.33].</li>

<li>Spindle start and end times are defined where the z-score signal surrounding a detected spindle drops below <code class="docutils literal notranslate">Selection threshold</code> [0.1].</li>

<li>Spindles separated by less than <code class="docutils literal notranslate">Min. interval</code> [0.25] sec are merged. Instead of merging spindles, authors excluded spindles beginning less than 0.25 s after of a detected spindle.</li>

<li>Spindles within <code class="docutils literal notranslate">Min. duration</code> [0.49] sec and <code class="docutils literal notranslate">Max. duration</code> [None] are retained.</li>

</ol>

Lacourse2018 - Lacourse, K. et al. (2018) J Neurosci. Meth.

<li>Artefact-free NREM sleep epochs are retained for analysis.</li>

<li>Signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code> to create the sigma signal, and between 0.3 Hz and 30 Hz to create the broadband signal.</li>

<li>All transformations are applied to the windowed signal, with a window duration of <code class="docutils literal notranslate">Window length</code> [0.3] s and with a step of <code class="docutils literal notranslate">Window step</code> [0.1] s.</li>

<li>The absolute sigma power signal is taken as the window average of the squared value of each data sample. The resulting signal is log10-transformed.</li>

<li>The relative sigma power signal is taken as the ratio of sigma band power to broadband power, log10-transformed and z-score normalized with respect to a 30-s window centered on the current window. In computing the z-score, only values between the 10th and 90th percentile are used to compute the standard deviation.</li>

<li>The covariance signal is taken as the window average of the sample-wise product of the detrended sigma signal and detrended broadband signal.</li>

<li>The normalized covariance signal is taken as the covariance signal log10-transformed and z-score normalized with respect to the 30-s window centered on the current window. In computing the z-score, only values between the 10th and 90th percentile are used to compute the standard deviation.</li>

<li>The correlation signal is taken as the covariance signal divided, window-wise, by the product of the standard deviations of the sigma and broadband signals.</li>

<li>Spindles are detected when all four of these conditions are met:

<li>the absolute sigma power signal exceeds <code class="docutils literal notranslate">Absolute power threshold</code> [1.25];</li>

<li>the relative sigma power signal exceeds <code class="docutils literal notranslate">Relative sigma threshold</code> [1.6];</li>

<li>the normalized covariance signal exceeds <code class="docutils literal notranslate">Covariance threshold</code> [1.3];</li>

<li>the correlation signal exceeds <code class="docutils literal notranslate">Correlation threshold</code> [0.69].</li>

</ul>

</li>

<li>Spindle start and end times are defined where the normalized covariance and correlation signals surrounding a detected spindle drop below their respective threshold.</li>

<li>Spindles shorter than <code class="docutils literal notranslate">Min. duration</code> [0.3] s and <code class="docutils literal notranslate">Max. duration</code> [2.5] s are discarded.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

<li>A context classifier, based on the spectral makeup of the signal surrounding each detected spindle, identifies those spindles likely to be true positives. This step (not implemented) is most useful in the absence of sleep staging.</li>

</ol>

FASST - Leclerq, Y. et al. (2011) Compu. Intel. Neurosci. 1-11

<li>Signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> [11] and <code class="docutils literal notranslate">Highcut</code> [18] using a zero-phase 8th (4x2) order Butterworth filter.</li>

<li>The detection threshold is set as the <code class="docutils literal notranslate">Detection threshold</code> th percentile of the filtered signal. Authors use only N2 signal to set the threshold.</li>

<li>The filtered signal is rectified, yielding the detection signal.</li>

<li>The detection signal is smoothed with a moving average of window size = <code class="docutils literal notranslate">Smoothing window length</code> [0.1].</li>

<li>Spindles are detected as rises in the detection signal above the detection threshold, lasting between <code class="docutils literal notranslate">Min. duration</code> [0.4] and <code class="docutils literal notranslate">Max. duration</code> [1.3].</li>

<li>Detected spindles separated by less than <code class="docutils literal notranslate">Min. interval</code> [1] s are merged.</li>

<li>Spindles overlapping across channels are merged.</li>

</ol>

FASST2 - Leclerq, Y. et al. (2011) Compu. Intel. Neurosci. 1-11

This method is identical to FASST, except step 3 is replaced with the following step:

<li>The root-mean-square of the filtered signal is taken, with a moving window of size = <code class="docutils literal notranslate">RMS window length</code> [0.1] s, yielding the detection signal.</li>

</ol>

UCSD - University of California, San Diego; unpublished

<li>The raw EEG signal is subjected to a time-frequency transformation using real wavelets with frequencies from <code class="docutils literal notranslate">Lowcut</code> to <code class="docutils literal notranslate">Highcut</code> at 0.5-Hz intervals, with wavelet duration = <code class="docutils literal notranslate">Wavelet duration</code> [1] s, width = <code class="docutils literal notranslate">Wavelet width</code> s and smoothing window duration = <code class="docutils literal notranslate">Smoothing window length</code>.</li>

<li>The resulting time-frequency signals are rectified and convolved with a Tukey window of size = 0.5 s, then averaged to produce a single time-frequency signal.</li>

<li>A threshold is defined as the signal median plus <code class="docutils literal notranslate">Detection threshold</code> [2] SDs.</li>

<li>Spindles are detected at each relative maximum in the signal which exceeds the threshold.</li>

<li>Steps 1-3 are repeated on the raw signal, this time with width = 0.2 s, with Tukey window size = 0.2 s, and with the threshold set at <code class="docutils literal notranslate">Selection threshold</code> [1] SD.</li>

<li>Spindle start and end times are defined at threshold crossings.</li>

<li>Spindles are retained if their duration is between <code class="docutils literal notranslate">Min. duration</code> and <code class="docutils literal notranslate">Max. duration</code>.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

</ol>

Concordia - Concordia University, Montreal; unpublished

<li>The root-mean-square of the signal is taken, with a moving window of size = <code class="docutils literal notranslate">RMs window length</code> [0.2] s.</li>

<li>The low and high detection thresholds are set at the mean of the RMS signal + <code class="docutils literal notranslate">Detection threshold</code> [1.5] x RMS signal SD, and mean + <code class="docutils literal notranslate">Detection threshold</code> [10] x SD, respectively.</li>

<li>RMS rises between the low and high detection thresholds are considered putative spindles, and those located within <code class="docutils literal notranslate">Tolerance</code> [0.2] s are merged.</li>

<li>Spindles are merged if within <code class="docutils literal notranslate">Min. interval</code> [0] s of each other.</li>

</ol>

</section>

<h2>Slow wave detection methods<a class="headerlink" href="#slow-wave-detection-methods" title="Permalink to this heading"></a></h2>

Massimini2004 - Massimini, M. et al. (2004) J Neurosci 24(31), 6862-70

<li>256-channel EEG is re-referenced to the average of the signals from the earlobes.</li>

<li>EEG signal is locally averaged over 4 non-overlapping regions of the scalp.</li>

<li>Detection is limited to NREM signal.</li>

<li>The signal is bandpass filtered between <code class="docutils literal notranslate">Lowcut</code> and <code class="docutils literal notranslate">Highcut</code>, using a zero-phase 4th (2x2) order Butterworth filter. Wonambi’s implementation applies the filter sequentially to avoid numerical instability: first lowpass, the highpass.</li>

<li>Slow waves are detected when the following 3 criteria are met:

<li>A positive-to-negative zero crossing and a subsequent negative-to-positive zero crossing separated by <code class="docutils literal notranslate">Min. trough duration</code> [0.3] and <code class="docutils literal notranslate">Max. trough duration</code> [1.0] s.</li>

<li>A negative peak between the two zero crossings with voltage less than <code class="docutils literal notranslate">Max. trough amplitude</code> [-80] μV</li>

<li>A peak-to-peak amplitude greater than <code class="docutils literal notranslate">Min. peak-to-peak amplitude</code> [140] μV.</li>

</ul>

</li>

</ol>

Note

Not all channels will show slow waves with the same polarity (e.g. F3-M2 and M2-F3 will be inverted).

Furthermore, the typical iEEG channels will show slow waves as positive-then-negative, contrary to surface EEG electrodes.

For these reasons, it is left to the user to set the correct polarity for slow wave detection.

The default polarity is negative-then-positive.

Wonambi displays signals negative-up, so on typical EEG channels, slow waves will be up-then-down.

To switch to positive-then-negative (down-then-up), check the <code class="docutils literal notranslate">Invert detection</code> box.

</div>

AASM/Massimini2004

This is the same as Massimini2004, except with default values for slow waves as defined by the American Academy of Sleep Medicine (AASM).

</section>

</div>

<a href="detect.html" class="btn btn-neutral float-left" title="Event detection" accesskey="p" rel="prev"> Previous</a>

</div>

<hr/>

</div>

</footer>

</div>

</section>

</div>

jQuery(function () {

SphinxRtdTheme.Navigation.enable(false);

});

</script>

</body>

</html>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

FilesExpand file tree

methods.html

Latest commit

History

methods.html

File metadata and controls