<p>MEDUSA Kernel 1.4 release. Changelog: <br></p> <p><strong>Breaking changes:</strong></p> <ul> <li>Eliminated Tensorflow dependency, which has been replaced by PyTorch</li> </ul> <p><br> <strong>Major changes:</strong></p> <ul> <li>Inclusion of decoding methods for ssvep spellers</li> <li>New timeplot to visualize signals with an advanced GUI</li> <li>Inclusion of native new biosignal types</li> <li>Added new schema for settings in medusa ecosystem</li> </ul> <p><br> <strong>Minor changes:</strong></p> <ul> <li>Improvements in M/EEG channel definition</li> <li>Fixed errors in several functions</li> <li>Normalization and detrend options have been removed from transforms.fourier_spectrogram</li> </ul>

<p>MEDUSA Kernel 1.3 release. Changelog: <br></p> <p><strong>Breaking changes:</strong></p> <ul> <li>Eliminated dependency from PyQt5</li> </ul> <p><br> <strong>Major changes:</strong></p> <ul> <li>Inclusion of bitwise reconstruction methods for c-VEP paradigms</li> </ul> <p><br> <strong>Minor changes:</strong></p> <ul> <li>Improvements in timeplot</li> <li>Fixed errors to serialize exceptions in tensorflow_integration module</li> <li>Fixed errors in bci.mi_paradigms</li> </ul>

<p>MEDUSA Kernel 1.2.5 release. Changelog: <br></p> <p><strong>Breaking changes:</strong></p> <ul> <li>Parameter standard in function set_standard_montage from meeg.meeg.EEGChannelSet</li> <li>Module plots.topography_plots has been renamed to plots.head_plots</li> <li>Tensorflow integration is now a package extra. Install with this command 'pip install medusa-kernel[TF]'</li> </ul> <p><br> <strong>Major changes:</strong></p> <ul> <li>Artifact removing with ICA is now incldued</li> <li>Inclusion of a plot to show biosginals with events, conditions, etc compatible with recording app of MEDUSA Platform</li> <li>Updates in head_plots</li> <li>Improved integration of tensorflow. Tensorflow version has to be &lt;= 2.11 to avoid problems with GPU in Windows</li> </ul> <p><br> <strong>Minor changes:</strong></p> <ul> <li>Error fixing in models of bci.mi_paradigms</li> <li>Error fixing in spatial.filetering.CSP</li> <li>Minor bugs fixed in connectivity metrics</li> <li>Improved compatibility of time plot with recordings from MEDUSA Platform</li> <li>The message of TF extras error has been updated</li> <li>Improved position of the sliders of the timeplot</li> <li>Fixed bug in artifact rejection for cvep spellers</li> </ul>

<p>MEDUSA Kernel 1.1.3 release. Changelog: <br></p> <p><strong>Breaking changes:</strong></p> <ul> <li>The spectral analysis, nonlinear analysis and connectivity analysis functions (both phase and amplitude based) support as input signals with dimensions [n_epochs, n_samples, n_channels].</li> <li>Update of Motor Imagery module: Changes in the MIData class.</li> <li>Update in Neurofeedback module: Now includes pipelines for real-time functional connectivity calculation.</li> <li>Laplacian surface filter update: Improved filter design in automatic mode.</li> </ul> <p><br> <strong>Major changes:</strong></p> <ul> <li>The algorithms of the nonlinear analysis and connectivity analysis functions have been optimized and calculations are now faster.</li> </ul> <p><br> <strong>Minor changes:</strong></p> <ul> <li>Legacy package medusa.bci.remote has been eliminated</li> <li>Tensorflow versions fixed to avoid problems: tensorflow==2.11; tensorflow-probability==0.16</li> </ul>

<p>First major update of MEDUSA Kernel, a complete suite of functions for biosignal processing.</p> <p><strong>Breaking changes:</strong></p> <ul> <li>Phase connectivity improved for real time calculation</li> <li>More functions accept both input shapes [epochs x time x channels] and [time x channels]</li> <li>Improved documentation</li> </ul>

<p>First public release of MEDUSA Kernel, a complete suite of functions for biosignal processing. Available in the Python Package Index (PyPI) repository.</p>