.. _changelog: Changelog =========== 4.1.0 ------ * Tetrahedral quality of the meshes was increased substantially to improve numerical accuracy of the FEM calculations and remove outliers in the calculated electric fields * Option added to use white matter and pial surfaces from FreeSurfer for more accurate representation of smaller sulci in the head meshes * I/O functions for neuronavigation data have been updated to support new Brainsight version 2.5.3 NOTES & Known issues: see 4.0.0 4.0.1 ------ * changed Brainsight position import/export to support only NIfTI:Aligned (to avoid ambiguities) * small bug fixes 4.0.0 ------ * New head segmentation and meshing pipeline *charm* with improved accuracy and robustness * Many new TMS coil models * New head models with additional tissue types, in particular spongy bone and large blood vessels * New flexible meshing approach to simplify manual editing and inclusion of custom tissue types in the head mesh * New command line tool *meshmesh* to support meshing of custom geometries * Support of Nx1 center-surround montages in python and Matlab * Added python examples for temporal interference simulations using pre-calculated leadfields for speed * Added basic I/O functions for neuronavigation data in python * Update to python 3.9 * Major code cleanup and restructuring under the hood * Tested on Windows 10, Linux and Macs with Intel and Apple Silicon * Headreco and mri2mesh are deprecated. NOTE: Simnibs 4 is NOT backwards compatible. Head models created with charm cannot be used in older versions. Head models from older versions need to be converted using the command line tool *convert_3_to_4* for use with SimNIBS 4. Known issues: * Installation fails on paths with non-standard characters, such as backslash, chinese characters, ... (workaround: provide another path) * simnibs_gui does not start on some linux systems, e.g. with wayland (workaround: ld preloading of libstdc++.so.6 seems to help; example: export LD_PRELOAD=/usr/lib/libstdc++.so.6 - path needs to be adjusted according to library path on local system) * The PARDISO solver (available as option for the FEM calculations) does not work on Apple Silicon 3.2.6 ------ * fixed a bug causing headreco to fail on MacOS * Note: On Apple M1s using .ccd coil files does not yet work, please use niftis instead. 3.2.5 ------ * added skin smoothing options for TMS optimization * more informative matlab error messages * small bug fixes 3.2.4 ------ * Small bug fix related to gmsh options. 3.2.3 ------ * Gmsh version changed to avoid issues with Big Sur (only MacOS) * Headreco bug fixes to make meshing more stable in the eye region and air cavities * Added a Nx1 example for a center-surround electrode montage 3.2.2 ------ * Added matlab and python examples for TACSchallenge * Added matlab examples to calculate TI amplitudes * TMS optimization now prints optimal position in log file * Added reading of leadfields in mesh_load_hdr5.m * Added work-around to enable installation on Mac OS 11 Big Sur; NOTE: If you attempted to install a previous version of SimNIBS 3.2, you have to wipe the installation folder (in "/Users/username/Applications/SimNIBS-3.2" before attempting to install again; NOTE: This work-around is temporary and will be removed once the issues in the underlying python packages have been resolved * Changed the skin smoothing iterations in headreco from 5 to 20, which should result in a smoother skin surface. Note: this changes the standard behavior of headreco, so results might differ slightly. * Per default, headreco does not print the cat summary pdf anymore. The --cat_no_print flag was removed; Instead, use --cat_print now in case you need the summary. Bug fixes: * TDCS Network Optimization: Fixed bugs to accept images with NaNs, binary images, and images of size NxMxKx1; weights of eyes are now set to 0. * Added buffered read for gmsh v2 files in python to resolve speed issue on clusters * Resolved a bug causing some points of the individual middle gm surface to be falsely interpreted as outside gm when interpolating results to individual gm surface * Further small fixes across the code * Fixed electrode meshing that sometimes caused some parts of the electrodes to be detached. Known issues: * mri2mesh does not work with Freesurfer 7; please use Freesurfer 6 for now * SimNIBS is so far not tested on Macs with Apple Silicon, and is likely to give errors on those machines 3.2 ---- * Added Auxiliary Dipole Method (ADM) TMS optimization (contributed by Luis Gomez) * Added TES magnetic field calculations for MRCDI/MREIT (contributed by Hassan Yazdanian) * Added TES optimization with field strength * Added TES optimization for brain network targeting * FMM-based coil A field calculations from :file:`.ccd` files * Refactoring of optimization code * Calculating coil-cortex distances during TMS simulations * New installers 3.1 ---- * Added TDCS and TMS Optimization * Added support to the MKL PARDISO solver * Minor bug fixes 3.0 ----- * Major update to SimNIBS * New integrated solver based on PETSc and Hypre with huge speed ups! * New installation procedure * Changed headreco call (:code:`-d no-conform` and CAT12 now standard) * New coil models * Moved to Python 3.7 * Updated documentation * Uncertainty quantification support * Improved results visualizations * SimNIBS is now installable as a python package 2.1.2 --------- * This upgrade focused on the MATLAB library for SimNIBS * The MATLAB library can be found in in the *matlab/functions/* folder in the simnibs directory * The MATLAB examples have been expanded to show the new features * We also made a few bug fixes to *get_fields_at_coordinates* and the graphical user interface 2.1.1 --------- * This upgrade focused on usability of SimNIBS * Simnibs is now shipped with spm12 and cat12. Installing those separately is no longer necessary to run **heareco**. When you have Matlab installed, you're ready to go after the SimNIBS installation. * Added features for automatically calculate eeg positions and easily using them to set-up simulations in the GUI. Added a script called **get_eeg_positions** in order to use the new features on head models ran with SimNIBS 2.1.0 * Improved Matlab and python scripting. Examples can be fount in $SIMNIBSDIR/matlab/examples and $SIMNIBSDIR/python_examples * Changed the sign of the normals when interpolating to the cortical surface * the SimNIBS python installation no longer requires scikit-image 2.1.0 --------- * 2.1.0 is a major update of SimNIBS 2 * New head segmentation script **headreco** * New post-processing options to transform fields to NIfTI volumes, MNI space, FreeSurfer overlays and FsAverage space * New scripts to calculate EEG 10-10 positions * New MATLAB library, including example scripts e.g. to set up simulations for ring electrodes * New example data sets, including an extended MNI template * Major refactoring under the hood, for a cleaner experience and quicker future updates * **Head segmentations and simulation files created with SimNIBS 2.0 are incompatible with SimNIBS 2.1** 2.0.1g --------- * More fixes to the GUI * Changes address to Miniconda during installation procedure * Changed bug in simnibs.py where it would look for files that didn't exist 2.0.1f --------- * Changed 3dcalc wrapper in Linux * Fixed bugs in the GUI related to the PySide->PyQt changes 2.0.1e --------- * Changed from PySide to PyQt * Changed getopt in osx to a wrapper script which will call getopt_o adjusting DYLD_LIBRARY_PATH * Changed scalp color on GUI 2.0.1d --------- * Fixed the intallation of qt on mac * Fixed an installation bug that occurred when reinstalling simnibs without starting a new terminal window 2.0.1c --------- * freeglut is no longer required * now a local verision of libXp, libXpm and libXmu is provided. This should make the installation easier * solved bug in the anisotropic conductivity calculations that would happen if there was a "." in the path 2.0.1b --------- * Now SimNIBS gui gives out a warning if there are any spaces in the file path * The installation procedure will now install freeglut on Linux 2.0.1a --------- * support of conductivity tensors for gray and white matter added to GUI * script dwi2cond added to estimate conductiviy tensors from diffusion MRI * automatic installation procedure changed to use miniconda * bug in mri2mesh fixed which prevented it to use the T2 image to reconstruct the skull * Changed standard colors in GUI * The GUI now lets you set TMS coil distances * Removed deprecated post processing options * Added new TMS coil files * Fixed bug where the electrode thickness would change every time the electrode edition window opened * Fixed bug where every simulation would use the same conductivities * GUI now supports advanced electrode modeling * **ATTENTION** old .simnibs files maybe incompatible with the new version