Release v3.1.3

Release Date: November 3, 2026

A patch release: simulations save less and finish sooner, the envelope maths is exact, and analyses report how focal the stimulation is as well as the modulation.

Update: install the new desktop app; it downloads the matching image (idossha/ti-toolbox:v3.1.3) on its next start. Update both together, since the image carries the server the app talks to. If you run the image yourself, pull it again: docker pull --platform linux/amd64 idossha/ti-toolbox:v3.1.3.

What changed

  • Leaner, faster simulations. A simulation saves TI_max, hf_peak and hf_rms by default, in one field mesh and one cortical surface mesh, and writes NIfTI volumes only when you tick Write NIfTI volumes. A later step that needs a field or a NIfTI that was not saved computes it from the simulation and saves it.
  • hf_rms replaces hf_sar. The carrier field is saved as its RMS over time, in V/m.
  • Exact envelope maths. mTI_max, mEx-search scores, 2-channel TI_avg and the tangential flex-search goal are now computed exactly (see below for what to re-run).
  • Stimulation focality beside modulation focality. Analyses and Ex/mEx-search results also report how focal the carrier field is, and the TI selectivity (above 1, the envelope is more focal than its carriers). Existing statistics and rankings are unchanged.
  • Electrodes. New jobs put electrodes on 1 mm of gel instead of 4 mm, as SimNIBS does for flex-search electrodes. You choose the electrode before generating a leadfield, and the leadfield records it.
  • Channels and the 3-D view. Channel k is electrodes Ek+ and Ek- everywhere. Pick a new montage’s electrodes by clicking them on the head; electrodes stand on the scalp, and Actual size shows them at their real size on their gel.
  • Your own Simulator defaults in Settings ▸ Simulator, used by every new Simulator job.
  • Multipolar TI takes at most 12 channels (24 electrodes).
  • Many bugs were squashed, across group statistics, the analyzer, EEG forward models, the optimizer, job scheduling, the desktop app and the scripting API.

Do I need to re-run anything?

  • mTI simulations and mEx-searches: mTI_max and mEx scores were slightly low in places (the grey-matter mean moves by under 0.1 %). Re-run them if exact element values matter.
  • 2-channel TI_avg: element values move by up to 4.6 % (the grey-matter mean by under 0.5 %).
  • Flex-searches with the tangential goal (dir_TI_tangential) optimised the wrong quantity: re-run them.
  • Group comparisons and EEG forward models can give different results: re-run them.
  • Cluster permutations: a cluster’s p-value could come out one permutation count too small; re-run one whose p sits just below alpha.
  • Visual Exporter: re-export TI_normal.ply; its arrows are now the simulation’s TI_normal.
  • Leadfields generated before this release used SimNIBS’s own electrode rather than your Simulator’s (they show “electrode not recorded”); generate them again to match.

TI_max, TI_normal and hf_peak (up to 8 channels) are unchanged; rescale nothing. A new simulation that relies on the default gel differs from one run before; set the gel to 4 mm to reproduce it.

Scripts: tit.calc is now envelope, envelope_max, envelope_mean, hf_peak and hf_power (get_TI_vectors, get_TI_avg, get_TI_dir, psi and tit.fields are gone); read hf_rms instead of hf_sar (hf_sar = 2 × hf_rms²); and drop map_to_surf, tissues_in_niftis and open_in_gmsh from SimulationConfig.

The full list is in the Changelog.

Desktop App (v3.1.3): macOS Intel · macOS Apple Silicon · Windows · Linux AppImage · Linux deb

Other:

  • Docker Image: docker pull --platform linux/amd64 idossha/ti-toolbox:v3.1.3
  • Source Code: GitHub Repository

For installation instructions, see the Installation Guide.

Getting Help

See Troubleshooting for known issues, or GitHub Discussions for help.