Sim4Life V9.2 Release Notes

Sim4Life V9.2 generates solver-ready anatomical models directly from MRI/CT scans with third-generation AI tools, adds a broadband skin model for absorbed power density from 10 to 110 GHz, and introduces a DBS/sEEG electrode generator.

Items apply to both editions unless tagged (Desktop) or (Web). The web edition was released first, on 7 October 2025. Desktop hardware and OS: see System Requirements.

Known Issues (V9.2)

  • Instability for some rare dispersive material configurations.
  • Job priorities may not correspond to the job execution order (fixed in V9.6).
  • Potentially decreased accuracy in rare cases for the sensor combiner of the far-field sensor.

V9.2 · 9 December 2025

Desktop (text version) · Web V9.2, Platform V1.86.0, 7 October 2025 (platform notes)

Added

Modeling
  • Third-generation deep learning model labels head, neck, and torso tissues from MRI/CT scans and generates solver-ready models from raw images, with about 50% less setup time and user interaction than before.
  • DBS and sEEG electrode generator: parameterized electrodes with adjustable diameter, contact length, spacing, arc angle, segmentation, and tip offset, interactively or from Python.
  • (Desktop) Angle measurement tool, and a Sample Surface tool that generates points on a meshed surface.
  • (Desktop) The 10-10 system tool includes the outermost ring of the specification, and surfaces generated during segmentation get material tags automatically.
  • (Desktop) Mesh smoothing in the Octree Mesher, and a preferred mouse-wheel zoom direction in the 3D view.

Simulation
  • Broadband Skin Power Absorption Model, implementing Christ et al. (2025) as adopted by IEC/IEEE standards, for absorbed power density from 10 to 110 GHz.
  • (Desktop) T-increase thermal transient simulations, focused on temperature rise rather than absolute temperature.
  • (Web) OpenFOAM plugin: run fluid and small-strain mechanics solvers from the Plugin Manager, with all dictionaries, logs, and results stored in the Sim4Life project.

Optimization(Web)
  • Upgraded optimizer combining surrogate modeling with a multi-objective genetic algorithm (MOGA).
  • Pareto front visualization of trade-offs between objectives such as safety, efficacy, and energy efficiency. In a spinal cord stimulation study, it found pulse shapes with the same neural recruitment at up to 5x lower energy.

Analysis
  • (Desktop) Entity groups that are not part of a simulation can be dragged into the masking filter.

Virtual Population
  • (Desktop) Upgraded ViP models (Duke, Ella, Fats, Billie, Thelonious, and Yoon-sun) deform more realistically at the joints in extreme postures and have additional spinal joints for neck and back bending.

Help Center(Web)
  • Integrated Help Center connected to Application Support, with opt-in sharing of logs and screenshots.

Fixed

  • (Desktop) Sim4Life connects to the material database when the Windows user name contains non-ASCII characters.
  • (Desktop) The 3D scaling bar no longer overlaps modeling-tool hints.
  • (Desktop) Closed coils are no longer flagged as open in low-frequency magneto vector potential simulations.
  • (Desktop) Changing the MIP Viewer slice index moves the projection slice.
  • (Desktop) Label field and image properties no longer disappear from the Controller.
  • (Desktop) Faster spline and edge rendering, and no more crashes when orbiting the 3D view repeatedly.
  • (Desktop) The centerline tool accepts mesh points as source or target.
  • (Desktop) Grids are recalculated after clearing results and resetting voxels of a low-frequency magneto-quasi-static simulation.
  • (Desktop) Planar cuts of static ViP models update grid and bounding box dimensions, and all entities of a group stretch correctly.
  • (Desktop) Transformations of imported HOC neurons are applied correctly when interpolating potential fields.
  • (Desktop) GPU-accelerated CEM43 calculations are accurate, and CEM43 values are no longer overwritten when Arrhenius recording is also enabled.

Note

  • (Desktop) V9.2 is the last version that supports NVIDIA Maxwell GPUs with the Acceleware libraries.