The highlight of this Sim4Life release for our power users is Optimizer V1, which includes a new generation of multi-parameter, multi-goal optimization frameworks. It provides an entire tool box ranging from stochastic response surface methods to Evolutionary Algorithms (e.g., Genetic Algorithms), harnessing, e.g., the power of Sandia National Lab’s Dakota library in combination with Sim4Life’s high performance computing (HPC) support. Optimizer V1 is open and flexible, and supports the automation of many tasks, including device optimization, virtual prototyping of treatments, and safety evaluations, enabling the user to connect any combination of objectives and constraints (predefined or expression-based) for a given set of design variables with assistance from the intuitive interface.
We have also incorporated into Sim4Life V4.0 a number of enhancements requested by our valuable users to improve the design speed and to manage the growing complexity of virtual prototyping. Highlighted examples include the parallelization/HPC support for coupled electromagnetic (EM)-neuron dynamics modeling, efficient anatomical model generation, modification and adaptation functionality for application-driven computational phantoms, and a new interface for the intuitive management of circuits/network simulation.
In addition to V4.0's novel features, there are several improvements as well as a number of bug fixes. Your feedback, as always, is greatly welcome!
Multi-Parameter Design Optimization
NOVEL MULTI-PARAMETER MULTI-GOAL OPTIMIZER
Sim4Life V4.0 incorporates a novel, effective, and easy-to use optimization framework, forming a dedicated suite/toolkit for optimization and/or tuning of virtual prototypes. The automated process, empowered by state-of-the-art technologies, thus enables engineers to reach design goals with minimal time and effort.
Focusing on fast convergence and robustness, the optimization engine ‒ harnessing among other schemes the powerful DAKOTA package from Sandia National Lab ‒ includes global optimization methods, such as stochastic response surface methods and Evolutionary Algorithms (e.g., Genetic Algorithms), for real-world engineering applications.
The guided mode in the new user interface for Optimizer V1 allows for an easy optimization setup with predefined goals. For users with more experience in Sim4Life who prefer unlimited flexibility, the expert mode is the method of choice as it allows for full customization of the setup through pipelines in the workbench.
Multiple objectives and constraints (reflection coefficient, averaged specific absorption rate (SAR), radiation efficiency, etc.) can be provided by the user as predefined goals or mathematical expressions.
Optimizer V1 allows modelers to specify a multitude of design variables including geometry, material parameters, lumped elements, simulation settings and postprocessing/analysis parameters which can be expressed as arbitrary mathematical formulae.
Customizable plots provide the user with status information (parameters, fitness, convergence, etc.) at any time during the optimization process.
Sim4Life V4.0 combines the superior speed of HPC enabled solvers with our powerful new optimizer/sweeper to perform effective optimization of entire computer aided design (CAD) derived devices, embedded within complex environments.
Faster & More Versatile Neuronal Dynamics Simulation
EM NEURON INTERACTION/ STIMULATION
Sim4Life V4.0 introduces process parallelization through threading for coupled EM-neuronal dynamics simulations to considerably cut down on the large computation time required for modeling the electrophysiological response of a large number of neuronal cells and fibers (e.g., within complex neuro-functionalized anatomical human models).
To enhance HPC support and computation performance for users, V4.0 now allows the submission of EM-neuronal coupled simulations for processing on remote computing hardware.
In order to investigate neuro-stimulation by means of multiple, complex electric fields (e.g., generated by multi-electrode arrays or the x/y/z MRI gradient coils), in V4.0 neuronal dynamics simulations can incorporate multiple incident EM-fields with independent modulation.
To permit neuronal model validation, Sim4Life V4.0 enables to run neuronal dynamics simulations also in absence of EM exposure. The precomputation of externally applied EM fields is therefore not a prerequisite anymore.
To better visualize electrophysiological quantities such as transmembrane potentials or intracellular currents, individual dynamic visualizations of neuronal data can be combined in the post-processing of V4.0.
Cloud-Based High-Performance Computing Made Easier
CLOUD COMPUTING ON NIMBIX
Cloud computing in Sim4Life enables users with extended hardware resource requirements to borrow computational power from third-party providers. We started to make Sim4Life available on Nimbix. Additional cloud computing platforms (e.g., Amazon/AWS) will be available soon.
To facilitate the use of cloud/remote resources, V4.0 now features integrated upload/download functionality of result files directly from the Nimbix cloud data partition.
A dedicated graphical user interface (GUI) provides user access to the Nimbix status console and all its functionality for real-time monitoring and controlling of running simulation tasks and costs in the cloud.
The Nimbix support includes on-the-fly calculation of optimal platform for a particular simulation hardware requirements and real-time resource availability information.
Solvers currently available in the Nimbix cloud are EM-FDTD/SW/AXE and Thermal for both single and double precision architectures.
Extended ViP & Easier, Faster Image-Based Modeling and Adaptation
Virtual Population (ViP) / COMPUTABLE PHANTOMS
New morphed versions of Ella (with BMI 22, 26, 30) and Fats (with BMI 29) extend the population coverage and the range of applications.
The modified Ella model, with high resolution segmented breast tissues in prone position and a posable head and neck (C1 and C2), is the anatomical female model for magnetic resonance imaging (MRI) breast-coil design, optimization, and compliance evaluation.
New mouse models are also available with Sim4Life V4.0.
IMAGE BASED MODELING AND ADAPTATION / SURFACE GENERATION
Boolean operations can now be used with complex surface-based anatomical models, such as the ViP v3.x, giving the user the needed degree of freedom to adapt and modify different organs for application and device specific simulations. Moreover, planar cuts and cage-based morphing can now be applied to the ViP and other license protected models.
The new surface extraction tool cuts-down time and memory requirements for large segmentation jobs such as high-resolution adult models from several days to a few hours.
The revised image registration tool allows to select image masks and combine multiple similarity metrics, for improved image-based model adaption/personalization.
A Python module was added which provides access to many useful image-based modeling functions via scripting.
The new release of iSEG together with Sim4Life V4.0 features an improved and much more robust live-link experience.
An automatic bone segmentation algorithm for faster computed tomography (CT)-image processing, a bias-correction algorithm for MR images, and the possibility to import and voxel STL-files are a few examples for new widgets improving segmentation time and quality.
Peak memory usage is substantially reduced during loading and saving of iSEG projects, making the tool more suitable for average workstations.
Fulfilling User Requests: Ease of Use & Resource Savings (Selected)
A new interactive circuit viewer/GUI in V4.0 cuts-down the time for visualization and operation of matching/tuning circuits, provided within a new intuitive interface.
The new version enables the user to save more disk space by selectively keeping or deleting voxels, raw or processed data, further allowing for an easy subsequent loading in read-only mode.
To widen flexibility and application range, the internal material database in Sim4Life V4.0 has been extended by various additional material descriptions (e.g., from industrial suppliers).
A more comprehensive list of improvements, fixes, and new features is included in the Release Notes.
To experience the power and elegance of Sim4Life V4.0 or for further information, please email us at firstname.lastname@example.org or call +41 44 245 9765.
The relevant Sim4Life installers for Win7/Win8/8.1/Win10 64-bit platforms can be downloaded here. Existing customers with an up-to-date annual maintenance and support plan will automatically receive all pertinent information regarding the Software Installer Download and updated licenses.
At ZMT we are committed to support our customers with the most innovative software solutions, testing equipment, and service.
Easy Setup of Multi-Parameter Multi-Goal Optimizations with Only a Few Clicks: The novel optimization framework allows for multi-parameter and multi-goal optimizations with real-time visualization of the results. In this example, the capacitance values of a 7T MRI birdcage coil are optimized. The reflection coefficient is minimized at 298 MHz while the SAR in the head region is maintained below a specified limit.
Robust, Effective Device Optimization in Complex Scenarios: Optimizing the performance (S11, SAR, RadEff) of a conformal printed dual band antenna for a wearable device with Optimizer V1 (using our subgridding and HPC solutions): 2D plot of the Reflection Coefficient showing extractions of several iterations (optimum highlighted). Far-field patterns are visualized, corresponding to different arm positions. Convergence plots help the user to monitor the status of the optimization study.
Faster Neuronal Dynamics Simulation, Extended Application Range: Spinal Cord Stimulation (SCS) induced by electric fields delivered via multi-contact electrodes (model of spinal cord with a multi-electrode paddle). The field generated by multiple independent electrodes determines the stimulation of fibers of different diameter, and the generation of action potentials (APs) propagating along the fiber with different conduction velocities.
Morphed versions of existing ViP anatomical models Fats (originally 120 kg, 3rd from left) and Ella (originally 57 kg, 3rd from right). Body mass index (BMI) values from left to right: 29, 36, and 49 kg/m2 (Fats) and 22, 26, and 30 kg/m2 (Ella).
Versatile Toolkit for Complex Mesh-Based Modeling in Sim4Life V4.0: Cage-based morphing (free-form deformation technique) allowing to modify the shape and size for multiple model entities: use of the new cage-based morphing tool on the heart and blood vessels of the Duke v3.1 model, creating a deformed version of the anatomy without holes or gaps between the tissues. Complex deformation fields (e.g. breathing, beating heart) can be loaded through a dedicated Python API.
Easy and Comprehensive Visualization and Operation of Matching/Tuning Circuits: In V4.0, users can - for the first time - interactively operate circuits using a new intuitive graphical circuit viewer/interface. In this example, the MATCH tool is used to attached matching, tuning, and decoupling circuit networks in the post-processing stage of multi-port simulations (parallel-transmit coil array).