MultiSim Services and Resources

WP- 9 Hypermodelling Integration and Framework Development.

These are all the data, service, and software resources that are made available to the researchers involved in the MultiSim project.  As it is now it constitutes a working plan.  As the resources will be made available, the actual on-line access will be linked to the term, and the resource will become accessible.  Thus, all linked resources are available now.

Data Repository

All the data generated in relation of the MultiSim project should be stored in digital format according to a predefined schema.  Each data object is stored in a single file, identified by a Unique Resource Identifier (URI), and annotated by a minimum set of metadata.

Each data object belongs to one or more Sets, collections of data, that are also identified by a URI and annotated with metadata.  For now sets will be defined with directories; to avoid data replication, soft links will be used if a data object belongs to multiple sets. A set can contain only data objects (files) or other sets (directories).

The results of Search functions on the metadata can be saved as sets; this makes easy to create a set which include all data objects of a certain type, for example to create a batch submission for a service that re-process with a new service all data objects available.

In term of usage, the data repository will be formed by two remote file systems that can be mounted on your PC, and a web interface.  The first file system will host one sandbox directory for each user.  you will be able to copy, rename, delete, any file inside it.

For now the data are organised according to scales: it is responsibility of the WP leaders to agree which data are stored and how they are organised.

Data repository for Multiscale simulations

  • General clinical data
  • Organism-organ coregistration data
  • Organ-tissue coregistration data
  • Hypermodel inputs
  • Risk of bone fracture outputs

Data repository for Organism scale

  • Motion Capture data
  • Medical imaging data
  • Activity monitoring data
  • Musculoskeletal Dynamics model inputs
  • Musculoskeletal Dynamics outputs
  • Musculo-articular force spectra outputs

Data repository for Organ scale

  • Medical imaging data
  • Segmentation data
  • Mineral Density data
  • FE model inputs
  • FE results outputs
  • Bone strength outputs
  • Bone strain outputs

Data repository for Tissue scale

  • Biopsies MicroCT data
  • HRpQCT data
  • Cropped data
  • Labelling data (binary)
  • Staged compression data
  • DVC model inputs
  • DVC results outputs
  • MicroFE model inputs
  • MicroFE model outputs
  • Elastic tensor outputs
  • Mechanical stimuli outputs
  • Bone remodelling model inputs
  • Bone remodelling outputs

Data repository for Cell scale

  • Clinical biomarkers data
  • RNA data
  • Mouse MicroCT data
  • Mouse registration model inputs
  • Mouse bone remodelling outputs
  • Mouse microFE model inputs
  • Mouse bone strength outputs
  • Bone remodelling model inputs
  • Bone remodelling outputs
  • Cell model inputs
  • Cell model outputs

Services Provided

A number of data processing and simulation services are provided to MultiSIM researchers. The services are invoked by submitting to them the URI of a valid input set.  If the service complete its execution, an output set will be automatically stored in the data repository, annotated with provenance metadata.

By creating a set which contains a list of valid inputs, and submitting it as input to the service, the service will run many times, one for each input in the set.

Services are also organised by scale:

Multiscale simulations

  • Co-registration service: the service output a co-registration pose matrix that spatially transform one set into another.  If the two input sets are paired landmark clouds, a Single Value Decomposition algorithm is used.  If the input sets are a landmark cloud and a polygonal surface, then an Interactive Closest Point algorithm is used.
  • Actual risk of fracture: organism-organ hypermodel that computes the probabilistic loading spectrum with an organism scale model, and then run the organ-scale model within a stochastic execution scheme to determine the actual risk of bone fracture.
  • Absolute Risk of Fracture: same as above, but also include a bone remodelling loop over time, which provides an actual risk of fracture for each year over five or ten years.

Organism scale services

  • Inverse kinematics: computes the kinematics of a kinematic chain (skeleton) that best fits the marker trajectories during motion capture.
  • Inverse dynamics: computes joint torques required to produce the measured kinematics.
  • Muscle dynamics: computes the forces that each muscle must exert in order to produce the joint torques required to provide the produce the measured kinematics.
  • Loading spectra predictions: using activity monitoring or other sources of frequency information, compute the probabilistic loading spectrum for the skeletal region of interest.

Organ scale services

  • Clinical CT calibration: transform CT images so that voxel values express HA-equivalent mineral density, or directly Elastic moduli, as defined by calibration images.
  • Automatic Meshing: using ShIRT elastic registration morph a template FE mesh on the CT 3D image.
  • Bonemat: project the morphed mesh onto the calibrated CT images, and compute the average elastic modulus to be assigned to each finite element.
  • Bone strength: takes in input a valid FE model of a bone, and an array of load cases; return predictions on strain distribution, an on global strength.

Tissue scale services

  • Crop-transfer: launch a script on the remote microCT machine that crop the selected calibrated dataset from a proprietary format to DICOM, crop it, and transfer to the data repository, creating a valid data object fully annotated.
  • Density processing service: generate binary masks of a microCT dataset based on density ranges.  Computes also frequency histograms, and calculates Bone Mineral Content (BMC) over a defined set of bounding boxes.
  • DVC analysis: this service takes in input a set of microCT dataset generated by a staged compression experiment, and return a displacement and strain fields, computed using Digital Volume Correlation.
  • MicroFE generation: this service takes a calibrated microCT dataset and output a valid FE model, generated using homogeneous or heterogeneous modelling approaches.
  • MicroFE solver: this service runs the microFE model, applying defined boundary conditions, and output the displacement, strain, stress, and stimuli fields.
  • Elastic tensor: this service takes a microFE model, and computes the elastic tensor of the bone tissue volume its represents.
  • Bone remodelling: this service takes in input a microCT dataset, the microFE model derived from it, and predict how the tissue morphology will change over time.  It outputs a series of synthetic microCT, and the relative microFE models.

Cell scale services

  • RNA analysis: wrap EBI web services required for RNA analysis.
  • Remodelling quantification: the service accept two or more in vivo microCT taken on the same mouse at different time points, and computes how the bone tissue has changed over time because of growth and adaptation.
  • Mouse tibia strength: this service takes as input an in vivo microCT of a mouse tibia, generate a homogeneous microFE, applies predefined boundary conditions, and computes the strength of the mouse tibia under physiological loading.
  • Mouse bone remodelling: this service predict the changes in bone mineral content in various regions of the mouse tibia.
  • Cell model: this service takes in input the calculated strain or fluid flow at the tissue level and calculates the load transfer within the cytoskeleton. It outputs the stresses and strains within the microtubules, actin filaments and nucleus to determine the change of gene expression and therefore the cell activity at the tissue level.

Software Tools

  • NMS Builder: this is a pre-processing software tool to generate musculoskeletal dynamics models. The application is based on the MAF2 framework, and it is maintained by Fulvia Taddei group at the Rizzoli institute in Bologna.
    • Application (32 bits – Windows only)
    • Documentation & Test Data
  • Builder M2O: Builder Mark II Organ (M2O) is a pre-processing software tool to generate organ-scale finite element models. The application is based on the MAF2 framework, and it is maintained by Kewei Duan, Insigneo CS&E.
    • Application (32 bits – Windows only)
    • Documentation & Test data
  • Builder T2O: Builder Mark II Tissue (M2T) is a pre-processing software tool to generate tissue-scale finite element models. The application is based on the MAF2 framework, and it is maintained by Kewei Duan, Insigneo CS&E.
    • Application (64 bits – Windows only)
    • Test data
    • Documentation
  • VPH-HF: the MultiSIM execution services framework is managed with the VPH Hypermodelling Framework (VPH-HF) an open source software co-developed by Insigneo CS&E and the CINECA Supercomputing Centre in Italy.  MultiSIM users should have no need to access the installation of this software, but for completeness, we make it available here in the version modified for MultiSIM.