Skip to content

Multiscale

Image Generation

The Multiscale Image Generation module offers an interface with various parameters for manipulating and configuring the MPSlib library. MPSlib features a set of algorithms based on multiple point statistical (MPS) models inferred from a training image. Currently, only the Generalized ENESIM algorithm with direct sampling (DS) mode is available.

Panels and their use

Figure 1
Figure 1: Multiscale Image Generation Module.

Input Data

  • Training Image: Volume that acts as a training image.

  • Hard Data Image: Volume that acts as "Hard Data", where values and points are fixed during the simulation.

  • Mask Image: Volume that acts as a mask for selecting the simulation area. Unselected segments will not be included in the simulation.

Simulation Options

  • Wrap HD into cylinder: If the "Hard Data" is a well image (2D), this option causes the image to be considered a cylinder and performs simulations as 3D.

  • Data type: Determines whether the data type is continuous or categorical. Segmentations and Labelmaps can be used for discrete and continuous simulations, but scalar volumes can only be used for continuous.

  • Categorical: Segmentations control regions and determine the class value of the Hard Data and Training Image volume. Unselected segments will be disregarded.

  • Continuous: Segmentations control which regions and volume values will be used as Hard Data or training data. Unselected segments will be disregarded.

  • Grid Resolution: Voxel resolution of the resulting image (mm).

  • Grid Dimensions: Dimensions of the resulting image.

  • Automatic resampling: Activates the automatic resizing functionality of the input data to the simulation grid's dimension and resolution. If the image is an imagelog, Y-axis resizing is disabled.

    • Spacing: New axis resolution after resizing.
    • Size: New axis dimension after resizing.
    • Ratio: Ratio of the new voxel resolution to the initial resolution.

Parameters

  • Number of Conditioning points: Number of conditioning points to be used in each iteration.

  • Number of realizations: Number of simulations and images to be generated.

  • Number of iterations: Maximum number of iterations when searching the training image.

  • Random seed: Initial value used to start the simulation. The same seed with the same parameters always generates the same result.

  • Colocate dimensions: For a 3D simulation, ensure that the order in the last dimensions is important, allowing a 2D co-simulation with conditional data in the third dimension.

  • Max search radius: Only conditional data within a defined radius is used as conditioning data.

  • Max distance (normalized): The maximum distance that will lead to the acceptance of a conditional model match. If the value is 0, a perfect match will be sought.

  • Distance power: Weights the conditioning data based on the distance from the central values. A value of 0 configures no weighting.

Output Options

  • Output prefix: Name of the generated volume or file. In case of multiple realizations, a number is added to indicate the realization.

  • Save: Options for saving the results.

  • First realization: Saves only the first realization as an individual volume.
  • All realizations: Saves all realizations as individual volumes.
  • As sequence: Saves the realizations in a sequence set. The "_proxy" output volume indicates it is a sequence and has the controllers for visualizing the realizations.

  • TIF files: Saves all realizations as tiff files.

  • Export directory: Directory where the tiff files will be saved. It is only enabled if the "TIF files" option is selected.

Buttons

  • Run: Executes the mps sequentially. The Geoslicer interface is locked during the execution of this option.
  • Run Parallel: Executes the mps in parallel. In this option, the execution runs in another thread, and the interface can be used during execution.
  • Cancel: Interrupts the simulation execution. Only when executed in parallel.

Post Processing

Module for data extraction after Multiscale simulation.

Figura 1
Figure 1: Multiscale Post Processing Module.

Methods

Porosity per Realization

Produces a table with the porosity percentage of each slice of a volume, across all volumes in a sequence.

Input Data and Parameters
  1. Result Volume: Volume for porosity calculation. If the volume is a proxy for a sequence of volumes, porosity will be calculated for all realizations.
  2. Training Image: Extra volume included in the calculations and added to the table as a reference.
  3. Pore segment Value: Value to be considered as a pore in scalar volumes (continuous data)
  4. Pore segment: Segment to be considered as a pore in Labelmaps (discrete data).


Pore Size distribution

Recalculates the distribution of pore size for frequency.

Input Data and Parameters
  1. PSD Sequence Table: Table or proxy of sequence of tables resulting from the Microtom module.
  2. PSD TI Table: Table result from microtom for the training image.