# Statistics on Object Differences User Tutorial

Authors: Kedar Madi, Beatriz Paniagua, Jared Vicory

Collaborators:

![](img/SlicerSALT-DifferenceStatistics-Tutorial_0.jpg)

![](img/SlicerSALT-SPHARM-PDM-Tutorial_2.png)

This module allows for the computation of correlation between how shapes change between two time points and various covariates. It does this by first computing point-to-point changes between the two given timepoints, then automatically running the [Covariate Significance Testing](CovarianceTesting.md) module on the results. Computed p-values at each point on the mesh are mapped onto a given template for visualization.

## Installation on SlicerSALT

Download the SlicerSALT packages for your respective operating system from the   _[SlicerSALT website ](http://salt.slicer.org)_  and install it\.   The module   will be ready to use then\. 

![](img/SlicerSALT-DifferenceStatistics-Tutorial_2.png)

![](img/SlicerSALT-DifferenceStatistics-Tutorial_3.png)

## Difference Statistics Module

The module can be found in the SlicerSALT menu under `Shape Analysis` -> `Statistics on Object Differences`

![](img/SlicerSALT-DifferenceStatistics-Tutorial_4.png)

### Inputs

There are three required inputs:

![](img/SlicerSALT-DifferenceStatistics-Tutorial_5.png)


- Input CSV file
  - A CSV file where:
    - The first two columns, `Timepoint 1` and `Timepoint 2`, are paths to the pairs of meshes between which differences will be computed. All meshes must have the same number of points and be in correspondence (i.e., point 1 in each mesh should correspond to roughly the same anatomical location)
    - All following columns describe covariates which will be compared with the computed differences to determine if there is correlation. These covariates can be continuous or categorical.
  - An example is shown below
- Template Mesh
  - This is the mesh onto which p-values will be mapped for visualization purposes
  - This is typically a sample object from the population and should have the same number of points/correspondence as the input files (and can be one of the input files)
  - The choice of template will not effect computation
- Output Directory
  - The directory where output files will be stored. Outputs will include
    - Computed pairwise differences between timepoints
    - The various outputs from the [Covariate Significance Testing](CovarianceTesting.md) module

Example Input CSV file:

![](img/SlicerSALT-DifferenceStatistics-Tutorial_6.png)

### Running the module and Outputs

Once the inputs are set, click the `Apply` button to begin computation. First, the meshes will be loaded and differences will be computed. Depending on the number of input files, this could take several minutes.

![](img/SlicerSALT-DifferenceStatistics-Tutorial_7.png)

Then, SlicerSALT will switch to the [Covariate Significance Testing](CovarianceTesting.md) module, with the inputs set automatically, to begin computing correlations.

![](img/SlicerSALT-DifferenceStatistics-Tutorial_8.png)

When this is finished, SlicerSALT will change to the Shape Population Viewer module which can be used to view the p-value maps for each input covariate. The data used to color the mesh can be set under `ColorMap and Vectors` -> `Attribute` on the bottom tab.

![](img/SlicerSALT-DifferenceStatistics-Tutorial_9.png)

The file with the mapped p-values is saved to out.vtk in the specified output folder.

## Acknowledgements \- Resources \- Questions



<ul>
  <li>The method described here was developed by Kedar Madi during his dissertation at Wake Forest with guidance from Beatriz Paniagua. It was adapted for SlicerSALT by Jared Vicory.</li>
  <li>Github repositories:</li>
      <ul>
            <li><a href="https://github.com/slicersalt/DifferenceStatistics">DifferenceStatistics</a></li>
            <li><a href="https://salt.slicer.org">SlicerSALT</a></li>
            <li><a href="https://github.com/Slicer/Slicer">3D Slicer</a></li>
      </ul>
  <li>Forums:</li>
      <ul>
            <li><a href="https://discourse.slicer.org/t/about-the-slicersalt-category/47">SlicerSALT</a></li>
            <li><a href="https://discourse.slicer.org/">3D Slicer</a></li>
      </ul>
  <li>For other remarks or questions, please email: beatriz.paniagua@kitware.com</li>
</ul>