# Population Analysis PCA Process and Export

**URL:** <https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006>\
**Category:** Support\
**Created:** [October 4, 2021, 10:47pm UTC](https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006 "2021-10-04T22:47:26Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![mhertz](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/mhertz/32/12678_2.png) [@mhertz](https://discourse.slicer.org/u/mhertz)\
**Post date:** [October 4, 2021, 10:47pm UTC](https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006/1 "2021-10-04T22:47:26Z")

</div>

Hello everyone, I have been having some trouble using the Population Analysis function in SlicerSALT. I have been following [this](https://docs.google.com/presentation/d/1CfFPKFMA3KtlOJY5hpdZ_HACavoSIsVxpnThNIuYhZE/present?slide=id.p1) tutorial, but am stuck at Step 3, generating the PCA model. I’ve uploaded the VTK files and generated the CSV describing the groups, but when I go to select the CSV under “selection of groups” in the PCA section and press “Process and Export”, nothing happens. Any help in resolving this issue would be much appreciated.

I’m primarily using this tool to superimpose my 3D data to create a mean shape model for morphometric analysis. If anyone knows another way to create a mean shape model from multiple surface scans that alternative would also be great! Thanks very much.

---

<div class="post-metadata">

**Author:** ![muratmaga](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/muratmaga/32/3622_2.png) [@muratmaga](https://discourse.slicer.org/u/muratmaga)\
**Post date:** [October 4, 2021, 10:51pm UTC](https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006/2 "2021-10-04T22:51:33Z")

</div>

If you have fixed/anatomical landmarks, you can create a mean shape model in SlicerMorph. ([GitHub - SlicerMorph/SlicerMorph: Extensions to conduct 3D morphometrics in Slicer](https://github.com/SlicerMorph/SlicerMorph))

---

<div class="post-metadata">

**Author:** ![mhertz](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/mhertz/32/12678_2.png) [@mhertz](https://discourse.slicer.org/u/mhertz)\
**Post date:** [October 4, 2021, 11:01pm UTC](https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006/3 "2021-10-04T23:01:21Z")

</div>

Hi, thanks @muratmaga for the idea. My issue is that I’m trying to do morphometrics on 3d surface scans of humans, focusing on the arms and legs, where there are not established and well-defined anatomical landmarks. I found a few interesting articles about landmarkless approaches of analysis, but the first step was creating a prototype (mean/average) model to then sequentially compare the dataset to.

These are what I’ve been drawing inspiration from:

> **[A Landmark-Free Method for Three-Dimensional Shape Analysis](https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0150368)**
>
> Background The tools and techniques used in morphometrics have always aimed to transform the physical shape of an object into a concise set of numerical data for mathematical analysis. The advent of landmark-based morphometrics opened new avenues of...

> **[Joint registration and averaging of multiple 3D anatomical surface models](https://www.sciencedirect.com/science/article/abs/pii/S1077314205000858)**
>
> In biological imaging, three-dimensional (3D) reconstruction from serial sections is a powerful technique that produces graphical surface models of an…

Would you happen to know anything about that?

---

<div class="post-metadata">

**Author:** ![muratmaga](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/muratmaga/32/3622_2.png) [@muratmaga](https://discourse.slicer.org/u/muratmaga)\
**Post date:** [October 4, 2021, 11:20pm UTC](https://discourse.slicer.org/t/population-analysis-pca-process-and-export/20006/4 "2021-10-04T23:20:07Z")

</div>

There is a method we are working on, which as a by-product can facilitate generation of such population averages. But that’s unfortunately couple months away.

Meanwhile, you can take one representative sample, run it through PseudoLMGenerator in SlicerMorph, and then use ALPACA to transfer points to other samples. If the resultant transfer is sufficiently good, you can take those points and run through GPA module to obtain a mean shape model.

This is all available through SLicerMorph.