Hi,

On behalf of SlicerMorph team, I wanted to point out to a new paper that 
come out of our group that extends the ALPACA pipeline by supporting 
multiple-templates. It is open-access and is available here: 
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0278035 

Single template-based LMing method like ALPACA works very well typically in 
single-species context where morphological disparity among samples is not 
high, but may be of limited utility in studies with multiple species, or 
multiple developmental time points. Multi-template ALPACA (MALPACA) 
overcomes this by using multiple templates during the registrations. There 
are couple benefits of this approach, some of which has been discussed 
recently in the forum in context of measurement errors:

1. You increase the accuracy of the final estimate of LM positions.
2. You avoid the risk of "biasing" your automated landmark outcome to your 
template choice.
3. With multiple estimates of LM positions for each sample, you can 
implement a post-hoc quality-control analysis. 

Along with MALPACA, ALPACA now has a "template selection" tab, which uses 
the point clouds derived from the 3D models and kmeans clustering to 
suggest samples from your study population to be used as templates. User 
specifies the grouping structure, as well as how many templates to be 
selected per group. (Note that any ALPACA analysis assumes that as input, 
user provides 3D models that are identical in content without missing or 
extraneous parts.) 

The method is available as part of the latest SlicerMorph extension. Paper 
provides a detailed step-by-step instruction both for template selection 
and MALPACA. An online version of this tutorial can also be found at 
https://github.com/SlicerMorph/Tutorials/tree/main/MALPACA, which will 
serve as the most up-to-date documentation. 

ALPACA proper has got a few updates as well: 

1. The default output format is now JSON as opposed to legacy FCSV (you can 
change it to FCSV using the settings under Advanced settings tab. However 
we encourage you to use JSON format for better compatibility with Slicer 
and future-proofing your datasets. You can use SlicerMorphR package to 
import JSON files into R with its read.markups.json function. 
2. Batch mode now supports replication of analysis using the same 
settings/templates, which can be useful in context of symmetry analyses. 
3. ALPACA now can optionally use  bayesian CPD for deformable registration 
phase, which increases the speed about 10X, which is useful for 
multi-template estimations as well as replications. Please look at the BCPD 
repo (https://github.com/ohirose/bcpd) for installation instructions on 
your OS. 

Finally, a new stable version (v5.2.1) of 3D Slicer is now available at, 
with lots of new features and bug fixes. All the changes to SlicerMorph 
listed above are only available for the latest stable. To benefit from 
them, please make sure to update your Slicer installation to the current 
stable available at: https://download.slicer.org. 

You can file bug reports or problems you encountered at 
https://download.slicer.org. 


Best wishes,
M

-- 
You received this message because you are subscribed to the Google Groups 
"Morphmet" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/morphmet2/0958edd7-811c-48ca-a280-1b2e2cbe4fadn%40googlegroups.com.

Reply via email to