https://github.com/insightsoftwareconsortium/itkthinshelldemons

Thin Shell Demons Mesh Registration
https://github.com/insightsoftwareconsortium/itkthinshelldemons

Last synced: 6 months ago
JSON representation

Thin Shell Demons Mesh Registration

• Host: GitHub
• URL: https://github.com/insightsoftwareconsortium/itkthinshelldemons
• Owner: InsightSoftwareConsortium
• License: apache-2.0
• Created: 2021-03-16T19:23:13.000Z (almost 5 years ago)
• Default Branch: main
• Last Pushed: 2025-02-06T15:20:04.000Z (12 months ago)
• Last Synced: 2025-07-27T08:32:21.519Z (6 months ago)
• Language: C++
• Size: 167 KB
• Stars: 5
• Watchers: 16
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
![Build Status](https://github.com/InsightSoftwareConsortium/ITKThinShellDemons/workflows/Build,%20test,%20package/badge.svg)

![PyPi Version](https://img.shields.io/pypi/v/itk-thinshelldemons.svg)

![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)

# ITKThinShellDemons

This module implements the Thin Shell Demons regularization proposed in

> Thin Shell Demons  

> Zhao Q, Price T, Pizer S, Niethammer M, Alterovitz R, Rosenman J  

> MIUA 2015  

> :warning: **This module requires to be compiled against an ITK version with**  

> - [PointSetToPointSetMetricWithIndexv4](https://github.com/InsightSoftwareConsortium/ITK/pull/2385)   

> 







## V4 version (TSHDv4)

The v4 version is implemented in ThinShellDemonsMetricv4. With this approach different

transformation can be combined with the thin shell regularization. For examples see:

1. [Affine](./test/itkThinShellDemonsTestv4_Affine.cxx)

2. [DisplacementField](./test/itkThinShellDemonsTestv4_Displacement.cxx)

3. [SyN Diffeomorphism](./test/itkThinShellDemonsTestv4_SyN.cxx)

For Python samples please see:

1. [Affine](./examples/test_tsd_affine.py)

2. [DisplacementField](./examples/test_tsd_displacement.py)

3. [SyN Diffeomorphism](./examples/test_tsd_syn.py)

Please be aware that the regularization is implemented on the fixed mesh. This is due to the 

nomenclature in of the moving transform being computed from fixed image to moving image. For 

re-sampling the moving image is pulled back to the fixed image domain. For point sets

it's easier to compute the push forward than the pull back (for some transformation the 

inverse might be difficult compute or not available). Hence, the nomenclature matches

the ITK approach for images and computes a transform from fixed to moving domain. However to

register a point set it is the fixed point set that is transformed to the moving point set 

domain and it is computationally more efficient to regularize on the fixed mesh.

## Requirements

Compile against ITK.

## Authors

Qingyu Zhao (original version)  

Samuel Gerber (v4 version and updates to original)  

Pranjal Sahu