Intro to Image Registration
Ella Bahry from Helmholtz Imaging / Max Delbrück Center (MDC) gives an overview of image registration techniques, explaining how to align different images in the same coordinate space for scientific analysis.
The video covers various transformation types, interpolation methods, and registration tools, different software tools and approaches for specific imaging modalities, emphasizing the importance of understanding data characteristics and testing different methods to achieve optimal registration results.
This workshop was offered as a HIDA course:
https://www.helmholtz-hida.de/en/
Presentation is available here:
https://ida-mdc.github.io/registratio...
Github repo for notebooks and presentation:
https://github.com/ida-mdc/registrati...
00:00:00 Intro to image registration - goal and uses
00:06:21 Types of transformations
00:11:26 Interpolation
00:18:03 Review of approaches
00:18:49 Intensity based, mutual information, ITKElastix
00:29:55 Feature based techniques - SIFT, deep learning based feature detection
00:44:51 Deep learning for registration @VoxelMorph
00:49:04 Considerations, tools, designing your pipeline
Corrections:
00:40:15 Not perfectly aligned but well aligned.
00:46:42 You do need training data, but you don't need ANNOTATED training data.