Documentation/Instructions to use DISA-LC2 to perform Multimodal image registration

ImFusion Suite
1

Hello,

Are there any instructions / documentations on how to use DISA-LC2 algorithm to perform multi-modal image registration? I am able to extract the DISA features on two different modalities but then I am unsure how to perform the registration on the ImFusionSuite.

Thanks in advance!

2

Hi,
If you are registering two volumes you can use the “Descriptors Registration” algorithm (in the menu “Registration > Descriptors Registration”). The controller of this algorithm allows you to select between DISA and MIND features and to run a local or global registration.

If instead you need to register an Ultrasound sweep to a volume, you can use the algorithm in “Ultrasound > Advanced > Ultrasound DISA registration”. This algorithm adds some ultrasound specific preprocessing and weighting to the “Descriptors Registration” algorithm.

If you encounter any issues with the registration I will be happy to help you.

Best,
Matteo Ronchetti

3

Hi,

Thank you for your reply. When I try to perform the Descriptors Registration with DISA, I get this error:

[DISAFeaturesAlgorithm] Preset model 'DISA-LC2' cannot be used since neither ONNX or Torch engine is available
[DescriptorsRegistrationAlgorithm] DISA Algorithm failed.
[DescriptorsRegistrationAlgorithm] Error in extracting descriptors
[DescriptorsRegistrationAlgorithm] Error in computing features

Am I supposed to install torch engine separately?

Also, just to confirm about the terms local vs. global registration, does it refer to the search space of the deformation parameters?

Thanks,
Manasi