How can I use GPU to accelerate General Registration (Elastix) module?

Hi all,
I’m using General Registration (Elastix) to align two different time DICOM CT of one patient, and it works very well!
However it takes such a long time, about 5min. I checked the CPU and GPU usage status. As you can see in the figures, memory is consumed 84% while GPU is only about 14%. How can I use more GPU to process this module?
Info of my computer is:

  1. CPU: 13th Gen Intel(R) Core™ i7-13650HX 2.60GHz
  2. RAM: 16.0GB
  3. GPU: NVIDIA GeForce RTX 4060 Laptop 8GB


It seems like only using Memory to process the computation. And I’ve checked the CUDA in Slicer’s Python Console.