CUDA for MacBook Pro, Windows and DICOM into STL?

Hi,

I am open source developer and someone is asking me the following questions:

a) If I use ImageJ / Fiji without CUDA support, would it render faster using a MacBook Pro?

b) If I enable CUDA support on my Windows machine, would it render faster?

c) Does ImageJ / Fiji fully support the feature of converting DICOM into STL after stacking images?

Thank you very much for your help,

Best regards :smiley:

Welcome to the forum, @olivern.

Depends. Would it render faster than which alternative? From your other questions, I infer this to be a comparison between macOS and Windows? The OS really is not important here—what matters mostly is the graphics card, as well as to a lesser extent CPU performance and available RAM.

ImageJ’s classic 3D rendering plugin, the 3D Viewer, leans on Java 3D. The new one under development, SciView, leans on OpenGL and also has a new Vulkan backend via its backing Scenery library. So really, this question is just about comparing OpenGL rendering et. al across graphics hardware specs.

No, ImageJ does not offer any CUDA-based acceleration. And no one is working on this yet, to my knowledge, at least not in the general case of volume rendering.

Probably not by the definition of “fully” which is desired here. But yes, you can import DICOM files and then export meshes using the 3D Viewer (File :arrow_forward: Export Surfaces :arrow_forward: STL menu items).

2 Likes
  1. Scenery/SciView has compute shaders under development. However, that support is through OpenCL, not CUDA. Nevertheless, a better graphics card will give you better performance for those operations. @skalarproduktraum would probably have more to say about that.

  2. Yes, meshing a 3D image, such as a DICOM is already available in SciView (demoed at a recent ImageJ meeting).

  3. SciView’s official update site will be up at the end of the month, and all of the features you are asking about already exist. Note that volume rendering may be released a bit later, but even that is already working through the Vulkan renderer of Scenery.

4 Likes