Hello all
For example I have 1000 dicom images. When I read that image folder into the pipeline for 3D rendering, how much resources will it take to load image data into the GPU?
DICOM files are usually not compressed, so the images will use similar amount of memory of the GPU as the total size of the files on disk.
Clinical 3D images (CT, MRI, US, PET) are usually easily handled on all modern GPUs. Size of microCT, microscopy, etc. images can be tens of gigabytes, so they may exceed your GPU memory size and rendering at high resolution can be slow.
For example, I have 1000 CT dicom images of 512x512, 16 bits per pixel. How much resources will it take to load image data into the GPU?
In theory, it should be 1000 * 512 * 512 * 32 = 1GB
Don’t worry too much about memory usage. If you work with images that are too large to comfortably fit in memory then you have a whole lot of other, more complex problems to solve.
1 Like
@lassoan Thanks for your answer