Image size in demos
shaghayegh1376 opened this issue · 7 comments
hi authors, in defining default geometry in TIGRE the image size is 256.256 but why in implementing algorithms we use head phantom with 64.64 image size?
The size you want to use is completely arbitrary, you can use whichever you want. In the demos, sometimes we have small images just so they run fast on your computer, but feel free to change the size as much as you wish.
I appreciate your guidance
Another question is that compared to the results of other articles using GPU, it seems that the execution time using this toolbox is a little longer. what is the reason
@shaghayegh1376 What do you mean "other articles"? The execution time will totally depend on your GPU, of course.
All the TIGRE times in all the TIGRE articles have been measured with as fast, or slower version of TIGRE than the current one, by me. In all TIGRE articles, you can also see which hardware I used.
I mean the results of articles about CBCT reconstruction algorithms using GPU.
i got your point so it depends on system hardware . i am using NVIDIA GeForce RTX 3050
@shaghayegh1376 the RTX XX50 is the slowest GPU of any generation XX of NVIDIA GPUs, so its reasonable that you get slower performance.
how we can measure image quality for FDK?
because with "Measure_Quality.m" function we can measure image quality in each iteration
how we can measure image quality compare with ground truth