I’m working on a preprocessing and a segmentation pipeline for vessel extraction from micro-scan medical images. The image volumes are around 2048x2048x1400 and 2560x2560x1400, with 5,5 Gb and 8,5 Gb in 8 bits. In my pipeline i’m using some Frangi , k-mean and morphological filters and i’m struggling with processing such volumes.
I’ve made some research and tried to use stream divisions in the filters to solve the problem but nothing changed ( maybe i’m using it wrong). Apparently, loading and processing the whole image is the problem, so i tried to split in different stacks but filters such as Frangi has different results working like that.
So, i was wondering what options could be more interesting to deal with such volumes of medical images. Could you guys help me on this?
I’m using Python 3.7.6 on windows 10.
Thank you for the attention!