Killed when appied in bigger data

[Whiteglad]

When I run the smooth script for bigger data（upsampled 2X ), it was killed. What is the maximum resolution that this tool can handle？

[yoda-vid]

Thanks for asking, @Whiteglad, and great question. In practice the image size is mainly limited by the system memory, so it may be platform-dependent. There are size limits based on file format, but the tool's readers (NumPy for NPY files, or ITK for other formats such as MHD/MHD, NIfTI, etc) should be able to read existing images regardless of size (as far as I know). What is the size of the image you are loading?

It could be that there was an out of memory error when loading the large image. Could you let me know what command you ran and the error message you encountered? Are you using Mac or Windows (or Linux)? Also, did it work for you on the original data, but was killed with the larger data?

[SaibotMagd]

I could easily load a 4.3 GB file into the GUI without the importer. I tried a 88.3 GB lightsheet tif-file using the importer, it needed about 30min to load but after that it worked pretty fine. Scrolling through slices, zooming works smooth.

[yoda-vid]

Thanks, @SaibotMagd, glad it's working!

Just to note that images formats loaded without requiring import (eg MHD/MHA files loaded through SimpleITK) are fully loaded into memory, so they can become memory constrained. There's a way to access these images by ROI but haven't had a chance to implement it yet.

[SaibotMagd]

There's a way to access these images by ROI but haven't had a chance to implement it yet.

1. Load the picture into e.g. ImageJ as VirtualImage (bio-image plugin), it only loads a single slice at the same time
2. use the command: " python run.py --img sample_region/myvolume --proc load --offset 30,35,30 --size 25,25,12 to set the to-be-loaded ROI in mag.

[yoda-vid]

This is super helpful, thanks for pointing it out, @SaibotMagd!