Slicer quits unexpectedly - Label statistics

Slicer Version: 4.10.2
OS: Ubuntu

Hi everyone,

I’m not sure if this is the same issue as this thread, but Slicer has been quitting on me unexpectedly as well. Particularly, every time I try to calculate label statistics from a scalar map (mean diffusivity, in particular). I do not get this issue when I try with the other maps of interest (in this case, FA, Trace, RA, Parallel and Perpendicular Diffusivity).

Any help would be greatly appreciated - thank you!

Label statistics module is superseded by Segment statistics module. Can you try if you experience any problems using Segment Statistics module on a recent Slicer Preview Release?

Hi Andras,

I downloaded 4.11.0 and installed the dMRI plugin. When I try to open MRML scenes that have the MeanDiffusivity label map created, Slicer crashes. It can open scenes that do not have the MeanDiffusivity label map, but when I try to make it (to test the segment statistics module), again it crashes. Not sure what the issue is.

I went back to 4.10.2 and it can be calculated with Segment statistics though, which is great.

Thanks for your help.

Can you share a scene that causes Slicer-4.11 to crash so that we can investigate and fix the issue? Does Slicer-4.11 work correctly if instead of loading the scene file (.mrml) you just load the MeanDiffusivity label map and other files?

Hi Andras,

I don’t think I can share the scenes I was working with, but I was able to replicate the problem using the DTI tutorial data from the 3DSlicer group. Is there a place I can upload it?

Using both my data the the tutorial data, I was unable to open the mean diffusivity files on 4.11 - still crashes. FA and the other scalar maps work fine though.

Yes, please upload that scene to onedrive/dropbox/googledrive and post the link here. Thanks!

Here you go.

Mean diffusivity image contains invalid floating-point value (NaN), which causes a VTK filter (vtkImageHistogramStatistics) to crash when it tries to figure out a good default window/level value. I’ve fixed that problem in VTK now, it’ll just take a few days to get that integrated into Slicer (you can track progress here).

How did you compute this mean diffusivity image? It would be probably better to change that algorithm to not have NaN values in the result (e.g., replace by 0 or -1), as such values rarely used and may cause unexpected behavior in various software.

Hi Andras,

Brilliant! Thank you for the fix, eager to see it integrated. :slight_smile:

I actually just use the SlicerDMRI plugin to compute the mean diffusivity map. Under Diffusion -> Quantify -> Diffusion Tensor Scalar Maps. I then choose my DTI and create the map. I’ve tried maybe half the maps, but only this one causes the issue so far.

Could you please report the issue here:
Thank you!

1 Like