# Pyradiomics GLRLM crashing at native 100um resolution vs 250um

**URL:** <https://discourse.slicer.org/t/pyradiomics-glrlm-crashing-at-native-100um-resolution-vs-250um/47500>\
**Category:** Support\
**Tags:** feature, pyradiomics\
**Created:** [June 29, 2026, 5:10pm UTC](https://discourse.slicer.org/t/pyradiomics-glrlm-crashing-at-native-100um-resolution-vs-250um/47500 "2026-06-29T17:10:30Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![oviya](https://avatars.discourse-cdn.com/v4/letter/o/e495f1/32.png) [@oviya](https://discourse.slicer.org/u/oviya)\
**Post date:** [June 29, 2026, 5:10pm UTC](https://discourse.slicer.org/t/pyradiomics-glrlm-crashing-at-native-100um-resolution-vs-250um/47500/1 "2026-06-29T17:10:30Z")

</div>

Hello everyone! My name is Oviya, I’m an undergrad hoping to get a bit of help.

I have been trying extract features from MRI images via a compute cluster, and have been successful with downsampled (250um) images (41MB). However, I am unable to run the same script on the native 100um resolution (542 MB). On the 100um resolution, I’ve been able to extract GLCM, GLSZM, NGTDM, and GLDM, but **not** GLRLM.

Tech support from the compute cluster indicated that the job is failing with a segmentation fault (exit code 139), which indicates a low-level crash in compiled code (e.g., ITK/SimpleITK/PyRadiomics or NumPy C extensions), not a Python exception.

From the logs, PyRadiomics completes feature extraction successfully (“Reading feature maps: 100%”), and the crash occurs immediately afterward during post-processing. So the failure might be happening in the step where extracted feature maps are converted into arrays and assembled into a feature matrix, or during cleanup between chunks.

I’m unsure of how to resolve this. If anyone could take a peek at the code/offer advice, it would be appreciated!

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/a/d/ad3ba1d2f6a2464d78eb94faedf43d9de6982651.png)

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/f/9/f998cda4ab2c2f93185a805c47835048b3ebc347.png)

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/9/2/9257a33ce06ee3606cb56b3a2bf91bd368389cbc.png)

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/d/c/dc9b40d152931aa586660bd0188404d5c8e47bcb.png)

---

<div class="post-metadata">

**Author:** ![PaoloZaffino](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/paolozaffino/32/81052_2.png) [@PaoloZaffino](https://discourse.slicer.org/u/PaoloZaffino)\
**Post date:** [June 29, 2026, 5:39pm UTC](https://discourse.slicer.org/t/pyradiomics-glrlm-crashing-at-native-100um-resolution-vs-250um/47500/2 "2026-06-29T17:39:04Z")

</div>

Just out of curiosity, have you tried [Radiomics.jl](https://www.radiomicsjl.org/)?

It’s written in Julia (so it is much more efficient) but it can be easily called from Python.

Of course, it is IBSI compliant!

---

<div class="post-metadata">

**Author:** ![oviya](https://avatars.discourse-cdn.com/v4/letter/o/e495f1/32.png) [@oviya](https://discourse.slicer.org/u/oviya)\
**Post date:** [June 29, 2026, 7:25pm UTC](https://discourse.slicer.org/t/pyradiomics-glrlm-crashing-at-native-100um-resolution-vs-250um/47500/3 "2026-06-29T19:25:43Z")

</div>

I haven’t tried radiomics.jl, I/my colleagues haven’t used it before, so there is some unfamiliarity. However it can definitely be a next step I can try! In the meantime if there are PyRadiomics solutions, I’d love to hear.

Best,  
Oviya