[FEAT EXTRACTION] Segmentation fault (core dumped) error

[arpitadu]

I am extracting voxel-wise features from MRI (.nii format) scans using PyRadiomics (in python). Getting the error "Segmentation fault (core dumped)", when PyRadiomics is computing GLCM features for Wavelet Image. I have data for 50+ patients, so I tried to run on multiple patients. But, I am getting the same error for them.

Please advise. Thanks in advance!

PyRadiomics configuration Here is my Parameter file:

---

imageType: Original: {} LoG: {'sigma' : [1.0, 3.0, 5.0]} Wavelet: binWidth: 10 Gradient: gradientUseSpacing: true LBP3D: lbp3DLevels: 2 lbp3DIcosphereRadius: 2 lbp3DIcosphereSubdivision: 1

featureClass: glcm: # Disable SumAverage by specifying all other GLCM features available

• 'Autocorrelation'
• 'JointAverage'
• 'ClusterProminence'
• 'ClusterShade'
• 'ClusterTendency'
• 'Contrast'
• 'Correlation'
• 'DifferenceAverage'
• 'DifferenceEntropy'
• 'DifferenceVariance'
• 'JointEnergy'
• 'JointEntropy'
• 'Imc1'
• 'Imc2'
• 'Idm'
• 'Idmn'
• 'Id'
• 'Idn'
• 'InverseVariance'
• 'MaximumProbability'
• 'SumEntropy'
• 'SumSquares' glrlm: glszm: gldm:

  ngtdm:

  firstorder: shape:

setting: binWidth: 25 padDistance: 4 preCrop: true label: 255

voxelSetting: kernelRadius: 4 maskedKernel: true initValue: nan voxelBatch: 10000

---

PyRadiomics log file

Version (please complete the following information):

• OS: Linux
• Python version: 3.9
• PyRadiomics version

Additional context It's taking a long time to process a single image, around ~5hr or more. Generally, I am getting this error after 4 hrs of the run. Please let me know is there any way to improve.

[JoostJM]

Sorry for the late reply.

What PyRadiomics version are you using? Does this happen for each image? Does it also happen if you process the case as a single image (i.e. without a csv file as input)?

If so, can you share an image and mask (anonymized) for me to debug with?

[JoostJM]

likely releated to #737