Wrong message in extracting radiomics about 64-bit float

When I extracted from 3D segment, some could be extracted but some failed, the wrong messgae is as follows:

[2020-05-22 21:40:45] E: radiomics.script: Feature extraction failed!
Traceback (most recent call last):
  File "D:\python37\lib\site-packages\pyradiomics-2.2.0.post7+gac7458e-py3.7-win-amd64.egg\radiomics\scripts\segment.py", line 70, in _extractFeatures
    feature_vector.update(extractor.execute(imageFilepath, maskFilepath, label, label_channel))
  File "D:\python37\lib\site-packages\pyradiomics-2.2.0.post7+gac7458e-py3.7-win-amd64.egg\radiomics\featureextractor.py", line 267, in execute
    image, mask = self.loadImage(imageFilepath, maskFilepath, generalInfo)
  File "D:\python37\lib\site-packages\pyradiomics-2.2.0.post7+gac7458e-py3.7-win-amd64.egg\radiomics\featureextractor.py", line 380, in loadImage
    image = imageoperations.normalizeImage(image, **self.settings)
  File "D:\python37\lib\site-packages\pyradiomics-2.2.0.post7+gac7458e-py3.7-win-amd64.egg\radiomics\imageoperations.py", line 595, in normalizeImage
    image *= scale
  File "D:\python37\lib\site-packages\SimpleITK\SimpleITK.py", line 4277, in __mul__
    return Multiply( self, float(other) )
  File "D:\python37\lib\site-packages\SimpleITK\SimpleITK.py", line 50876, in Multiply
    return _SimpleITK.Multiply(*args)
RuntimeError: Exception thrown in SimpleITK Multiply: c:\d\vs14-win64-pkg\simpleitk\code\common\include\sitkMemberFunctionFactory.hxx:196:
sitk::ERROR: Pixel type: vector of 64-bit float is not supported in 3D byclass itk::simple::MultiplyImageFilter

Your pixeltype is a vector of 64-bit float, i.e. a color image. PyRadiomics is designed to extract from gray-scale images. What is the input data you are using? You could consider extracting a single color channel, or taking the mean across color channels.

I used MRI image. It was a gray-scale image.

What file type did you use? If it is something like .png or .jpg, your image may indeed be grayscale and look like that, but because the underlying format is a color format, it is still stored as a color image and therefore not accepted by PyRadiomics.
Moreover, formats like .png and .jpg do not contain the geometric space information (i.e. how the image data array translates to a real-world volume). This is especially important when calculating shape features, but also if you want to resample to different spacing or use distance weighting in GLRLM and/or GLCM.

When possible, I usually advise to use NRRD or NIFTII format, which can store grayscale as truly grayscale, and contains the geometry information. There are several tools available to convert DICOMs into this format.