[pyrodiomics] the extracted image omics characteristic value is abnormal (many 0 and 1)

use Pyradiomics Many of the extracted image features are 0 and 1（ The figure below excel surface ）, This doesn't seem normal , It may also affect the modeling results .

  The reason is the problem of normalization . The parameter file needs to be modified *.yaml.

If normalization is set normalize: True（ Or normalize yourself ）, Then the range of gray value is 0-1 Between , At this time, if binWidth: 32（ Other values are the same ）, Then the above situation will occur . This is because the program will set the gray values at intervals 32 Classified as 1 individual bin, That is to say 0-32 Classified as 1 Gray scale , By analogy .

So there are two solutions ：

1. binWidth: 32  Change to  binCount: 32 ( Regardless of the grayscale range , It is stipulated that only 32 Gray scale , This method program will alarm, but it doesn't matter , It's just Pyradiomics The author thinks that binWidth Better , Many articles are also used binCount)

2.  Simultaneous setting ：

normalize: True

normalizeScale: 1024 （ Other values can also ）

binWidth: 32

For the discussion of these two methods, please refer to the documentation ：https://pyradiomics.readthedocs.io/en/latest/faq.html?highlight=binCount#what-about-gray-value-discretization-fixed-bin-width-fixed-bin-count