# Normalize Pyradiomics features by the number of pixels (Volume)

**URL:** <https://discourse.slicer.org/t/normalize-pyradiomics-features-by-the-number-of-pixels-volume/18617>\
**Category:** Radiomics\
**Tags:** radiomics, pyradiomics\
**Created:** [July 10, 2021, 2:17pm UTC](https://discourse.slicer.org/t/normalize-pyradiomics-features-by-the-number-of-pixels-volume/18617 "2021-07-10T14:17:06Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![MachadoL](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/machadol/32/3372_2.png) [@MachadoL](https://discourse.slicer.org/u/MachadoL)\
**Post date:** [July 10, 2021, 2:17pm UTC](https://discourse.slicer.org/t/normalize-pyradiomics-features-by-the-number-of-pixels-volume/18617/1 "2021-07-10T14:17:06Z")

</div>

Hey guys,  
I am extracting radiomics from several different volume ROIS. I am afraid the ROI volume affects my evaluation, since many radiomic features are not divided by the number of pixels. I did even read an article speaking about the bias caused by volume yet present in many features.  
I believe a reasonable solution would be to divide those non-volume-normalized features by the number of pixels.

What do you guys think?

ARTICLE: Radiomics features of the primary tumor fail to improve prediction of overall survival in large cohorts of CT- and PET-imaged head and neck cancer patients, by Rachel B. Ger. 2019.