# False Positive value

**URL:** <https://discourse.itk.org/t/false-positive-value/5640>\
**Category:** Beginner Questions\
**Tags:** segementation, metric, python\
**Created:** [January 22, 2023, 9:44pm UTC](https://discourse.itk.org/t/false-positive-value/5640 "2023-01-22T21:44:39Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Moha\_data](https://discourse.itk.org/letter_avatar_proxy/v4/letter/m/ecc23a/32.png) [@Moha\_data](https://discourse.itk.org/u/Moha_data)\
**Post date:** [January 22, 2023, 9:44pm UTC](https://discourse.itk.org/t/false-positive-value/5640/1 "2023-01-22T21:44:39Z")

</div>

I have a binary image (background = 0) and to see how False positive works, I applied the dilation to the image (as a test image) and calculated the False Positive.

![image](https://discourse.itk.org/uploads/default/original/2X/b/b175dde3997dcf3d1c972dc6ca2901a056527263.png)

The left image is the reference and the right one is the test. I got the False positive value of 0. Shouldn’t be a value bigger than zero? I appreciate any comment. Thanks

---

<div class="post-metadata">

**Author:** ![dzenanz](https://discourse.itk.org/user_avatar/discourse.itk.org/dzenanz/32/1093_2.png) [@dzenanz](https://discourse.itk.org/u/dzenanz)\
**Post date:** [January 23, 2023, 3:33pm UTC](https://discourse.itk.org/t/false-positive-value/5640/2 "2023-01-23T15:33:58Z")

</div>

That image should have many false positives, and no false negatives. If you reverse the reference/test image, than there should be many false negatives and no false positives. Double check what you are using as reference, and try it the other way around for sanity check.

---

<div class="post-metadata">

**Author:** ![Moha\_data](https://discourse.itk.org/letter_avatar_proxy/v4/letter/m/ecc23a/32.png) [@Moha\_data](https://discourse.itk.org/u/Moha_data)\
**Post date:** [January 24, 2023, 4:41pm UTC](https://discourse.itk.org/t/false-positive-value/5640/3 "2023-01-24T16:41:52Z")

</div>

Right! that will fix the problem.  
The document says: “LabelOverlapMeasuresImageFilte computes overlap measures between the set same set of labels of pixels of two images. Background is assumed to be 0”.  
To wrap up, while for some metrics like Dice, it is not important how to pass the Test and Reference images, it can affect False positive/Negative results.
