# Uniform spacing for prediction task

**URL:** <https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374>\
**Category:** Support\
**Created:** [March 14, 2023, 1:31pm UTC](https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374 "2023-03-14T13:31:06Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Nima\_Yousefi](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/nima_yousefi/32/66845_2.png) [@Nima\_Yousefi](https://discourse.slicer.org/u/Nima_Yousefi)\
**Post date:** [March 14, 2023, 1:31pm UTC](https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374/1 "2023-03-14T13:31:06Z")

</div>

Hi,  
In case of prediction and feature extraction,  
is it necessary to have uniform spacing for ct images by doing resampling?

@lassoan I would appreciate if you share your opinions with me.

Best regard

---

<div class="post-metadata">

**Author:** ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)\
**Post date:** [March 14, 2023, 2:20pm UTC](https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374/2 "2023-03-14T14:20:32Z")

</div>

Yes, generally for any 3D processing or analysis you can avoid some problems if you resample the image to have isotropic resolution. Resampling is a lossy operation, so try to keep the current resolution along axes (e.g, if you have a 0.8x0.8x3mm spacing then resample to 0.8x0.8x0.8 instead of 1.0x1.0x1.0).

---

<div class="post-metadata">

**Author:** ![Nima\_Yousefi](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/nima_yousefi/32/66845_2.png) [@Nima\_Yousefi](https://discourse.slicer.org/u/Nima_Yousefi)\
**Post date:** [March 14, 2023, 5:31pm UTC](https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374/3 "2023-03-14T17:31:17Z")

</div>

Thank you. How about filtering such as wavelet or LoG?

---

<div class="post-metadata">

**Author:** ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)\
**Post date:** [March 14, 2023, 6:57pm UTC](https://discourse.slicer.org/t/uniform-spacing-for-prediction-task/28374/4 "2023-03-14T18:57:54Z")

</div>

Noise filtering or other arbitrary enhancement or feature extraction filters are probably unnecessary and may even lead to worse end results.