# MONAI Auto3DSeg inference sources

**URL:** <https://discourse.slicer.org/t/monai-auto3dseg-inference-sources/36929>\
**Category:** Support\
**Tags:** segmentation, monai-auto3dseg\
**Created:** [June 20, 2024, 9:31pm UTC](https://discourse.slicer.org/t/monai-auto3dseg-inference-sources/36929 "2024-06-20T21:31:38Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![wesselk](https://avatars.discourse-cdn.com/v4/letter/w/ac8455/32.png) [@wesselk](https://discourse.slicer.org/u/wesselk)\
**Post date:** [June 20, 2024, 9:31pm UTC](https://discourse.slicer.org/t/monai-auto3dseg-inference-sources/36929/1 "2024-06-20T21:31:38Z")

</div>

Hi all,

I have been working with the new MONAI Auto3DSeg and have a few questions if someone could help me out:

1- Where can I find the source for the models used? For example, “Madiastinal anatomy TS2” is not in MONAI’s [model zoo](https://monai.io/model-zoo) and it’s also not an available subtask or roi\_subset in [Total Segmentator](https://github.com/wasserth/TotalSegmentator).

2- Were some of these models trained by Slicer community contributors?

3- Is there a way to run Auto3DSeg’s inference from cmd line without launching Slicer and switching modules?

Thanks in advance for any help!

---

<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:** [June 21, 2024, 7:32pm UTC](https://discourse.slicer.org/t/monai-auto3dseg-inference-sources/36929/2 "2024-06-21T19:32:37Z")

</div>

Most models that you can find online in various model zoos are not [optimized for real-world usage](https://github.com/lassoan/SlicerMONAIAuto3DSeg/tree/main?tab=readme-ov-file#general-guidelines). We created these models with @diazandr3s specifically to make them practically, clinically useful.

You can download these models and run them without Slicer GUI, from the command-line or via network requests within Slicer’s Python environment; or even completely independently from Slicer. To make it really easy to use without Slicer, we have separated the inference script into a [standalone Python file](https://github.com/lassoan/SlicerMONAIAuto3DSeg/blob/main/MONAIAuto3DSeg/Scripts/auto3dseg_segresnet_inference.py).
