Error in Nvidia AIAA

Since my Nvidia AIAA has lacked of the model of annotation_mri_brain_tumors_t1ce_tc, I reinstalled the Nvidia AIAA.
After I reinstall the Nvidia AIAA, it happen a error call “Failed to fetch models from remote server. Make sure server address is correct and <server_uri>/v1/models is accessible in browser”. How to fix it?

What server address do you use?
If you open that server address with /v1/models appended in the web browser (e.g., http://perklabseg.asuscomm.com:5000/v1/models) then what do you see?

[
  {
    "name": "clara_pt_brain_mri_annotation_t1c",
    "labels": [
      "brain tumor core"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the tumor from MRI T1C image using DEXtr3D",
    "version": "1",
    "type": "annotation"
  },
  {
    "name": "clara_pt_brain_mri_segmentation_t1c",
    "labels": [
      "tumor"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the tumor core from MRI T1c image",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_covid19_ct_lesion_segmentation",
    "labels": [
      "COVID-19 affected regions"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of COVID-19 affected regions in the lung from CT image",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_covid19_ct_lung_annotation",
    "labels": [
      "lung"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the lung from CT image using DEXtr3D",
    "version": "1",
    "type": "annotation"
  },
  {
    "name": "clara_pt_covid19_ct_lung_segmentation",
    "labels": [
      "lung"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the Lung from CT image",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_deepgrow_2d_annotation",
    "labels": [],
    "description": "2D DeepGrow model based on Unet",
    "version": "1",
    "type": "deepgrow",
    "deepgrow": "2D"
  },
  {
    "name": "clara_pt_deepgrow_3d_annotation",
    "labels": [],
    "description": "3D DeepGrow model based on Unet",
    "version": "1",
    "type": "deepgrow",
    "deepgrow": "3D"
  },
  {
    "name": "clara_pt_liver_and_tumor_ct_segmentation",
    "labels": [
      "liver",
      "liver tumor"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the liver and tumor from CT",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_pancreas_and_tumor_ct_segmentation",
    "labels": [
      "pancreas",
      "pancreas tumor"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the pancreas and tumor from portal venous phase CT",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_prostate_mri_segmentation",
    "labels": [
      "prostate central gland",
      "prostate peripheral zone"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the liver and tumor from MRI",
    "version": "1",
    "type": "segmentation"
  },
  {
    "name": "clara_pt_spleen_ct_annotation",
    "labels": [
      "spleen"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the spleen from CT image using DEXtr3D",
    "version": "1",
    "type": "annotation"
  },
  {
    "name": "clara_pt_spleen_ct_segmentation",
    "labels": [
      "spleen"
    ],
    "description": "A pre-trained model for volumetric (3D) segmentation of the spleen from CT image",
    "version": "1",
    "type": "segmentation"
  }
]

I am using default server address

The server response looks good. Which Slicer version are you using? Are you behind a hospital/corporate firewall or proxy server?

1 Like

version 4.13.0, I am using university network.

Which Slicer-4.13.0 version exactly? I’ve just tested latest Slicer Preview Release and it works well.
Do you use a proxy server to connect to the internet?

2021-11-21
I can use it now, after I shutdown the Anti-Virus software.
But it still lack of the model of annotation_mri_brain_tumors_t1ce_tc

I’ve installed all the models that I’ve found on NGC for latest Clara4 a few months ago. If you know about any additional pretrained models that you would like to see on the server then let me know.

After I download the models, Is it just put it into the extension file of Nvidia and it’s fine?

The models must be installed on the server. If there are any Clara v4 pre-trained models that are not installed on the server already then send me the download link and I’ll install on the server.

I would like to these two model.

clara_pt_brain_mri_segmentation_t1c - already installed

clara_pt_fed_learning_brain_tumor_mri_segmentation - this model requires 4 channel MRI (4 aligned MRIs T1c, T1, T2, FLAIR at 1x1x1 mm), which GUI clients, such as the NVidia AIAA Segment Editor effect in 3D Slicer does not support. You need to do some custom Python scripting, as described here:

You can add a comment to this issue requesting this feature to be implemented in the GUI, but since NVidia the developers currently mostly focus on MONAILabel, most likely you’ll get more help if you ask for making this model available in the MONAILabel Slicer extension.

Is clara_pt_brain_mri_segmentation_t1c only use in auto segmentation? I would to use it in segment from boundary points.

Each model is for a specific segmentation method. For example, a model created for “auto-segmentation” can only be used for “auto-segmentation”, and it cannot be used for “segment from boundary points”.

Dear Andres,

I have the same error. I was using the NVIDIA AIAA segmenting the lung tumour using the annotation lung tumor model (Segment from boundary points) this morning, but now this error appears.
I installed it on another computer, it doesn’t have any error, but now it doesn’t conssist the annotation_lung_tumor (it just has the covid models for lung) and there is no model for the lung tumour anymore. I would be grateful if you help me, please.

Best regards,
Parisa

For lung segmentation, on the default Slicer AIAA segmentation server you can find the clara_pt_covid19_ct_lung_segmentation automatic segmentation model and the clara_pt_covid19_ct_lung_annotation point based segmentation model.

You can find more pre-trained models for MONAILabel.

You can also segment lungs using LungCTAnalyzer and DensityLungSegmentation extensions.

Are you sure this model is available for the latest NVidia Clara AIAA server version? I don’t see it on NGC. If you can send a download link on NGC then I can install it on the server.

Thanks for your reply.
No, unfortunately it’s not available anymore.
There is no model for lung tumor (not the whole lung).
It was very helpful. But, it seems this model is not available now.

Yesterday I could use it for a couple of hours. Because I had installed the NVIDIA AIAA on 3D-Slicer a long time ago and I was using the lung tumor annotation model. But suddenly it stopped and this error appeared.