NV-Segment-CT

NV-Segment-CT

Purpose

• Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence. Not for clinical interpretation.
• Use the wrapper exactly as documented; do not replace the upstream entrypoint with a handwritten implementation.
• Manifest I/O: inputs are ct_volume; outputs are label_map and result_json.

Instructions

• Read skill_manifest.yaml before changing arguments, side effects, or validation gates.
• Run scripts/run_vista3d.py through the documented command below; keep outputs under a caller-provided run directory.
• If a host agent exposes run_script, use run_script("scripts/run_vista3d.py", args=[...]); otherwise run the Bash/Python command shown below.
• Create the documented Python 3.10 virtual environment and invoke its binaries directly; do not install model dependencies into the caller's active environment.
• Check the emitted JSON and paired verifier guidance before treating the run as evidence.

Available Scripts

Prerequisites

• Runtime requirements: Python 3.10 with venv support and GPU/CUDA when declared by the manifest. Model packages come from the pinned upstream requirements file; only wrapper-specific packages are added locally.
• Side effects: creates an isolated environment under ~/.cache/nvidia-skills/venvs/nv-segment-ct-f9f5f51/, writes the downloaded bundle under skills/nv-segment-ct/bundle/, may cache model assets under ~/.cache/huggingface/, and may contact https://huggingface.co and https://raw.githubusercontent.com during first setup; the optional spleen fixture fetcher downloads MSD09 from https://msd-for-monai.s3-us-west-2.amazonaws.com.
• Run commands from the repository root unless an existing section below says otherwise.

Limitations

• This is a thin wrapper. Inference, preprocessing, and postprocessing are delegated entirely to the official hugging_face_pipeline.HuggingFacePipelineHelper in bundle/. Do not modify code under bundle/.
• transformers==4.46.3 is the wrapper compatibility overlay tested with the upstream requirements' Torch 2.0.1; newer Transformers releases can disable that older Torch backend.
• The pinned upstream requirements include Torch 2.0.1. Use only the pinned NVIDIA model assets; do not load untrusted checkpoints in this legacy reproduction environment.
• Device auto-detected (cuda if available, else cpu); --device flag overrides.
• Output may be schema-valid but semantically empty (e.g. label prompts that do not match the input anatomy). Sanity gates assert at least one foreground voxel per requested anatomy.
• Not for clinical deployment, clinical interpretation, autonomous diagnosis, regulatory submission.

Troubleshooting

| Error | Cause | Fix |