nv-generate-ct-rflow

Files (33)

• BENCHMARK.md
• SKILL.md
• evals.json · evals/
• README.md · fixtures/
• abdomen_hepatic_tumor.json · fixtures/
• abdomen_liver_spleen.json · fixtures/
• chest_lung_lobes.json · fixtures/
• chest_lung_tumor_controllable.json · fixtures/
• ct_from_mask_request_example.json · fixtures/
• ct_image_only_default.json · fixtures/
• ct_mask_lung_tumor.json · fixtures/
• default_config_infer.json · fixtures/
• head_brain.json · fixtures/
• pelvis.json · fixtures/
• ct-from-mask-format.md · references/
• ct-mask-label-space.md · references/
• fov-and-downloads.md · references/
• _anatomy.py · scripts/
• _summary_card.py · scripts/
• list_anatomies.py · scripts/
• run_ct_from_mask.py · scripts/
• run_ct_image.py · scripts/
• run_ct_mask.py · scripts/
• run_rflow_ct.py · scripts/
• wrapper_utils.py · scripts/
• skill-card.md
• skill.oms.sig
• skill_manifest.yaml
• test_run_ct_from_mask.py · tests/
• test_run_ct_image.py · tests/
• test_run_ct_mask.py · tests/
• test_run_rflow_ct.py · tests/
• output_schema.json · validators/

SKILL.md (first lines)

# NV-Generate-CT (rflow-ct)

## Purpose
- Used for generating synthetic CT volumes and masks with NV-Generate-CTMR rflow-ct. Not for production training data without review.
- Use the wrapper exactly as documented; do not replace the upstream entrypoint with a handwritten implementation.
- Do not write custom inference code for normal runs. The wrapper owns config staging, output paths, label mapping evidence, and validation.
- Manifest I/O: inputs are `config_infer_override`; outputs are `synthetic_ct_volumes` and `result_json`.

## Instructions
- Read `skill_manifest.yaml` before changing arguments, side effects, or validation gates.
- Run `scripts/run_rflow_ct.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_rflow_ct.py", args=[...])`; otherwise run the Bash/Python command shown below.
- Emit a single bash code block, and keep the `python -m pip install -r "$NV_GENERATE_ROOT/requirements.txt"` step in that same command — the runtime may be a fresh environment without `nibabel`/MONAI, so dropping the install fails with `ModuleNotFoundError`.
- Do not add `rm`, `mkdir`, or any cleanup of `--output-dir`; the wrapper creates it. Use a fresh `--output-dir` instead of deleting one.
- Check the emitted JSON and paired verifier guidance before treating the run as evidence.

## Available Scripts
| Script | Purpose | Arguments |
|---|---|---|
| `scripts/_anatomy.py` | Internal helper used by the primary entrypoint. | Imported only; do not call directly. |
| `scripts/_summary_card.py` | Internal helper used by the primary entrypoint. | Imported only; do not call directly. |
| `scripts/list_anatomies.py` | Helper command for catalog or anatomy lookup. | `[--region REGION] [--filter TEXT] [--controllable]` |
| `scripts/run_rflow_ct.py` | Primary entrypoint declared by skill_manifest.yaml. | `CONFIG_INFER.json --output-dir OUT_DIR [--random-seed N] [--version rflow-ct] [--yes]` |
| `scripts/run_ct_mask.py` | Advanced diagnostic helper for standalone raw MAISI mask generation. | `REQUEST.json --output-dir OUT_DIR [--random-seed N] [--preflight-only] [--yes]` |
| `scripts/run_ct_from_mask.py` | Advanced helper for CT image generation from a MAISI label mask. | `REQUEST.json --output-dir OUT_DIR [--random-seed N] [--yes]` |
| `scripts/run_ct_image.py` | Advanced helper for CT image-only generation without paired labels. | `MODEL_CONFIG.json --output-dir OUT_DIR [--version rflow-ct] [--random-seed N] [--yes]` |

## Prerequisites
- Required environment variables: `NV_GENERATE_ROOT`.
- Runtime requirements: GPU/CUDA when declared by the manifest; Python packages listed in `runtime.side_effects.pip_packages`.
- Side effects: writes generated outputs under the caller's `--output-dir`, may cache model assets under `~/.cache/huggingface/`, and may contact `https://huggingface.co` or `https://github.com` during setup.
- Run commands from the repository root unless an existing section below says otherwise.

## Limitations
- This is a thin wrapper. Inference, sampling, and decoding are delegated entirely to NVIDIA-Medtech/NV-Generate-CTMR's `scripts.inference`. Do not modify code under $NV_GENERATE_ROOT.
- rflow-ct requires CUDA and ≈ 16 GB VRAM minimum for the default 256³ output_size. Larger output_size (e.g. 512×512×768) needs an A100/H100.
- Output volumes are synthetic. They are not safe to use as training data for production medtech models without an independent quality review.
- Not for clinical deployment, clinical interpretation, autonomous diagnosis, regulatory submission.

## Troubleshooting
| Error | Cause | Fix |
|---|---|---|
| Missing dependency or import error | Runtime package drift from `skill_manifest.yaml`. | Install the packages declared in the manifest or use the documented setup command. |
| Empty or schema-invalid output | Wrong input path, unsupported modality, or upstream failure. | Re-run with a known fixture and inspect the wrapper JSON plus stderr. |
| Validation gate failure | Output violated a declared engineering invariant. | Keep the failed evidence pack and use the gate message to repair inputs or wrapper code. |

Wraps the upstream
[`NVIDIA-Medtech/NV-Generate-CTMR`](https://github.com/NVIDIA-Medtech/NV-Generate-CTMR/tree/61c4ec709b84cad468852243c48e250bec732074)
rectified-flow synthesis pipeline. The wrapper does not reimplement diffusion,
sampling, or autoencoder decoding — it shells out to the upstream
`scripts.inference` entry point exactly as the project's README documents and
inspects the produced image/mask pairs.

## Preconditions

1. If `NV_GENERATE_ROOT` already names a local checkout, the wrapper uses it
   and records its current commit in the result. Otherwise, create the
   recommended pinned default checkout (one-time):

   ```bash
   if [ -z "${NV_GENERATE_ROOT:-}" ]; then
     export NV_GENERATE_COMMIT=61c4ec709b84cad468852243c48e250bec732074
     export NV_GENERATE_ROOT="$HOME/.cache/nvidia-skills/upstreams/NV-Generate-CTMR-61c4ec7"
     if [ ! -d "$NV_GENERATE_ROOT/.git" ]; then
       git clone https://github.com/NVIDIA-Medtech/NV-Generate-CTMR.git "$NV_GENERATE_ROOT"
       git -C "$NV_GENERATE_ROOT" checkout --detach "$NV_GENERATE_COMMIT"
     fi
   fi
   python -m pip install -r "$NV_GENERATE_ROOT/requirements.txt"
   ```

2. Download the `rflow-ct` weights **and** the mask-candidate datasets
   into the clone (one-time, ≈ 5.5 GB):

   ```bash
   cd "$NV_GENERATE_ROOT"
   python -m scripts.download_model_data --version rflow-ct --root_dir "./"
   ```

   The mask candidates (`datasets/all_masks_flexible_size_and_spacing_4000`)
   condition the diffusion sampler; omitting them via `--model_only` will
   make the inference script fail with a missing-file error at startup.
   The anatomy-size condition file is also part of the full CT download and is
   needed for controllable mask generation.

3. NVIDIA GPU with ≥ 16 GB VRAM and CUDA. There is no CPU fallback.

For agent-generated user run commands, prefer the short wrapper command in
Usage. Do not prepend clone or model-download setup steps when `NV_GENERATE_ROOT`
or the repo-local upstream cache is already present. In a fresh Python
environment, still include `python -m pip install -r "$NV_GENERATE_ROOT/requirements.txt"`
before the wrapper unless the active environment has already proven those
imports are available; cached weights do not imply cached Python packages. Run
the wrapper from the medical-AI-skills repo root. If setup requires `cd "$NV_GENERATE_ROOT"`, return to the Medical AI Skills repo before invoking
`skills/nv-generate-ct-rflow/scripts/run_rflow_ct.py`.

## Usage

```bash
export NV_GENERATE_ROOT="${NV_GENERATE_ROOT:-$HOME/.cache/nvidia-skills/upstreams/NV-Generate-CTMR-61c4ec7}" && \
python -m pip install -r "$NV_GENERATE_ROOT/requirements.txt" && \
python skills/nv-generate-ct-rflow/scripts/run_rflow_ct.py \
  PATH_TO_CONFIG_INFER.json \
  --output-dir runs/nv_generate_ct_rflow_demo \
  --random-seed 0 \
  --version rflow-ct
```

Replace `PATH_TO_CONFIG_INFER.json` with the user's actual request/config
path. Do not copy the fixture path from this document unless the user
explicitly asked to run that fixture. If the user says "the case request is at
`runs/.../chest_lung_tumor_controllable.json`", that exact path is the first
positional argument to `scripts/run_rflow_ct.py`.

The fixture argument is a `config_infer.json` override file: it can replace
`num_output_samples`, `body_region`, `anatomy_list`, `controllable_anatomy_size`,
`output_size`, and `spacing`. Pass `default` to use the upstream config
verbatim. The wrapper stages the override into the upstream tree before
running.

From NVIDIA/skills, Apache-2.0.

More from NVIDIA/skills